Daily Papers

Video Large Language Models (Video LLMs) have recently exhibited remarkable capabilities in general video understanding. However, they mainly focus on holistic comprehension and struggle with capturing fine-grained spatial and temporal details. Besides, the lack of high-quality object-level video instruction data and a comprehensive benchmark further hinders their advancements. To tackle these challenges, we introduce the VideoRefer Suite to empower Video LLM for finer-level spatial-temporal video understanding, i.e., enabling perception and reasoning on any objects throughout the video. Specially, we thoroughly develop VideoRefer Suite across three essential aspects: dataset, model, and benchmark. Firstly, we introduce a multi-agent data engine to meticulously curate a large-scale, high-quality object-level video instruction dataset, termed VideoRefer-700K. Next, we present the VideoRefer model, which equips a versatile spatial-temporal object encoder to capture precise regional and sequential representations. Finally, we meticulously create a VideoRefer-Bench to comprehensively assess the spatial-temporal understanding capability of a Video LLM, evaluating it across various aspects. Extensive experiments and analyses demonstrate that our VideoRefer model not only achieves promising performance on video referring benchmarks but also facilitates general video understanding capabilities.

Pre-training on large-scale video data has become a common recipe for learning transferable spatiotemporal representations in recent years. Despite some progress, existing methods are mostly limited to highly curated datasets (e.g., K400) and exhibit unsatisfactory out-of-the-box representations. We argue that it is due to the fact that they only capture pixel-level knowledge rather than spatiotemporal semantics, which hinders further progress in video understanding. Inspired by the great success of image-text pre-training (e.g., CLIP), we take the first step to exploit language semantics to boost transferable spatiotemporal representation learning. We introduce a new pretext task, Turning to Video for Transcript Sorting (TVTS), which sorts shuffled ASR scripts by attending to learned video representations. We do not rely on descriptive captions and learn purely from video, i.e., leveraging the natural transcribed speech knowledge to provide noisy but useful semantics over time. Our method enforces the vision model to contextualize what is happening over time so that it can re-organize the narrative transcripts, and can seamlessly apply to large-scale uncurated video data in the real world. Our method demonstrates strong out-of-the-box spatiotemporal representations on diverse benchmarks, e.g., +13.6% gains over VideoMAE on SSV2 via linear probing. The code is available at https://github.com/TencentARC/TVTS.

We present a large-scale study on unsupervised spatiotemporal representation learning from videos. With a unified perspective on four recent image-based frameworks, we study a simple objective that can easily generalize all these methods to space-time. Our objective encourages temporally-persistent features in the same video, and in spite of its simplicity, it works surprisingly well across: (i) different unsupervised frameworks, (ii) pre-training datasets, (iii) downstream datasets, and (iv) backbone architectures. We draw a series of intriguing observations from this study, e.g., we discover that encouraging long-spanned persistency can be effective even if the timespan is 60 seconds. In addition to state-of-the-art results in multiple benchmarks, we report a few promising cases in which unsupervised pre-training can outperform its supervised counterpart. Code is made available at https://github.com/facebookresearch/SlowFast

Transformers have recently been popular for learning and inference in the spatial-temporal domain. However, their performance relies on storing and applying attention to the feature tensor of each frame in video. Hence, their space and time complexity increase linearly as the length of video grows, which could be very costly for long videos. We propose a novel visual memory network architecture for the learning and inference problem in the spatial-temporal domain. We maintain a fixed set of memory slots in our memory network and propose an algorithm based on Gumbel-Softmax to learn an adaptive strategy to update this memory. Finally, this architecture is benchmarked on the video object segmentation (VOS) and video prediction problems. We demonstrate that our memory architecture achieves state-of-the-art results, outperforming transformer-based methods on VOS and other recent methods on video prediction while maintaining constant memory capacity independent of the sequence length.

We present EmerNeRF, a simple yet powerful approach for learning spatial-temporal representations of dynamic driving scenes. Grounded in neural fields, EmerNeRF simultaneously captures scene geometry, appearance, motion, and semantics via self-bootstrapping. EmerNeRF hinges upon two core components: First, it stratifies scenes into static and dynamic fields. This decomposition emerges purely from self-supervision, enabling our model to learn from general, in-the-wild data sources. Second, EmerNeRF parameterizes an induced flow field from the dynamic field and uses this flow field to further aggregate multi-frame features, amplifying the rendering precision of dynamic objects. Coupling these three fields (static, dynamic, and flow) enables EmerNeRF to represent highly-dynamic scenes self-sufficiently, without relying on ground truth object annotations or pre-trained models for dynamic object segmentation or optical flow estimation. Our method achieves state-of-the-art performance in sensor simulation, significantly outperforming previous methods when reconstructing static (+2.93 PSNR) and dynamic (+3.70 PSNR) scenes. In addition, to bolster EmerNeRF's semantic generalization, we lift 2D visual foundation model features into 4D space-time and address a general positional bias in modern Transformers, significantly boosting 3D perception performance (e.g., 37.50% relative improvement in occupancy prediction accuracy on average). Finally, we construct a diverse and challenging 120-sequence dataset to benchmark neural fields under extreme and highly-dynamic settings.

This paper studies a conceptually simple extension of Masked Autoencoders (MAE) to spatiotemporal representation learning from videos. We randomly mask out spacetime patches in videos and learn an autoencoder to reconstruct them in pixels. Interestingly, we show that our MAE method can learn strong representations with almost no inductive bias on spacetime (only except for patch and positional embeddings), and spacetime-agnostic random masking performs the best. We observe that the optimal masking ratio is as high as 90% (vs. 75% on images), supporting the hypothesis that this ratio is related to information redundancy of the data. A high masking ratio leads to a large speedup, e.g., > 4x in wall-clock time or even more. We report competitive results on several challenging video datasets using vanilla Vision Transformers. We observe that MAE can outperform supervised pre-training by large margins. We further report encouraging results of training on real-world, uncurated Instagram data. Our study suggests that the general framework of masked autoencoding (BERT, MAE, etc.) can be a unified methodology for representation learning with minimal domain knowledge.

Building on the advances of language models, Large Multimodal Models (LMMs) have contributed significant improvements in video understanding. While the current video LMMs utilize advanced Large Language Models (LLMs), they rely on either image or video encoders to process visual inputs, each of which has its own limitations. Image encoders excel at capturing rich spatial details from frame sequences but lack explicit temporal context, which can be important in videos with intricate action sequences. On the other hand, video encoders provide temporal context but are often limited by computational constraints that lead to processing only sparse frames at lower resolutions, resulting in reduced contextual and spatial understanding. To this end, we introduce VideoGPT+, which combines the complementary benefits of the image encoder (for detailed spatial understanding) and the video encoder (for global temporal context modeling). The model processes videos by dividing them into smaller segments and applies an adaptive pooling strategy on features extracted by both image and video encoders. Our architecture showcases improved performance across multiple video benchmarks, including VCGBench, MVBench and Zero-shot question-answering. Further, we develop 112K video-instruction set using a novel semi-automatic annotation pipeline which further improves the model performance. Additionally, to comprehensively evaluate video LMMs, we present VCGBench-Diverse, covering 18 broad video categories such as lifestyle, sports, science, gaming, and surveillance videos. This benchmark with 4,354 question-answer pairs evaluates the generalization of existing LMMs on dense video captioning, spatial and temporal understanding, and complex reasoning, ensuring comprehensive assessment across diverse video types and dynamics. Code: https://github.com/mbzuai-oryx/VideoGPT-plus.

We present a self-supervised Contrastive Video Representation Learning (CVRL) method to learn spatiotemporal visual representations from unlabeled videos. Our representations are learned using a contrastive loss, where two augmented clips from the same short video are pulled together in the embedding space, while clips from different videos are pushed away. We study what makes for good data augmentations for video self-supervised learning and find that both spatial and temporal information are crucial. We carefully design data augmentations involving spatial and temporal cues. Concretely, we propose a temporally consistent spatial augmentation method to impose strong spatial augmentations on each frame of the video while maintaining the temporal consistency across frames. We also propose a sampling-based temporal augmentation method to avoid overly enforcing invariance on clips that are distant in time. On Kinetics-600, a linear classifier trained on the representations learned by CVRL achieves 70.4% top-1 accuracy with a 3D-ResNet-50 (R3D-50) backbone, outperforming ImageNet supervised pre-training by 15.7% and SimCLR unsupervised pre-training by 18.8% using the same inflated R3D-50. The performance of CVRL can be further improved to 72.9% with a larger R3D-152 (2x filters) backbone, significantly closing the gap between unsupervised and supervised video representation learning. Our code and models will be available at https://github.com/tensorflow/models/tree/master/official/.

Convolutional Neural Networks define an exceptionally powerful class of models, but are still limited by the lack of ability to be spatially invariant to the input data in a computationally and parameter efficient manner. In this work we introduce a new learnable module, the Spatial Transformer, which explicitly allows the spatial manipulation of data within the network. This differentiable module can be inserted into existing convolutional architectures, giving neural networks the ability to actively spatially transform feature maps, conditional on the feature map itself, without any extra training supervision or modification to the optimisation process. We show that the use of spatial transformers results in models which learn invariance to translation, scale, rotation and more generic warping, resulting in state-of-the-art performance on several benchmarks, and for a number of classes of transformations.

This work explores the use of spatial context as a source of free and plentiful supervisory signal for training a rich visual representation. Given only a large, unlabeled image collection, we extract random pairs of patches from each image and train a convolutional neural net to predict the position of the second patch relative to the first. We argue that doing well on this task requires the model to learn to recognize objects and their parts. We demonstrate that the feature representation learned using this within-image context indeed captures visual similarity across images. For example, this representation allows us to perform unsupervised visual discovery of objects like cats, people, and even birds from the Pascal VOC 2011 detection dataset. Furthermore, we show that the learned ConvNet can be used in the R-CNN framework and provides a significant boost over a randomly-initialized ConvNet, resulting in state-of-the-art performance among algorithms which use only Pascal-provided training set annotations.

We present a novel technique for self-supervised video representation learning by: (a) decoupling the learning objective into two contrastive subtasks respectively emphasizing spatial and temporal features, and (b) performing it hierarchically to encourage multi-scale understanding. Motivated by their effectiveness in supervised learning, we first introduce spatial-temporal feature learning decoupling and hierarchical learning to the context of unsupervised video learning. We show by experiments that augmentations can be manipulated as regularization to guide the network to learn desired semantics in contrastive learning, and we propose a way for the model to separately capture spatial and temporal features at multiple scales. We also introduce an approach to overcome the problem of divergent levels of instance invariance at different hierarchies by modeling the invariance as loss weights for objective re-weighting. Experiments on downstream action recognition benchmarks on UCF101 and HMDB51 show that our proposed Hierarchically Decoupled Spatial-Temporal Contrast (HDC) makes substantial improvements over directly learning spatial-temporal features as a whole and achieves competitive performance when compared with other state-of-the-art unsupervised methods. Code will be made available.

Applying Multimodal Large Language Models (MLLMs) to video understanding presents significant challenges due to the need to model temporal relations across frames. Existing approaches adopt either implicit temporal modeling, relying solely on the LLM decoder, or explicit temporal modeling, employing auxiliary temporal encoders. To investigate this debate between the two paradigms, we propose the Stackable Temporal Encoder (STE). STE enables flexible explicit temporal modeling with adjustable temporal receptive fields and token compression ratios. Using STE, we systematically compare implicit and explicit temporal modeling across dimensions such as overall performance, token compression effectiveness, and temporal-specific understanding. We also explore STE's design considerations and broader impacts as a plug-in module and in image modalities. Our findings emphasize the critical role of explicit temporal modeling, providing actionable insights to advance video MLLMs.

This paper presents a simple yet effective approach to modeling space-time correspondences in the context of video object segmentation. Unlike most existing approaches, we establish correspondences directly between frames without re-encoding the mask features for every object, leading to a highly efficient and robust framework. With the correspondences, every node in the current query frame is inferred by aggregating features from the past in an associative fashion. We cast the aggregation process as a voting problem and find that the existing inner-product affinity leads to poor use of memory with a small (fixed) subset of memory nodes dominating the votes, regardless of the query. In light of this phenomenon, we propose using the negative squared Euclidean distance instead to compute the affinities. We validated that every memory node now has a chance to contribute, and experimentally showed that such diversified voting is beneficial to both memory efficiency and inference accuracy. The synergy of correspondence networks and diversified voting works exceedingly well, achieves new state-of-the-art results on both DAVIS and YouTubeVOS datasets while running significantly faster at 20+ FPS for multiple objects without bells and whistles.

Recently, large-scale pre-trained language-image models like CLIP have shown extraordinary capabilities for understanding spatial contents, but naively transferring such models to video recognition still suffers from unsatisfactory temporal modeling capabilities. Existing methods insert tunable structures into or in parallel with the pre-trained model, which either requires back-propagation through the whole pre-trained model and is thus resource-demanding, or is limited by the temporal reasoning capability of the pre-trained structure. In this work, we present DiST, which disentangles the learning of spatial and temporal aspects of videos. Specifically, DiST uses a dual-encoder structure, where a pre-trained foundation model acts as the spatial encoder, and a lightweight network is introduced as the temporal encoder. An integration branch is inserted between the encoders to fuse spatio-temporal information. The disentangled spatial and temporal learning in DiST is highly efficient because it avoids the back-propagation of massive pre-trained parameters. Meanwhile, we empirically show that disentangled learning with an extra network for integration benefits both spatial and temporal understanding. Extensive experiments on five benchmarks show that DiST delivers better performance than existing state-of-the-art methods by convincing gaps. When pre-training on the large-scale Kinetics-710, we achieve 89.7% on Kinetics-400 with a frozen ViT-L model, which verifies the scalability of DiST. Codes and models can be found in https://github.com/alibaba-mmai-research/DiST.

We present an unsupervised visual feature learning algorithm driven by context-based pixel prediction. By analogy with auto-encoders, we propose Context Encoders -- a convolutional neural network trained to generate the contents of an arbitrary image region conditioned on its surroundings. In order to succeed at this task, context encoders need to both understand the content of the entire image, as well as produce a plausible hypothesis for the missing part(s). When training context encoders, we have experimented with both a standard pixel-wise reconstruction loss, as well as a reconstruction plus an adversarial loss. The latter produces much sharper results because it can better handle multiple modes in the output. We found that a context encoder learns a representation that captures not just appearance but also the semantics of visual structures. We quantitatively demonstrate the effectiveness of our learned features for CNN pre-training on classification, detection, and segmentation tasks. Furthermore, context encoders can be used for semantic inpainting tasks, either stand-alone or as initialization for non-parametric methods.

We introduce an object-aware decoder for improving the performance of spatio-temporal representations on ego-centric videos. The key idea is to enhance object-awareness during training by tasking the model to predict hand positions, object positions, and the semantic label of the objects using paired captions when available. At inference time the model only requires RGB frames as inputs, and is able to track and ground objects (although it has not been trained explicitly for this). We demonstrate the performance of the object-aware representations learnt by our model, by: (i) evaluating it for strong transfer, i.e. through zero-shot testing, on a number of downstream video-text retrieval and classification benchmarks; and (ii) by using the representations learned as input for long-term video understanding tasks (e.g. Episodic Memory in Ego4D). In all cases the performance improves over the state of the art -- even compared to networks trained with far larger batch sizes. We also show that by using noisy image-level detection as pseudo-labels in training, the model learns to provide better bounding boxes using video consistency, as well as grounding the words in the associated text descriptions. Overall, we show that the model can act as a drop-in replacement for an ego-centric video model to improve performance through visual-text grounding.

We present an unsupervised representation learning approach that compactly encodes the motion dependencies in videos. Given a pair of images from a video clip, our framework learns to predict the long-term 3D motions. To reduce the complexity of the learning framework, we propose to describe the motion as a sequence of atomic 3D flows computed with RGB-D modality. We use a Recurrent Neural Network based Encoder-Decoder framework to predict these sequences of flows. We argue that in order for the decoder to reconstruct these sequences, the encoder must learn a robust video representation that captures long-term motion dependencies and spatial-temporal relations. We demonstrate the effectiveness of our learned temporal representations on activity classification across multiple modalities and datasets such as NTU RGB+D and MSR Daily Activity 3D. Our framework is generic to any input modality, i.e., RGB, Depth, and RGB-D videos.

We consider the task of Image-to-Video (I2V) generation, which involves transforming static images into realistic video sequences based on a textual description. While recent advancements produce photorealistic outputs, they frequently struggle to create videos with accurate and consistent object motion, especially in multi-object scenarios. To address these limitations, we propose a two-stage compositional framework that decomposes I2V generation into: (i) An explicit intermediate representation generation stage, followed by (ii) A video generation stage that is conditioned on this representation. Our key innovation is the introduction of a mask-based motion trajectory as an intermediate representation, that captures both semantic object information and motion, enabling an expressive but compact representation of motion and semantics. To incorporate the learned representation in the second stage, we utilize object-level attention objectives. Specifically, we consider a spatial, per-object, masked-cross attention objective, integrating object-specific prompts into corresponding latent space regions and a masked spatio-temporal self-attention objective, ensuring frame-to-frame consistency for each object. We evaluate our method on challenging benchmarks with multi-object and high-motion scenarios and empirically demonstrate that the proposed method achieves state-of-the-art results in temporal coherence, motion realism, and text-prompt faithfulness. Additionally, we introduce \benchmark, a new challenging benchmark for single-object and multi-object I2V generation, and demonstrate our method's superiority on this benchmark. Project page is available at https://guyyariv.github.io/TTM/.

For computer vision systems to operate in dynamic situations, they need to be able to represent and reason about object permanence. We introduce a framework for learning to estimate 4D visual representations from monocular RGB-D, which is able to persist objects, even once they become obstructed by occlusions. Unlike traditional video representations, we encode point clouds into a continuous representation, which permits the model to attend across the spatiotemporal context to resolve occlusions. On two large video datasets that we release along with this paper, our experiments show that the representation is able to successfully reveal occlusions for several tasks, without any architectural changes. Visualizations show that the attention mechanism automatically learns to follow occluded objects. Since our approach can be trained end-to-end and is easily adaptable, we believe it will be useful for handling occlusions in many video understanding tasks. Data, code, and models are available at https://occlusions.cs.columbia.edu/.

High-quality video inpainting that completes missing regions in video frames is a promising yet challenging task. State-of-the-art approaches adopt attention models to complete a frame by searching missing contents from reference frames, and further complete whole videos frame by frame. However, these approaches can suffer from inconsistent attention results along spatial and temporal dimensions, which often leads to blurriness and temporal artifacts in videos. In this paper, we propose to learn a joint Spatial-Temporal Transformer Network (STTN) for video inpainting. Specifically, we simultaneously fill missing regions in all input frames by self-attention, and propose to optimize STTN by a spatial-temporal adversarial loss. To show the superiority of the proposed model, we conduct both quantitative and qualitative evaluations by using standard stationary masks and more realistic moving object masks. Demo videos are available at https://github.com/researchmm/STTN.

Recent temporal LiDAR-based 3D object detectors achieve promising performance based on the two-stage proposal-based approach. They generate 3D box candidates from the first-stage dense detector, followed by different temporal aggregation methods. However, these approaches require per-frame objects or whole point clouds, posing challenges related to memory bank utilization. Moreover, point clouds and trajectory features are combined solely based on concatenation, which may neglect effective interactions between them. In this paper, we propose a point-trajectory transformer with long short-term memory for efficient temporal 3D object detection. To this end, we only utilize point clouds of current-frame objects and their historical trajectories as input to minimize the memory bank storage requirement. Furthermore, we introduce modules to encode trajectory features, focusing on long short-term and future-aware perspectives, and then effectively aggregate them with point cloud features. We conduct extensive experiments on the large-scale Waymo dataset to demonstrate that our approach performs well against state-of-the-art methods. Code and models will be made publicly available at https://github.com/kuanchihhuang/PTT.

Recent advancements in multimodal large language models (MLLMs) have shown promising results, yet existing approaches struggle to effectively handle both temporal and spatial localization simultaneously. This challenge stems from two key issues: first, incorporating spatial-temporal localization introduces a vast number of coordinate combinations, complicating the alignment of linguistic and visual coordinate representations; second, encoding fine-grained temporal and spatial information during video feature compression is inherently difficult. To address these issues, we propose LLaVA-ST, a MLLM for fine-grained spatial-temporal multimodal understanding. In LLaVA-ST, we propose Language-Aligned Positional Embedding, which embeds the textual coordinate special token into the visual space, simplifying the alignment of fine-grained spatial-temporal correspondences. Additionally, we design the Spatial-Temporal Packer, which decouples the feature compression of temporal and spatial resolutions into two distinct point-to-region attention processing streams. Furthermore, we propose ST-Align dataset with 4.3M training samples for fine-grained spatial-temporal multimodal understanding. With ST-align, we present a progressive training pipeline that aligns the visual and textual feature through sequential coarse-to-fine stages.Additionally, we introduce an ST-Align benchmark to evaluate spatial-temporal interleaved fine-grained understanding tasks, which include Spatial-Temporal Video Grounding (STVG) , Event Localization and Captioning (ELC) and Spatial Video Grounding (SVG). LLaVA-ST achieves outstanding performance on 11 benchmarks requiring fine-grained temporal, spatial, or spatial-temporal interleaving multimodal understanding. Our code, data and benchmark will be released at Our code, data and benchmark will be released at https://github.com/appletea233/LLaVA-ST .

Automatically discovering composable abstractions from raw perceptual data is a long-standing challenge in machine learning. Recent slot-based neural networks that learn about objects in a self-supervised manner have made exciting progress in this direction. However, they typically fall short at adequately capturing spatial symmetries present in the visual world, which leads to sample inefficiency, such as when entangling object appearance and pose. In this paper, we present a simple yet highly effective method for incorporating spatial symmetries via slot-centric reference frames. We incorporate equivariance to per-object pose transformations into the attention and generation mechanism of Slot Attention by translating, scaling, and rotating position encodings. These changes result in little computational overhead, are easy to implement, and can result in large gains in terms of data efficiency and overall improvements to object discovery. We evaluate our method on a wide range of synthetic object discovery benchmarks namely CLEVR, Tetrominoes, CLEVRTex, Objects Room and MultiShapeNet, and show promising improvements on the challenging real-world Waymo Open dataset.

Very recently, a variety of vision transformer architectures for dense prediction tasks have been proposed and they show that the design of spatial attention is critical to their success in these tasks. In this work, we revisit the design of the spatial attention and demonstrate that a carefully-devised yet simple spatial attention mechanism performs favourably against the state-of-the-art schemes. As a result, we propose two vision transformer architectures, namely, Twins-PCPVT and Twins-SVT. Our proposed architectures are highly-efficient and easy to implement, only involving matrix multiplications that are highly optimized in modern deep learning frameworks. More importantly, the proposed architectures achieve excellent performance on a wide range of visual tasks, including image level classification as well as dense detection and segmentation. The simplicity and strong performance suggest that our proposed architectures may serve as stronger backbones for many vision tasks. Our code is released at https://github.com/Meituan-AutoML/Twins .

Popular representation learning methods encourage feature invariance under transformations applied at the input. However, in 3D perception tasks like object localization and segmentation, outputs are naturally equivariant to some transformations, such as rotation. Using pre-training loss functions that encourage equivariance of features under certain transformations provides a strong self-supervision signal while also retaining information of geometric relationships between transformed feature representations. This can enable improved performance in downstream tasks that are equivariant to such transformations. In this paper, we propose a spatio-temporal equivariant learning framework by considering both spatial and temporal augmentations jointly. Our experiments show that the best performance arises with a pre-training approach that encourages equivariance to translation, scaling, and flip, rotation and scene flow. For spatial augmentations, we find that depending on the transformation, either a contrastive objective or an equivariance-by-classification objective yields best results. To leverage real-world object deformations and motion, we consider sequential LiDAR scene pairs and develop a novel 3D scene flow-based equivariance objective that leads to improved performance overall. We show our pre-training method for 3D object detection which outperforms existing equivariant and invariant approaches in many settings.

Recent progress in using recurrent neural networks (RNNs) for image description has motivated the exploration of their application for video description. However, while images are static, working with videos requires modeling their dynamic temporal structure and then properly integrating that information into a natural language description. In this context, we propose an approach that successfully takes into account both the local and global temporal structure of videos to produce descriptions. First, our approach incorporates a spatial temporal 3-D convolutional neural network (3-D CNN) representation of the short temporal dynamics. The 3-D CNN representation is trained on video action recognition tasks, so as to produce a representation that is tuned to human motion and behavior. Second we propose a temporal attention mechanism that allows to go beyond local temporal modeling and learns to automatically select the most relevant temporal segments given the text-generating RNN. Our approach exceeds the current state-of-art for both BLEU and METEOR metrics on the Youtube2Text dataset. We also present results on a new, larger and more challenging dataset of paired video and natural language descriptions.

In this paper, we propose a simple yet effective approach for self-supervised video object segmentation (VOS). Our key insight is that the inherent structural dependencies present in DINO-pretrained Transformers can be leveraged to establish robust spatio-temporal correspondences in videos. Furthermore, simple clustering on this correspondence cue is sufficient to yield competitive segmentation results. Previous self-supervised VOS techniques majorly resort to auxiliary modalities or utilize iterative slot attention to assist in object discovery, which restricts their general applicability and imposes higher computational requirements. To deal with these challenges, we develop a simplified architecture that capitalizes on the emerging objectness from DINO-pretrained Transformers, bypassing the need for additional modalities or slot attention. Specifically, we first introduce a single spatio-temporal Transformer block to process the frame-wise DINO features and establish spatio-temporal dependencies in the form of self-attention. Subsequently, utilizing these attention maps, we implement hierarchical clustering to generate object segmentation masks. To train the spatio-temporal block in a fully self-supervised manner, we employ semantic and dynamic motion consistency coupled with entropy normalization. Our method demonstrates state-of-the-art performance across multiple unsupervised VOS benchmarks and particularly excels in complex real-world multi-object video segmentation tasks such as DAVIS-17-Unsupervised and YouTube-VIS-19. The code and model checkpoints will be released at https://github.com/shvdiwnkozbw/SSL-UVOS.

We wish to automatically predict the "speediness" of moving objects in videos---whether they move faster, at, or slower than their "natural" speed. The core component in our approach is SpeedNet---a novel deep network trained to detect if a video is playing at normal rate, or if it is sped up. SpeedNet is trained on a large corpus of natural videos in a self-supervised manner, without requiring any manual annotations. We show how this single, binary classification network can be used to detect arbitrary rates of speediness of objects. We demonstrate prediction results by SpeedNet on a wide range of videos containing complex natural motions, and examine the visual cues it utilizes for making those predictions. Importantly, we show that through predicting the speed of videos, the model learns a powerful and meaningful space-time representation that goes beyond simple motion cues. We demonstrate how those learned features can boost the performance of self-supervised action recognition, and can be used for video retrieval. Furthermore, we also apply SpeedNet for generating time-varying, adaptive video speedups, which can allow viewers to watch videos faster, but with less of the jittery, unnatural motions typical to videos that are sped up uniformly.

We present VideoOrion, a Video Large Language Model (Video-LLM) that explicitly captures the key semantic information in videos--the spatial-temporal dynamics of objects throughout the videos. VideoOrion employs expert vision models to extract object dynamics through a detect-segment-track pipeline, encoding them into a set of object tokens by aggregating spatial-temporal object features. Our method addresses the persistent challenge in Video-LLMs of efficiently compressing high-dimensional video data into semantic tokens that are comprehensible to LLMs. Compared to prior methods which resort to downsampling the original video or aggregating visual tokens using resamplers, leading to information loss and entangled semantics, VideoOrion not only offers a more natural and efficient way to derive compact, disentangled semantic representations but also enables explicit object modeling of video content with minimal computational cost. Moreover, the introduced object tokens naturally allow VideoOrion to accomplish video-based referring tasks. Experimental results show that VideoOrion can learn to make good use of the object tokens, and achieves competitive results on both general video question answering and video-based referring benchmarks.

The scarcity of annotated data in LiDAR point cloud understanding hinders effective representation learning. Consequently, scholars have been actively investigating efficacious self-supervised pre-training paradigms. Nevertheless, temporal information, which is inherent in the LiDAR point cloud sequence, is consistently disregarded. To better utilize this property, we propose an effective pre-training strategy, namely Temporal Masked Auto-Encoders (T-MAE), which takes as input temporally adjacent frames and learns temporal dependency. A SiamWCA backbone, containing a Siamese encoder and a windowed cross-attention (WCA) module, is established for the two-frame input. Considering that the movement of an ego-vehicle alters the view of the same instance, temporal modeling also serves as a robust and natural data augmentation, enhancing the comprehension of target objects. SiamWCA is a powerful architecture but heavily relies on annotated data. Our T-MAE pre-training strategy alleviates its demand for annotated data. Comprehensive experiments demonstrate that T-MAE achieves the best performance on both Waymo and ONCE datasets among competitive self-supervised approaches. Codes will be released at https://github.com/codename1995/T-MAE

Self-supervised methods have shown remarkable progress in learning high-level semantics and low-level temporal correspondence. Building on these results, we take one step further and explore the possibility of integrating these two features to enhance object-centric representations. Our preliminary experiments indicate that query slot attention can extract different semantic components from the RGB feature map, while random sampling based slot attention can exploit temporal correspondence cues between frames to assist instance identification. Motivated by this, we propose a novel semantic-aware masked slot attention on top of the fused semantic features and correspondence maps. It comprises two slot attention stages with a set of shared learnable Gaussian distributions. In the first stage, we use the mean vectors as slot initialization to decompose potential semantics and generate semantic segmentation masks through iterative attention. In the second stage, for each semantics, we randomly sample slots from the corresponding Gaussian distribution and perform masked feature aggregation within the semantic area to exploit temporal correspondence patterns for instance identification. We adopt semantic- and instance-level temporal consistency as self-supervision to encourage temporally coherent object-centric representations. Our model effectively identifies multiple object instances with semantic structure, reaching promising results on unsupervised video object discovery. Furthermore, we achieve state-of-the-art performance on dense label propagation tasks, demonstrating the potential for object-centric analysis. The code is released at https://github.com/shvdiwnkozbw/SMTC.

We address the problem of video representation learning without human-annotated labels. While previous efforts address the problem by designing novel self-supervised tasks using video data, the learned features are merely on a frame-by-frame basis, which are not applicable to many video analytic tasks where spatio-temporal features are prevailing. In this paper we propose a novel self-supervised approach to learn spatio-temporal features for video representation. Inspired by the success of two-stream approaches in video classification, we propose to learn visual features by regressing both motion and appearance statistics along spatial and temporal dimensions, given only the input video data. Specifically, we extract statistical concepts (fast-motion region and the corresponding dominant direction, spatio-temporal color diversity, dominant color, etc.) from simple patterns in both spatial and temporal domains. Unlike prior puzzles that are even hard for humans to solve, the proposed approach is consistent with human inherent visual habits and therefore easy to answer. We conduct extensive experiments with C3D to validate the effectiveness of our proposed approach. The experiments show that our approach can significantly improve the performance of C3D when applied to video classification tasks. Code is available at https://github.com/laura-wang/video_repres_mas.

Self-supervised tasks such as colorization, inpainting and zigsaw puzzle have been utilized for visual representation learning for still images, when the number of labeled images is limited or absent at all. Recently, this worthwhile stream of study extends to video domain where the cost of human labeling is even more expensive. However, the most of existing methods are still based on 2D CNN architectures that can not directly capture spatio-temporal information for video applications. In this paper, we introduce a new self-supervised task called as Space-Time Cubic Puzzles to train 3D CNNs using large scale video dataset. This task requires a network to arrange permuted 3D spatio-temporal crops. By completing Space-Time Cubic Puzzles, the network learns both spatial appearance and temporal relation of video frames, which is our final goal. In experiments, we demonstrate that our learned 3D representation is well transferred to action recognition tasks, and outperforms state-of-the-art 2D CNN-based competitors on UCF101 and HMDB51 datasets.

Despite the significant advancements made by Diffusion Transformer (DiT)-based methods in video generation, there remains a notable gap with controllable camera pose perspectives. Existing works such as OpenSora do NOT adhere precisely to anticipated trajectories and physical interactions, thereby limiting the flexibility in downstream applications. To alleviate this issue, we introduce CPA, a unified camera-pose-awareness text-to-video generation approach that elaborates the camera movement and integrates the textual, visual, and spatial conditions. Specifically, we deploy the Sparse Motion Encoding (SME) module to transform camera pose information into a spatial-temporal embedding and activate the Temporal Attention Injection (TAI) module to inject motion patches into each ST-DiT block. Our plug-in architecture accommodates the original DiT parameters, facilitating diverse types of camera poses and flexible object movement. Extensive qualitative and quantitative experiments demonstrate that our method outperforms LDM-based methods for long video generation while achieving optimal performance in trajectory consistency and object consistency.

Extracting informative representations from videos is fundamental for effectively learning various downstream tasks. We present a novel approach for unsupervised learning of meaningful representations from videos, leveraging the concept of image spatial entropy (ISE) that quantifies the per-pixel information in an image. We argue that local entropy of pixel neighborhoods and their temporal evolution create valuable intrinsic supervisory signals for learning prominent features. Building on this idea, we abstract visual features into a concise representation of keypoints that act as dynamic information transmitters, and design a deep learning model that learns, purely unsupervised, spatially and temporally consistent representations directly from video frames. Two original information-theoretic losses, computed from local entropy, guide our model to discover consistent keypoint representations; a loss that maximizes the spatial information covered by the keypoints and a loss that optimizes the keypoints' information transportation over time. We compare our keypoint representation to strong baselines for various downstream tasks, \eg, learning object dynamics. Our empirical results show superior performance for our information-driven keypoints that resolve challenges like attendance to static and dynamic objects or objects abruptly entering and leaving the scene.

Contrastive language-image pretraining (CLIP) has demonstrated remarkable success in various image tasks. However, how to extend CLIP with effective temporal modeling is still an open and crucial problem. Existing factorized or joint spatial-temporal modeling trades off between the efficiency and performance. While modeling temporal information within straight through tube is widely adopted in literature, we find that simple frame alignment already provides enough essence without temporal attention. To this end, in this paper, we proposed a novel Implicit Learnable Alignment (ILA) method, which minimizes the temporal modeling effort while achieving incredibly high performance. Specifically, for a frame pair, an interactive point is predicted in each frame, serving as a mutual information rich region. By enhancing the features around the interactive point, two frames are implicitly aligned. The aligned features are then pooled into a single token, which is leveraged in the subsequent spatial self-attention. Our method allows eliminating the costly or insufficient temporal self-attention in video. Extensive experiments on benchmarks demonstrate the superiority and generality of our module. Particularly, the proposed ILA achieves a top-1 accuracy of 88.7% on Kinetics-400 with much fewer FLOPs compared with Swin-L and ViViT-H. Code is released at https://github.com/Francis-Rings/ILA .

This paper studies referring video object segmentation (RVOS) by boosting video-level visual-linguistic alignment. Recent approaches model the RVOS task as a sequence prediction problem and perform multi-modal interaction as well as segmentation for each frame separately. However, the lack of a global view of video content leads to difficulties in effectively utilizing inter-frame relationships and understanding textual descriptions of object temporal variations. To address this issue, we propose Semantic-assisted Object Cluster (SOC), which aggregates video content and textual guidance for unified temporal modeling and cross-modal alignment. By associating a group of frame-level object embeddings with language tokens, SOC facilitates joint space learning across modalities and time steps. Moreover, we present multi-modal contrastive supervision to help construct well-aligned joint space at the video level. We conduct extensive experiments on popular RVOS benchmarks, and our method outperforms state-of-the-art competitors on all benchmarks by a remarkable margin. Besides, the emphasis on temporal coherence enhances the segmentation stability and adaptability of our method in processing text expressions with temporal variations. Code will be available.

Convolutional networks for single-view object reconstruction have shown impressive performance and have become a popular subject of research. All existing techniques are united by the idea of having an encoder-decoder network that performs non-trivial reasoning about the 3D structure of the output space. In this work, we set up two alternative approaches that perform image classification and retrieval respectively. These simple baselines yield better results than state-of-the-art methods, both qualitatively and quantitatively. We show that encoder-decoder methods are statistically indistinguishable from these baselines, thus indicating that the current state of the art in single-view object reconstruction does not actually perform reconstruction but image classification. We identify aspects of popular experimental procedures that elicit this behavior and discuss ways to improve the current state of research.

Current vision models typically maintain a fixed correspondence between their representation structure and image space. Each layer comprises a set of tokens arranged "on-the-grid," which biases patches or tokens to encode information at a specific spatio(-temporal) location. In this work we present Moving Off-the-Grid (MooG), a self-supervised video representation model that offers an alternative approach, allowing tokens to move "off-the-grid" to better enable them to represent scene elements consistently, even as they move across the image plane through time. By using a combination of cross-attention and positional embeddings we disentangle the representation structure and image structure. We find that a simple self-supervised objective--next frame prediction--trained on video data, results in a set of latent tokens which bind to specific scene structures and track them as they move. We demonstrate the usefulness of MooG's learned representation both qualitatively and quantitatively by training readouts on top of the learned representation on a variety of downstream tasks. We show that MooG can provide a strong foundation for different vision tasks when compared to "on-the-grid" baselines.

We present the task of Spatio-Temporal Video Question Answering, which requires intelligent systems to simultaneously retrieve relevant moments and detect referenced visual concepts (people and objects) to answer natural language questions about videos. We first augment the TVQA dataset with 310.8K bounding boxes, linking depicted objects to visual concepts in questions and answers. We name this augmented version as TVQA+. We then propose Spatio-Temporal Answerer with Grounded Evidence (STAGE), a unified framework that grounds evidence in both spatial and temporal domains to answer questions about videos. Comprehensive experiments and analyses demonstrate the effectiveness of our framework and how the rich annotations in our TVQA+ dataset can contribute to the question answering task. Moreover, by performing this joint task, our model is able to produce insightful and interpretable spatio-temporal attention visualizations. Dataset and code are publicly available at: http: //tvqa.cs.unc.edu, https://github.com/jayleicn/TVQAplus

Recent advances in video-based multimodal large language models (Video-LLMs) have significantly improved video understanding by processing videos as sequences of image frames. However, many existing methods treat frames independently in the vision backbone, lacking explicit temporal modeling, which limits their ability to capture dynamic patterns and efficiently handle long videos. To address these limitations, we introduce STORM (Spatiotemporal TOken Reduction for Multimodal LLMs), a novel architecture incorporating a dedicated temporal encoder between the image encoder and the LLM. Our temporal encoder leverages the Mamba State Space Model to integrate temporal information into image tokens, generating enriched representations that preserve inter-frame dynamics across the entire video sequence. This enriched encoding not only enhances video reasoning capabilities but also enables effective token reduction strategies, including test-time sampling and training-based temporal and spatial pooling, substantially reducing computational demands on the LLM without sacrificing key temporal information. By integrating these techniques, our approach simultaneously reduces training and inference latency while improving performance, enabling efficient and robust video understanding over extended temporal contexts. Extensive evaluations show that STORM achieves state-of-the-art results across various long video understanding benchmarks (more than 5\% improvement on MLVU and LongVideoBench) while reducing the computation costs by up to 8times and the decoding latency by 2.4-2.9times for the fixed numbers of input frames. Project page is available at https://research.nvidia.com/labs/lpr/storm

Current video-based Masked Autoencoders (MAEs) primarily focus on learning effective spatiotemporal representations from a visual perspective, which may lead the model to prioritize general spatial-temporal patterns but often overlook nuanced semantic attributes like specific interactions or sequences that define actions - such as action-specific features that align more closely with human cognition for space-time correspondence. This can limit the model's ability to capture the essence of certain actions that are contextually rich and continuous. Humans are capable of mapping visual concepts, object view invariance, and semantic attributes available in static instances to comprehend natural dynamic scenes or videos. Existing MAEs for videos and static images rely on separate datasets for videos and images, which may lack the rich semantic attributes necessary for fully understanding the learned concepts, especially when compared to using video and corresponding sampled frame images together. To this end, we propose CrossVideoMAE an end-to-end self-supervised cross-modal contrastive learning MAE that effectively learns both video-level and frame-level rich spatiotemporal representations and semantic attributes. Our method integrates mutual spatiotemporal information from videos with spatial information from sampled frames within a feature-invariant space, while encouraging invariance to augmentations within the video domain. This objective is achieved through jointly embedding features of visible tokens and combining feature correspondence within and across modalities, which is critical for acquiring rich, label-free guiding signals from both video and frame image modalities in a self-supervised manner. Extensive experiments demonstrate that our approach surpasses previous state-of-the-art methods and ablation studies validate the effectiveness of our approach.

Recent two-stream deep Convolutional Neural Networks (ConvNets) have made significant progress in recognizing human actions in videos. Despite their success, methods extending the basic two-stream ConvNet have not systematically explored possible network architectures to further exploit spatiotemporal dynamics within video sequences. Further, such networks often use different baseline two-stream networks. Therefore, the differences and the distinguishing factors between various methods using Recurrent Neural Networks (RNN) or convolutional networks on temporally-constructed feature vectors (Temporal-ConvNet) are unclear. In this work, we first demonstrate a strong baseline two-stream ConvNet using ResNet-101. We use this baseline to thoroughly examine the use of both RNNs and Temporal-ConvNets for extracting spatiotemporal information. Building upon our experimental results, we then propose and investigate two different networks to further integrate spatiotemporal information: 1) temporal segment RNN and 2) Inception-style Temporal-ConvNet. We demonstrate that using both RNNs (using LSTMs) and Temporal-ConvNets on spatiotemporal feature matrices are able to exploit spatiotemporal dynamics to improve the overall performance. However, each of these methods require proper care to achieve state-of-the-art performance; for example, LSTMs require pre-segmented data or else they cannot fully exploit temporal information. Our analysis identifies specific limitations for each method that could form the basis of future work. Our experimental results on UCF101 and HMDB51 datasets achieve state-of-the-art performances, 94.1% and 69.0%, respectively, without requiring extensive temporal augmentation.

Learning a robust video Variational Autoencoder (VAE) is essential for reducing video redundancy and facilitating efficient video generation. Directly applying image VAEs to individual frames in isolation can result in temporal inconsistencies and suboptimal compression rates due to a lack of temporal compression. Existing Video VAEs have begun to address temporal compression; however, they often suffer from inadequate reconstruction performance. In this paper, we present a novel and powerful video autoencoder capable of high-fidelity video encoding. First, we observe that entangling spatial and temporal compression by merely extending the image VAE to a 3D VAE can introduce motion blur and detail distortion artifacts. Thus, we propose temporal-aware spatial compression to better encode and decode the spatial information. Additionally, we integrate a lightweight motion compression model for further temporal compression. Second, we propose to leverage the textual information inherent in text-to-video datasets and incorporate text guidance into our model. This significantly enhances reconstruction quality, particularly in terms of detail preservation and temporal stability. Third, we further improve the versatility of our model through joint training on both images and videos, which not only enhances reconstruction quality but also enables the model to perform both image and video autoencoding. Extensive evaluations against strong recent baselines demonstrate the superior performance of our method. The project website can be found at~https://yzxing87.github.io/vae/{https://yzxing87.github.io/vae/}.

Efficient neural representations for dynamic video scenes are critical for applications ranging from video compression to interactive simulations. Yet, existing methods often face challenges related to high memory usage, lengthy training times, and temporal consistency. To address these issues, we introduce a novel neural video representation that combines 3D Gaussian splatting with continuous camera motion modeling. By leveraging Neural ODEs, our approach learns smooth camera trajectories while maintaining an explicit 3D scene representation through Gaussians. Additionally, we introduce a spatiotemporal hierarchical learning strategy, progressively refining spatial and temporal features to enhance reconstruction quality and accelerate convergence. This memory-efficient approach achieves high-quality rendering at impressive speeds. Experimental results show that our hierarchical learning, combined with robust camera motion modeling, captures complex dynamic scenes with strong temporal consistency, achieving state-of-the-art performance across diverse video datasets in both high- and low-motion scenarios.

We consider the problem of localizing a spatio-temporal tube in a video corresponding to a given text query. This is a challenging task that requires the joint and efficient modeling of temporal, spatial and multi-modal interactions. To address this task, we propose TubeDETR, a transformer-based architecture inspired by the recent success of such models for text-conditioned object detection. Our model notably includes: (i) an efficient video and text encoder that models spatial multi-modal interactions over sparsely sampled frames and (ii) a space-time decoder that jointly performs spatio-temporal localization. We demonstrate the advantage of our proposed components through an extensive ablation study. We also evaluate our full approach on the spatio-temporal video grounding task and demonstrate improvements over the state of the art on the challenging VidSTG and HC-STVG benchmarks. Code and trained models are publicly available at https://antoyang.github.io/tubedetr.html.

In this paper, we address the challenges posed by the substantial training time and memory consumption associated with video transformers, focusing on the ViViT (Video Vision Transformer) model, in particular the Factorised Encoder version, as our baseline for action recognition tasks. The factorised encoder variant follows the late-fusion approach that is adopted by many state of the art approaches. Despite standing out for its favorable speed/accuracy tradeoffs among the different variants of ViViT, its considerable training time and memory requirements still pose a significant barrier to entry. Our method is designed to lower this barrier and is based on the idea of freezing the spatial transformer during training. This leads to a low accuracy model if naively done. But we show that by (1) appropriately initializing the temporal transformer (a module responsible for processing temporal information) (2) introducing a compact adapter model connecting frozen spatial representations ((a module that selectively focuses on regions of the input image) to the temporal transformer, we can enjoy the benefits of freezing the spatial transformer without sacrificing accuracy. Through extensive experimentation over 6 benchmarks, we demonstrate that our proposed training strategy significantly reduces training costs (by sim 50%) and memory consumption while maintaining or slightly improving performance by up to 1.79\% compared to the baseline model. Our approach additionally unlocks the capability to utilize larger image transformer models as our spatial transformer and access more frames with the same memory consumption.

Learning to reliably perceive and understand the scene is an integral enabler for robots to operate in the real-world. This problem is inherently challenging due to the multitude of object types as well as appearance changes caused by varying illumination and weather conditions. Leveraging complementary modalities can enable learning of semantically richer representations that are resilient to such perturbations. Despite the tremendous progress in recent years, most multimodal convolutional neural network approaches directly concatenate feature maps from individual modality streams rendering the model incapable of focusing only on relevant complementary information for fusion. To address this limitation, we propose a mutimodal semantic segmentation framework that dynamically adapts the fusion of modality-specific features while being sensitive to the object category, spatial location and scene context in a self-supervised manner. Specifically, we propose an architecture consisting of two modality-specific encoder streams that fuse intermediate encoder representations into a single decoder using our proposed self-supervised model adaptation fusion mechanism which optimally combines complementary features. As intermediate representations are not aligned across modalities, we introduce an attention scheme for better correlation. In addition, we propose a computationally efficient unimodal segmentation architecture termed AdapNet++ that incorporates a new encoder with multiscale residual units and an efficient atrous spatial pyramid pooling that has a larger effective receptive field with more than 10x fewer parameters, complemented with a strong decoder with a multi-resolution supervision scheme that recovers high-resolution details. Comprehensive empirical evaluations on several benchmarks demonstrate that both our unimodal and multimodal architectures achieve state-of-the-art performance.

This work addresses continuous space-time video super-resolution (C-STVSR) that aims to up-scale an input video both spatially and temporally by any scaling factors. One key challenge of C-STVSR is to propagate information temporally among the input video frames. To this end, we introduce a space-time local implicit neural function. It has the striking feature of learning forward motion for a continuum of pixels. We motivate the use of forward motion from the perspective of learning individual motion trajectories, as opposed to learning a mixture of motion trajectories with backward motion. To ease motion interpolation, we encode sparsely sampled forward motion extracted from the input video as the contextual input. Along with a reliability-aware splatting and decoding scheme, our framework, termed MoTIF, achieves the state-of-the-art performance on C-STVSR. The source code of MoTIF is available at https://github.com/sichun233746/MoTIF.

Spatial pyramid pooling module or encode-decoder structure are used in deep neural networks for semantic segmentation task. The former networks are able to encode multi-scale contextual information by probing the incoming features with filters or pooling operations at multiple rates and multiple effective fields-of-view, while the latter networks can capture sharper object boundaries by gradually recovering the spatial information. In this work, we propose to combine the advantages from both methods. Specifically, our proposed model, DeepLabv3+, extends DeepLabv3 by adding a simple yet effective decoder module to refine the segmentation results especially along object boundaries. We further explore the Xception model and apply the depthwise separable convolution to both Atrous Spatial Pyramid Pooling and decoder modules, resulting in a faster and stronger encoder-decoder network. We demonstrate the effectiveness of the proposed model on PASCAL VOC 2012 and Cityscapes datasets, achieving the test set performance of 89.0\% and 82.1\% without any post-processing. Our paper is accompanied with a publicly available reference implementation of the proposed models in Tensorflow at https://github.com/tensorflow/models/tree/master/research/deeplab.

Accurately perceiving and tracking instances over time is essential for the decision-making processes of autonomous agents interacting safely in dynamic environments. With this intention, we propose Mask4Former for the challenging task of 4D panoptic segmentation of LiDAR point clouds. Mask4Former is the first transformer-based approach unifying semantic instance segmentation and tracking of sparse and irregular sequences of 3D point clouds into a single joint model. Our model directly predicts semantic instances and their temporal associations without relying on hand-crafted non-learned association strategies such as probabilistic clustering or voting-based center prediction. Instead, Mask4Former introduces spatio-temporal instance queries that encode the semantic and geometric properties of each semantic tracklet in the sequence. In an in-depth study, we find that promoting spatially compact instance predictions is critical as spatio-temporal instance queries tend to merge multiple semantically similar instances, even if they are spatially distant. To this end, we regress 6-DOF bounding box parameters from spatio-temporal instance queries, which are used as an auxiliary task to foster spatially compact predictions. Mask4Former achieves a new state-of-the-art on the SemanticKITTI test set with a score of 68.4 LSTQ.

This paper aims to improve the performance of video multimodal large language models (MLLM) via long and rich context (LRC) modeling. As a result, we develop a new version of InternVideo2.5 with a focus on enhancing the original MLLMs' ability to perceive fine-grained details and capture long-form temporal structure in videos. Specifically, our approach incorporates dense vision task annotations into MLLMs using direct preference optimization and develops compact spatiotemporal representations through adaptive hierarchical token compression. Experimental results demonstrate this unique design of LRC greatly improves the results of video MLLM in mainstream video understanding benchmarks (short & long), enabling the MLLM to memorize significantly longer video inputs (at least 6x longer than the original), and master specialized vision capabilities like object tracking and segmentation. Our work highlights the importance of multimodal context richness (length and fineness) in empowering MLLM's innate abilites (focus and memory), providing new insights for future research on video MLLM. Code and models are available at https://github.com/OpenGVLab/InternVideo/tree/main/InternVideo2.5

Recent works have successfully extended large-scale text-to-image models to the video domain, producing promising results but at a high computational cost and requiring a large amount of video data. In this work, we introduce ConditionVideo, a training-free approach to text-to-video generation based on the provided condition, video, and input text, by leveraging the power of off-the-shelf text-to-image generation methods (e.g., Stable Diffusion). ConditionVideo generates realistic dynamic videos from random noise or given scene videos. Our method explicitly disentangles the motion representation into condition-guided and scenery motion components. To this end, the ConditionVideo model is designed with a UNet branch and a control branch. To improve temporal coherence, we introduce sparse bi-directional spatial-temporal attention (sBiST-Attn). The 3D control network extends the conventional 2D controlnet model, aiming to strengthen conditional generation accuracy by additionally leveraging the bi-directional frames in the temporal domain. Our method exhibits superior performance in terms of frame consistency, clip score, and conditional accuracy, outperforming other compared methods.

The explosive growth of videos on streaming media platforms has underscored the urgent need for effective video quality assessment (VQA) algorithms to monitor and perceptually optimize the quality of streaming videos. However, VQA remains an extremely challenging task due to the diverse video content and the complex spatial and temporal distortions, thus necessitating more advanced methods to address these issues. Nowadays, large multimodal models (LMMs), such as GPT-4V, have exhibited strong capabilities for various visual understanding tasks, motivating us to leverage the powerful multimodal representation ability of LMMs to solve the VQA task. Therefore, we propose the first Large Multi-Modal Video Quality Assessment (LMM-VQA) model, which introduces a novel spatiotemporal visual modeling strategy for quality-aware feature extraction. Specifically, we first reformulate the quality regression problem into a question and answering (Q&A) task and construct Q&A prompts for VQA instruction tuning. Then, we design a spatiotemporal vision encoder to extract spatial and temporal features to represent the quality characteristics of videos, which are subsequently mapped into the language space by the spatiotemporal projector for modality alignment. Finally, the aligned visual tokens and the quality-inquired text tokens are aggregated as inputs for the large language model (LLM) to generate the quality score and level. Extensive experiments demonstrate that LMM-VQA achieves state-of-the-art performance across five VQA benchmarks, exhibiting an average improvement of 5% in generalization ability over existing methods. Furthermore, due to the advanced design of the spatiotemporal encoder and projector, LMM-VQA also performs exceptionally well on general video understanding tasks, further validating its effectiveness. Our code will be released at https://github.com/Sueqk/LMM-VQA.

This paper presents the Large Vision Diffusion Transformer (LaVin-DiT), a scalable and unified foundation model designed to tackle over 20 computer vision tasks in a generative framework. Unlike existing large vision models directly adapted from natural language processing architectures, which rely on less efficient autoregressive techniques and disrupt spatial relationships essential for vision data, LaVin-DiT introduces key innovations to optimize generative performance for vision tasks. First, to address the high dimensionality of visual data, we incorporate a spatial-temporal variational autoencoder that encodes data into a continuous latent space. Second, for generative modeling, we develop a joint diffusion transformer that progressively produces vision outputs. Third, for unified multi-task training, in-context learning is implemented. Input-target pairs serve as task context, which guides the diffusion transformer to align outputs with specific tasks within the latent space. During inference, a task-specific context set and test data as queries allow LaVin-DiT to generalize across tasks without fine-tuning. Trained on extensive vision datasets, the model is scaled from 0.1B to 3.4B parameters, demonstrating substantial scalability and state-of-the-art performance across diverse vision tasks. This work introduces a novel pathway for large vision foundation models, underscoring the promising potential of diffusion transformers. The code and models will be open-sourced.

We present an efficient encoder-free approach for video-language understanding that achieves competitive performance while significantly reducing computational overhead. Current video-language models typically rely on heavyweight image encoders (300M-1.1B parameters) or video encoders (1B-1.4B parameters), creating a substantial computational burden when processing multi-frame videos. Our method introduces a novel Spatio-Temporal Alignment Block (STAB) that directly processes video inputs without requiring pre-trained encoders while using only 45M parameters for visual processing - at least a 6.5times reduction compared to traditional approaches. The STAB architecture combines Local Spatio-Temporal Encoding for fine-grained feature extraction, efficient spatial downsampling through learned attention and separate mechanisms for modeling frame-level and video-level relationships. Our model achieves comparable or superior performance to encoder-based approaches for open-ended video question answering on standard benchmarks. The fine-grained video question-answering evaluation demonstrates our model's effectiveness, outperforming the encoder-based approaches Video-ChatGPT and Video-LLaVA in key aspects like correctness and temporal understanding. Extensive ablation studies validate our architectural choices and demonstrate the effectiveness of our spatio-temporal modeling approach while achieving 3-4times faster processing speeds than previous methods. Code is available at https://github.com/jh-yi/Video-Panda.

LiDAR-based semantic perception tasks are critical yet challenging for autonomous driving. Due to the motion of objects and static/dynamic occlusion, temporal information plays an essential role in reinforcing perception by enhancing and completing single-frame knowledge. Previous approaches either directly stack historical frames to the current frame or build a 4D spatio-temporal neighborhood using KNN, which duplicates computation and hinders realtime performance. Based on our observation that stacking all the historical points would damage performance due to a large amount of redundant and misleading information, we propose the Sparse Voxel-Adjacent Query Network (SVQNet) for 4D LiDAR semantic segmentation. To take full advantage of the historical frames high-efficiently, we shunt the historical points into two groups with reference to the current points. One is the Voxel-Adjacent Neighborhood carrying local enhancing knowledge. The other is the Historical Context completing the global knowledge. Then we propose new modules to select and extract the instructive features from the two groups. Our SVQNet achieves state-of-the-art performance in LiDAR semantic segmentation of the SemanticKITTI benchmark and the nuScenes dataset.

Despite progress in video-language modeling, the computational challenge of interpreting long-form videos in response to task-specific linguistic queries persists, largely due to the complexity of high-dimensional video data and the misalignment between language and visual cues over space and time. To tackle this issue, we introduce a novel approach called Language-guided Spatial-Temporal Prompt Learning (LSTP). This approach features two key components: a Temporal Prompt Sampler (TPS) with optical flow prior that leverages temporal information to efficiently extract relevant video content, and a Spatial Prompt Solver (SPS) that adeptly captures the intricate spatial relationships between visual and textual elements. By harmonizing TPS and SPS with a cohesive training strategy, our framework significantly enhances computational efficiency, temporal understanding, and spatial-temporal alignment. Empirical evaluations across two challenging tasks--video question answering and temporal question grounding in videos--using a variety of video-language pretrainings (VLPs) and large language models (LLMs) demonstrate the superior performance, speed, and versatility of our proposed LSTP paradigm.

How to learn discriminative video representation from unlabeled videos is challenging but crucial for video analysis. The latest attempts seek to learn a representation model by predicting the appearance contents in the masked regions. However, simply masking and recovering appearance contents may not be sufficient to model temporal clues as the appearance contents can be easily reconstructed from a single frame. To overcome this limitation, we present Masked Motion Encoding (MME), a new pre-training paradigm that reconstructs both appearance and motion information to explore temporal clues. In MME, we focus on addressing two critical challenges to improve the representation performance: 1) how to well represent the possible long-term motion across multiple frames; and 2) how to obtain fine-grained temporal clues from sparsely sampled videos. Motivated by the fact that human is able to recognize an action by tracking objects' position changes and shape changes, we propose to reconstruct a motion trajectory that represents these two kinds of change in the masked regions. Besides, given the sparse video input, we enforce the model to reconstruct dense motion trajectories in both spatial and temporal dimensions. Pre-trained with our MME paradigm, the model is able to anticipate long-term and fine-grained motion details. Code is available at https://github.com/XinyuSun/MME.

Vision-language large models have achieved remarkable success in various multi-modal tasks, yet applying them to video understanding remains challenging due to the inherent complexity and computational demands of video data. While training-based video-LLMs deliver high performance, they often require substantial resources for training and inference. Conversely, training-free approaches offer a more efficient alternative by adapting pre-trained image-LLMs models for video tasks without additional training, but they face inference efficiency bottlenecks due to the large number of visual tokens generated from video frames. In this work, we present a novel prompt-guided visual perception framework (abbreviated as Free Video-LLM) for efficient inference of training-free video LLMs. The proposed framework decouples spatial-temporal dimension and performs temporal frame sampling and spatial RoI cropping respectively based on task-specific prompts. Our method effectively reduces the number of visual tokens while maintaining high performance across multiple video question-answering benchmarks. Extensive experiments demonstrate that our approach achieves competitive results with significantly fewer tokens, offering an optimal trade-off between accuracy and computational efficiency compared to state-of-the-art video LLMs. The code will be available at https://github.com/contrastive/FreeVideoLLM.

Effective modeling of complex spatiotemporal dependencies in long-form videos remains an open problem. The recently proposed Structured State-Space Sequence (S4) model with its linear complexity offers a promising direction in this space. However, we demonstrate that treating all image-tokens equally as done by S4 model can adversely affect its efficiency and accuracy. To address this limitation, we present a novel Selective S4 (i.e., S5) model that employs a lightweight mask generator to adaptively select informative image tokens resulting in more efficient and accurate modeling of long-term spatiotemporal dependencies in videos. Unlike previous mask-based token reduction methods used in transformers, our S5 model avoids the dense self-attention calculation by making use of the guidance of the momentum-updated S4 model. This enables our model to efficiently discard less informative tokens and adapt to various long-form video understanding tasks more effectively. However, as is the case for most token reduction methods, the informative image tokens could be dropped incorrectly. To improve the robustness and the temporal horizon of our model, we propose a novel long-short masked contrastive learning (LSMCL) approach that enables our model to predict longer temporal context using shorter input videos. We present extensive comparative results using three challenging long-form video understanding datasets (LVU, COIN and Breakfast), demonstrating that our approach consistently outperforms the previous state-of-the-art S4 model by up to 9.6% accuracy while reducing its memory footprint by 23%.

Advancements in language foundation models have primarily fueled the recent surge in artificial intelligence. In contrast, generative learning of non-textual modalities, especially videos, significantly trails behind language modeling. This thesis chronicles our endeavor to build multi-task models for generating videos and other modalities under diverse conditions, as well as for understanding and compression applications. Given the high dimensionality of visual data, we pursue concise and accurate latent representations. Our video-native spatial-temporal tokenizers preserve high fidelity. We unveil a novel approach to mapping bidirectionally between visual observation and interpretable lexical terms. Furthermore, our scalable visual token representation proves beneficial across generation, compression, and understanding tasks. This achievement marks the first instances of language models surpassing diffusion models in visual synthesis and a video tokenizer outperforming industry-standard codecs. Within these multi-modal latent spaces, we study the design of multi-task generative models. Our masked multi-task transformer excels at the quality, efficiency, and flexibility of video generation. We enable a frozen language model, trained solely on text, to generate visual content. Finally, we build a scalable generative multi-modal transformer trained from scratch, enabling the generation of videos containing high-fidelity motion with the corresponding audio given diverse conditions. Throughout the course, we have shown the effectiveness of integrating multiple tasks, crafting high-fidelity latent representation, and generating multiple modalities. This work suggests intriguing potential for future exploration in generating non-textual data and enabling real-time, interactive experiences across various media forms.

Temporal interpolation often plays a crucial role to learn meaningful representations in dynamic scenes. In this paper, we propose a novel method to train spatiotemporal neural radiance fields of dynamic scenes based on temporal interpolation of feature vectors. Two feature interpolation methods are suggested depending on underlying representations, neural networks or grids. In the neural representation, we extract features from space-time inputs via multiple neural network modules and interpolate them based on time frames. The proposed multi-level feature interpolation network effectively captures features of both short-term and long-term time ranges. In the grid representation, space-time features are learned via four-dimensional hash grids, which remarkably reduces training time. The grid representation shows more than 100 times faster training speed than the previous neural-net-based methods while maintaining the rendering quality. Concatenating static and dynamic features and adding a simple smoothness term further improve the performance of our proposed models. Despite the simplicity of the model architectures, our method achieved state-of-the-art performance both in rendering quality for the neural representation and in training speed for the grid representation.

We propose a flow-guided transformer, which innovatively leverage the motion discrepancy exposed by optical flows to instruct the attention retrieval in transformer for high fidelity video inpainting. More specially, we design a novel flow completion network to complete the corrupted flows by exploiting the relevant flow features in a local temporal window. With the completed flows, we propagate the content across video frames, and adopt the flow-guided transformer to synthesize the rest corrupted regions. We decouple transformers along temporal and spatial dimension, so that we can easily integrate the locally relevant completed flows to instruct spatial attention only. Furthermore, we design a flow-reweight module to precisely control the impact of completed flows on each spatial transformer. For the sake of efficiency, we introduce window partition strategy to both spatial and temporal transformers. Especially in spatial transformer, we design a dual perspective spatial MHSA, which integrates the global tokens to the window-based attention. Extensive experiments demonstrate the effectiveness of the proposed method qualitatively and quantitatively. Codes are available at https://github.com/hitachinsk/FGT.

This paper proposes a novel pretext task to address the self-supervised video representation learning problem. Specifically, given an unlabeled video clip, we compute a series of spatio-temporal statistical summaries, such as the spatial location and dominant direction of the largest motion, the spatial location and dominant color of the largest color diversity along the temporal axis, etc. Then a neural network is built and trained to yield the statistical summaries given the video frames as inputs. In order to alleviate the learning difficulty, we employ several spatial partitioning patterns to encode rough spatial locations instead of exact spatial Cartesian coordinates. Our approach is inspired by the observation that human visual system is sensitive to rapidly changing contents in the visual field, and only needs impressions about rough spatial locations to understand the visual contents. To validate the effectiveness of the proposed approach, we conduct extensive experiments with four 3D backbone networks, i.e., C3D, 3D-ResNet, R(2+1)D and S3D-G. The results show that our approach outperforms the existing approaches across these backbone networks on four downstream video analysis tasks including action recognition, video retrieval, dynamic scene recognition, and action similarity labeling. The source code is publicly available at: https://github.com/laura-wang/video_repres_sts.

Recently, Vision Large Language Models (VLLMs) integrated with vision encoders have shown promising performance in vision understanding. The key of VLLMs is to encode visual content into sequences of visual tokens, enabling VLLMs to simultaneously process both visual and textual content. However, understanding videos, especially long videos, remain a challenge to VLLMs as the number of visual tokens grows rapidly when encoding videos, resulting in the risk of exceeding the context window of VLLMs and introducing heavy computation burden. To restrict the number of visual tokens, existing VLLMs either: (1) uniformly downsample videos into a fixed number of frames or (2) reducing the number of visual tokens encoded from each frame. We argue the former solution neglects the rich temporal cue in videos and the later overlooks the spatial details in each frame. In this work, we present Balanced-VLLM (B-VLLM): a novel VLLM framework that aims to effectively leverage task relevant spatio-temporal cues while restricting the number of visual tokens under the VLLM context window length. At the core of our method, we devise a text-conditioned adaptive frame selection module to identify frames relevant to the visual understanding task. The selected frames are then de-duplicated using a temporal frame token merging technique. The visual tokens of the selected frames are processed through a spatial token sampling module and an optional spatial token merging strategy to achieve precise control over the token count. Experimental results show that B-VLLM is effective in balancing the number of frames and visual tokens in video understanding, yielding superior performance on various video understanding benchmarks. Our code is available at https://github.com/zhuqiangLu/B-VLLM.

In text-video retrieval, recent works have benefited from the powerful learning capabilities of pre-trained text-image foundation models (e.g., CLIP) by adapting them to the video domain. A critical problem for them is how to effectively capture the rich semantics inside the video using the image encoder of CLIP. To tackle this, state-of-the-art methods adopt complex cross-modal modeling techniques to fuse the text information into video frame representations, which, however, incurs severe efficiency issues in large-scale retrieval systems as the video representations must be recomputed online for every text query. In this paper, we discard this problematic cross-modal fusion process and aim to learn semantically-enhanced representations purely from the video, so that the video representations can be computed offline and reused for different texts. Concretely, we first introduce a spatial-temporal "Prompt Cube" into the CLIP image encoder and iteratively switch it within the encoder layers to efficiently incorporate the global video semantics into frame representations. We then propose to apply an auxiliary video captioning objective to train the frame representations, which facilitates the learning of detailed video semantics by providing fine-grained guidance in the semantic space. With a naive temporal fusion strategy (i.e., mean-pooling) on the enhanced frame representations, we obtain state-of-the-art performances on three benchmark datasets, i.e., MSR-VTT, MSVD, and LSMDC.

Deep saliency prediction algorithms complement the object recognition features, they typically rely on additional information, such as scene context, semantic relationships, gaze direction, and object dissimilarity. However, none of these models consider the temporal nature of gaze shifts during image observation. We introduce a novel saliency prediction model that learns to output saliency maps in sequential time intervals by exploiting human temporal attention patterns. Our approach locally modulates the saliency predictions by combining the learned temporal maps. Our experiments show that our method outperforms the state-of-the-art models, including a multi-duration saliency model, on the SALICON benchmark. Our code will be publicly available on GitHub.

Convolutional neural networks (CNNs) are usually built by stacking convolutional operations layer-by-layer. Although CNN has shown strong capability to extract semantics from raw pixels, its capacity to capture spatial relationships of pixels across rows and columns of an image is not fully explored. These relationships are important to learn semantic objects with strong shape priors but weak appearance coherences, such as traffic lanes, which are often occluded or not even painted on the road surface as shown in Fig. 1 (a). In this paper, we propose Spatial CNN (SCNN), which generalizes traditional deep layer-by-layer convolutions to slice-byslice convolutions within feature maps, thus enabling message passings between pixels across rows and columns in a layer. Such SCNN is particular suitable for long continuous shape structure or large objects, with strong spatial relationship but less appearance clues, such as traffic lanes, poles, and wall. We apply SCNN on a newly released very challenging traffic lane detection dataset and Cityscapse dataset. The results show that SCNN could learn the spatial relationship for structure output and significantly improves the performance. We show that SCNN outperforms the recurrent neural network (RNN) based ReNet and MRF+CNN (MRFNet) in the lane detection dataset by 8.7% and 4.6% respectively. Moreover, our SCNN won the 1st place on the TuSimple Benchmark Lane Detection Challenge, with an accuracy of 96.53%.

Predicting diverse human motions given a sequence of historical poses has received increasing attention. Despite rapid progress, existing work captures the multi-modal nature of human motions primarily through likelihood-based sampling, where the mode collapse has been widely observed. In this paper, we propose a simple yet effective approach that disentangles randomly sampled codes with a deterministic learnable component named anchors to promote sample precision and diversity. Anchors are further factorized into spatial anchors and temporal anchors, which provide attractively interpretable control over spatial-temporal disparity. In principle, our spatial-temporal anchor-based sampling (STARS) can be applied to different motion predictors. Here we propose an interaction-enhanced spatial-temporal graph convolutional network (IE-STGCN) that encodes prior knowledge of human motions (e.g., spatial locality), and incorporate the anchors into it. Extensive experiments demonstrate that our approach outperforms state of the art in both stochastic and deterministic prediction, suggesting it as a unified framework for modeling human motions. Our code and pretrained models are available at https://github.com/Sirui-Xu/STARS.

In this research, we present a novel approach to motion customization in video generation, addressing the widespread gap in the thorough exploration of motion representation within video generative models. Recognizing the unique challenges posed by video's spatiotemporal nature, our method introduces Motion Embeddings, a set of explicit, temporally coherent one-dimensional embeddings derived from a given video. These embeddings are designed to integrate seamlessly with the temporal transformer modules of video diffusion models, modulating self-attention computations across frames without compromising spatial integrity. Our approach offers a compact and efficient solution to motion representation and enables complex manipulations of motion characteristics through vector arithmetic in the embedding space. Furthermore, we identify the Temporal Discrepancy in video generative models, which refers to variations in how different motion modules process temporal relationships between frames. We leverage this understanding to optimize the integration of our motion embeddings. Our contributions include the introduction of a tailored motion embedding for customization tasks, insights into the temporal processing differences in video models, and a demonstration of the practical advantages and effectiveness of our method through extensive experiments.

We present a convolution-free approach to video classification built exclusively on self-attention over space and time. Our method, named "TimeSformer," adapts the standard Transformer architecture to video by enabling spatiotemporal feature learning directly from a sequence of frame-level patches. Our experimental study compares different self-attention schemes and suggests that "divided attention," where temporal attention and spatial attention are separately applied within each block, leads to the best video classification accuracy among the design choices considered. Despite the radically new design, TimeSformer achieves state-of-the-art results on several action recognition benchmarks, including the best reported accuracy on Kinetics-400 and Kinetics-600. Finally, compared to 3D convolutional networks, our model is faster to train, it can achieve dramatically higher test efficiency (at a small drop in accuracy), and it can also be applied to much longer video clips (over one minute long). Code and models are available at: https://github.com/facebookresearch/TimeSformer.

Current vision systems typically assign fixed-length representations to images, regardless of the information content. This contrasts with human intelligence - and even large language models - which allocate varying representational capacities based on entropy, context and familiarity. Inspired by this, we propose an approach to learn variable-length token representations for 2D images. Our encoder-decoder architecture recursively processes 2D image tokens, distilling them into 1D latent tokens over multiple iterations of recurrent rollouts. Each iteration refines the 2D tokens, updates the existing 1D latent tokens, and adaptively increases representational capacity by adding new tokens. This enables compression of images into a variable number of tokens, ranging from 32 to 256. We validate our tokenizer using reconstruction loss and FID metrics, demonstrating that token count aligns with image entropy, familiarity and downstream task requirements. Recurrent token processing with increasing representational capacity in each iteration shows signs of token specialization, revealing potential for object / part discovery.

We introduce Sketch-based Video Object Localization (SVOL), a new task aimed at localizing spatio-temporal object boxes in video queried by the input sketch. We first outline the challenges in the SVOL task and build the Sketch-Video Attention Network (SVANet) with the following design principles: (i) to consider temporal information of video and bridge the domain gap between sketch and video; (ii) to accurately identify and localize multiple objects simultaneously; (iii) to handle various styles of sketches; (iv) to be classification-free. In particular, SVANet is equipped with a Cross-modal Transformer that models the interaction between learnable object tokens, query sketch, and video through attention operations, and learns upon a per-frame set matching strategy that enables frame-wise prediction while utilizing global video context. We evaluate SVANet on a newly curated SVOL dataset. By design, SVANet successfully learns the mapping between the query sketches and video objects, achieving state-of-the-art results on the SVOL benchmark. We further confirm the effectiveness of SVANet via extensive ablation studies and visualizations. Lastly, we demonstrate its transfer capability on unseen datasets and novel categories, suggesting its high scalability in real-world applications.

We introduce Ferret, a new Multimodal Large Language Model (MLLM) capable of understanding spatial referring of any shape or granularity within an image and accurately grounding open-vocabulary descriptions. To unify referring and grounding in the LLM paradigm, Ferret employs a novel and powerful hybrid region representation that integrates discrete coordinates and continuous features jointly to represent a region in the image. To extract the continuous features of versatile regions, we propose a spatial-aware visual sampler, adept at handling varying sparsity across different shapes. Consequently, Ferret can accept diverse region inputs, such as points, bounding boxes, and free-form shapes. To bolster the desired capability of Ferret, we curate GRIT, a comprehensive refer-and-ground instruction tuning dataset including 1.1M samples that contain rich hierarchical spatial knowledge, with 95K hard negative data to promote model robustness. The resulting model not only achieves superior performance in classical referring and grounding tasks, but also greatly outperforms existing MLLMs in region-based and localization-demanded multimodal chatting. Our evaluations also reveal a significantly improved capability of describing image details and a remarkable alleviation in object hallucination. Code and data will be available at https://github.com/apple/ml-ferret

We present a unified perspective on tackling various human-centric video tasks by learning human motion representations from large-scale and heterogeneous data resources. Specifically, we propose a pretraining stage in which a motion encoder is trained to recover the underlying 3D motion from noisy partial 2D observations. The motion representations acquired in this way incorporate geometric, kinematic, and physical knowledge about human motion, which can be easily transferred to multiple downstream tasks. We implement the motion encoder with a Dual-stream Spatio-temporal Transformer (DSTformer) neural network. It could capture long-range spatio-temporal relationships among the skeletal joints comprehensively and adaptively, exemplified by the lowest 3D pose estimation error so far when trained from scratch. Furthermore, our proposed framework achieves state-of-the-art performance on all three downstream tasks by simply finetuning the pretrained motion encoder with a simple regression head (1-2 layers), which demonstrates the versatility of the learned motion representations.

We present VideoFactory, an innovative framework for generating high-quality open-domain videos. VideoFactory excels in producing high-definition (1376x768), widescreen (16:9) videos without watermarks, creating an engaging user experience. Generating videos guided by text instructions poses significant challenges, such as modeling the complex relationship between space and time, and the lack of large-scale text-video paired data. Previous approaches extend pretrained text-to-image generation models by adding temporal 1D convolution/attention modules for video generation. However, these approaches overlook the importance of jointly modeling space and time, inevitably leading to temporal distortions and misalignment between texts and videos. In this paper, we propose a novel approach that strengthens the interaction between spatial and temporal perceptions. In particular, we utilize a swapped cross-attention mechanism in 3D windows that alternates the "query" role between spatial and temporal blocks, enabling mutual reinforcement for each other. To fully unlock model capabilities for high-quality video generation, we curate a large-scale video dataset called HD-VG-130M. This dataset comprises 130 million text-video pairs from the open-domain, ensuring high-definition, widescreen and watermark-free characters. Objective metrics and user studies demonstrate the superiority of our approach in terms of per-frame quality, temporal correlation, and text-video alignment, with clear margins.

In this paper, we present BEVerse, a unified framework for 3D perception and prediction based on multi-camera systems. Unlike existing studies focusing on the improvement of single-task approaches, BEVerse features in producing spatio-temporal Birds-Eye-View (BEV) representations from multi-camera videos and jointly reasoning about multiple tasks for vision-centric autonomous driving. Specifically, BEVerse first performs shared feature extraction and lifting to generate 4D BEV representations from multi-timestamp and multi-view images. After the ego-motion alignment, the spatio-temporal encoder is utilized for further feature extraction in BEV. Finally, multiple task decoders are attached for joint reasoning and prediction. Within the decoders, we propose the grid sampler to generate BEV features with different ranges and granularities for different tasks. Also, we design the method of iterative flow for memory-efficient future prediction. We show that the temporal information improves 3D object detection and semantic map construction, while the multi-task learning can implicitly benefit motion prediction. With extensive experiments on the nuScenes dataset, we show that the multi-task BEVerse outperforms existing single-task methods on 3D object detection, semantic map construction, and motion prediction. Compared with the sequential paradigm, BEVerse also favors in significantly improved efficiency. The code and trained models will be released at https://github.com/zhangyp15/BEVerse.

This paper tackles the problem of semi-supervised video object segmentation on resource-constrained devices, such as mobile phones. We formulate this problem as a distillation task, whereby we demonstrate that small space-time-memory networks with finite memory can achieve competitive results with state of the art, but at a fraction of the computational cost (32 milliseconds per frame on a Samsung Galaxy S22). Specifically, we provide a theoretically grounded framework that unifies knowledge distillation with supervised contrastive representation learning. These models are able to jointly benefit from both pixel-wise contrastive learning and distillation from a pre-trained teacher. We validate this loss by achieving competitive J&F to state of the art on both the standard DAVIS and YouTube benchmarks, despite running up to 5x faster, and with 32x fewer parameters.

Detecting abnormal activities in real-world surveillance videos is an important yet challenging task as the prior knowledge about video anomalies is usually limited or unavailable. Despite that many approaches have been developed to resolve this problem, few of them can capture the normal spatio-temporal patterns effectively and efficiently. Moreover, existing works seldom explicitly consider the local consistency at frame level and global coherence of temporal dynamics in video sequences. To this end, we propose Convolutional Transformer based Dual Discriminator Generative Adversarial Networks (CT-D2GAN) to perform unsupervised video anomaly detection. Specifically, we first present a convolutional transformer to perform future frame prediction. It contains three key components, i.e., a convolutional encoder to capture the spatial information of the input video clips, a temporal self-attention module to encode the temporal dynamics, and a convolutional decoder to integrate spatio-temporal features and predict the future frame. Next, a dual discriminator based adversarial training procedure, which jointly considers an image discriminator that can maintain the local consistency at frame-level and a video discriminator that can enforce the global coherence of temporal dynamics, is employed to enhance the future frame prediction. Finally, the prediction error is used to identify abnormal video frames. Thoroughly empirical studies on three public video anomaly detection datasets, i.e., UCSD Ped2, CUHK Avenue, and Shanghai Tech Campus, demonstrate the effectiveness of the proposed adversarial spatio-temporal modeling framework.

In this paper we discuss several forms of spatiotemporal convolutions for video analysis and study their effects on action recognition. Our motivation stems from the observation that 2D CNNs applied to individual frames of the video have remained solid performers in action recognition. In this work we empirically demonstrate the accuracy advantages of 3D CNNs over 2D CNNs within the framework of residual learning. Furthermore, we show that factorizing the 3D convolutional filters into separate spatial and temporal components yields significantly advantages in accuracy. Our empirical study leads to the design of a new spatiotemporal convolutional block "R(2+1)D" which gives rise to CNNs that achieve results comparable or superior to the state-of-the-art on Sports-1M, Kinetics, UCF101 and HMDB51.

Continuous space-time video super-resolution (C-STVSR) aims to simultaneously enhance video resolution and frame rate at an arbitrary scale. Recently, implicit neural representation (INR) has been applied to video restoration, representing videos as implicit fields that can be decoded at an arbitrary scale. However, the highly ill-posed nature of C-STVSR limits the effectiveness of current INR-based methods: they assume linear motion between frames and use interpolation or feature warping to generate features at arbitrary spatiotemporal positions with two consecutive frames. This restrains C-STVSR from capturing rapid and nonlinear motion and long-term dependencies (involving more than two frames) in complex dynamic scenes. In this paper, we propose a novel C-STVSR framework, called HR-INR, which captures both holistic dependencies and regional motions based on INR. It is assisted by an event camera, a novel sensor renowned for its high temporal resolution and low latency. To fully utilize the rich temporal information from events, we design a feature extraction consisting of (1) a regional event feature extractor - taking events as inputs via the proposed event temporal pyramid representation to capture the regional nonlinear motion and (2) a holistic event-frame feature extractor for long-term dependence and continuity motion. We then propose a novel INR-based decoder with spatiotemporal embeddings to capture long-term dependencies with a larger temporal perception field. We validate the effectiveness and generalization of our method on four datasets (both simulated and real data), showing the superiority of our method.

There has been tremendous progress in multimodal Large Language Models (LLMs). Recent works have extended these models to video input with promising instruction following capabilities. However, an important missing piece is temporal localization. These models cannot accurately answer the "When?" questions. We identify three key aspects that limit their temporal localization capabilities: (i) time representation, (ii) architecture, and (iii) data. We address these shortcomings by proposing Language Instructed Temporal-Localization Assistant (LITA) with the following features: (1) We introduce time tokens that encode timestamps relative to the video length to better represent time in videos. (2) We introduce SlowFast tokens in the architecture to capture temporal information at fine temporal resolution. (3) We emphasize temporal localization data for LITA. In addition to leveraging existing video datasets with timestamps, we propose a new task, Reasoning Temporal Localization (RTL), along with the dataset, ActivityNet-RTL, for learning and evaluating this task. Reasoning temporal localization requires both the reasoning and temporal localization of Video LLMs. LITA demonstrates strong performance on this challenging task, nearly doubling the temporal mean intersection-over-union (mIoU) of baselines. In addition, we show that our emphasis on temporal localization also substantially improves video-based text generation compared to existing Video LLMs, including a 36% relative improvement of Temporal Understanding. Code is available at: https://github.com/NVlabs/LITA

Existing video-language pre-training methods primarily focus on instance-level alignment between video clips and captions via global contrastive learning but neglect rich fine-grained local information in both videos and text, which is of importance to downstream tasks requiring temporal localization and semantic reasoning. A powerful model is expected to be capable of capturing region-object correspondences and recognizing scene changes in a video clip, reflecting spatial and temporal granularity, respectively. To strengthen model's understanding into such fine-grained details, we propose a simple yet effective video-language modeling framework, S-ViLM, by exploiting the intrinsic structures of these two modalities. It includes two novel designs, inter-clip spatial grounding and intra-clip temporal grouping, to promote learning region-object alignment and temporal-aware features, simultaneously. Comprehensive evaluations demonstrate that S-ViLM performs favorably against existing approaches in learning more expressive representations. Specifically, S-ViLM surpasses the state-of-the-art methods substantially on four representative downstream tasks, covering text-video retrieval, video question answering, video action recognition, and temporal action localization.

Object discovery is a core task in computer vision. While fast progresses have been made in supervised object detection, its unsupervised counterpart remains largely unexplored. With the growth of data volume, the expensive cost of annotations is the major limitation hindering further study. Therefore, discovering objects without annotations has great significance. However, this task seems impractical on still-image or point cloud alone due to the lack of discriminative information. Previous studies underlook the crucial temporal information and constraints naturally behind multi-modal inputs. In this paper, we propose 4D unsupervised object discovery, jointly discovering objects from 4D data -- 3D point clouds and 2D RGB images with temporal information. We present the first practical approach for this task by proposing a ClusterNet on 3D point clouds, which is jointly iteratively optimized with a 2D localization network. Extensive experiments on the large-scale Waymo Open Dataset suggest that the localization network and ClusterNet achieve competitive performance on both class-agnostic 2D object detection and 3D instance segmentation, bridging the gap between unsupervised methods and full supervised ones. Codes and models will be made available at https://github.com/Robertwyq/LSMOL.

In this paper, we target at the problem of learning a generalizable dynamic radiance field from monocular videos. Different from most existing NeRF methods that are based on multiple views, monocular videos only contain one view at each timestamp, thereby suffering from ambiguity along the view direction in estimating point features and scene flows. Previous studies such as DynNeRF disambiguate point features by positional encoding, which is not transferable and severely limits the generalization ability. As a result, these methods have to train one independent model for each scene and suffer from heavy computational costs when applying to increasing monocular videos in real-world applications. To address this, We propose MonoNeRF to simultaneously learn point features and scene flows with point trajectory and feature correspondence constraints across frames. More specifically, we learn an implicit velocity field to estimate point trajectory from temporal features with Neural ODE, which is followed by a flow-based feature aggregation module to obtain spatial features along the point trajectory. We jointly optimize temporal and spatial features in an end-to-end manner. Experiments show that our MonoNeRF is able to learn from multiple scenes and support new applications such as scene editing, unseen frame synthesis, and fast novel scene adaptation. Codes are available at https://github.com/tianfr/MonoNeRF.

Vision-Language Models (VLMs) have achieved remarkable success across a range of multimodal tasks; however, their practical deployment is often constrained by high computational costs and prolonged inference times. Since the vision modality typically carries more information than the text modality, compressing visual prompts offers a promising solution to alleviate these challenges. Existing approaches predominantly focus on refining model architectures or directly reducing the number of visual tokens. However, these methods often compromise inference performance due to a lack of consideration for the unique spatial and temporal characteristics of visual data. In this work, we propose a token compression paradigm that operates on both spatial and temporal dimensions. Our approach includes a learning-free, plug-and-play compression pipeline that can be seamlessly integrated into most Multimodal Large Language Model (MLLM) frameworks. By leveraging this method, we enhance the model inference capability while simultaneously reducing its computational cost. Experimental results on the Video-QA task demonstrate the effectiveness of the proposed approach, showcasing significant improvements in efficiency without sacrificing performance.

We introduce VideoMamba, a novel adaptation of the pure Mamba architecture, specifically designed for video recognition. Unlike transformers that rely on self-attention mechanisms leading to high computational costs by quadratic complexity, VideoMamba leverages Mamba's linear complexity and selective SSM mechanism for more efficient processing. The proposed Spatio-Temporal Forward and Backward SSM allows the model to effectively capture the complex relationship between non-sequential spatial and sequential temporal information in video. Consequently, VideoMamba is not only resource-efficient but also effective in capturing long-range dependency in videos, demonstrated by competitive performance and outstanding efficiency on a variety of video understanding benchmarks. Our work highlights the potential of VideoMamba as a powerful tool for video understanding, offering a simple yet effective baseline for future research in video analysis.

Enhancing low-resolution, low-frame-rate videos to high-resolution, high-frame-rate quality is essential for a seamless user experience, motivating advancements in Continuous Spatial-Temporal Video Super Resolution (C-STVSR). While prior methods employ Implicit Neural Representation (INR) for continuous encoding, they often struggle to capture the complexity of video data, relying on simple coordinate concatenation and pre-trained optical flow network for motion representation. Interestingly, we find that adding position encoding, contrary to common observations, does not improve-and even degrade performance. This issue becomes particularly pronounced when combined with pre-trained optical flow networks, which can limit the model's flexibility. To address these issues, we propose BF-STVSR, a C-STVSR framework with two key modules tailored to better represent spatial and temporal characteristics of video: 1) B-spline Mapper for smooth temporal interpolation, and 2) Fourier Mapper for capturing dominant spatial frequencies. Our approach achieves state-of-the-art PSNR and SSIM performance, showing enhanced spatial details and natural temporal consistency.

In this paper, we address the semantic segmentation problem with a focus on the context aggregation strategy. Our motivation is that the label of a pixel is the category of the object that the pixel belongs to. We present a simple yet effective approach, object-contextual representations, characterizing a pixel by exploiting the representation of the corresponding object class. First, we learn object regions under the supervision of ground-truth segmentation. Second, we compute the object region representation by aggregating the representations of the pixels lying in the object region. Last, % the representation similarity we compute the relation between each pixel and each object region and augment the representation of each pixel with the object-contextual representation which is a weighted aggregation of all the object region representations according to their relations with the pixel. We empirically demonstrate that the proposed approach achieves competitive performance on various challenging semantic segmentation benchmarks: Cityscapes, ADE20K, LIP, PASCAL-Context, and COCO-Stuff. Cityscapes, ADE20K, LIP, PASCAL-Context, and COCO-Stuff. Our submission "HRNet + OCR + SegFix" achieves 1-st place on the Cityscapes leaderboard by the time of submission. Code is available at: https://git.io/openseg and https://git.io/HRNet.OCR. We rephrase the object-contextual representation scheme using the Transformer encoder-decoder framework. The details are presented in~Section3.3.

Our work explores temporal self-supervision for GAN-based video generation tasks. While adversarial training successfully yields generative models for a variety of areas, temporal relationships in the generated data are much less explored. Natural temporal changes are crucial for sequential generation tasks, e.g. video super-resolution and unpaired video translation. For the former, state-of-the-art methods often favor simpler norm losses such as L^2 over adversarial training. However, their averaging nature easily leads to temporally smooth results with an undesirable lack of spatial detail. For unpaired video translation, existing approaches modify the generator networks to form spatio-temporal cycle consistencies. In contrast, we focus on improving learning objectives and propose a temporally self-supervised algorithm. For both tasks, we show that temporal adversarial learning is key to achieving temporally coherent solutions without sacrificing spatial detail. We also propose a novel Ping-Pong loss to improve the long-term temporal consistency. It effectively prevents recurrent networks from accumulating artifacts temporally without depressing detailed features. Additionally, we propose a first set of metrics to quantitatively evaluate the accuracy as well as the perceptual quality of the temporal evolution. A series of user studies confirm the rankings computed with these metrics. Code, data, models, and results are provided at https://github.com/thunil/TecoGAN. The project page https://ge.in.tum.de/publications/2019-tecogan-chu/ contains supplemental materials.

In this paper, we re-examine the task of cross-modal clip-sentence retrieval, where the clip is part of a longer untrimmed video. When the clip is short or visually ambiguous, knowledge of its local temporal context (i.e. surrounding video segments) can be used to improve the retrieval performance. We propose Context Transformer (ConTra); an encoder architecture that models the interaction between a video clip and its local temporal context in order to enhance its embedded representations. Importantly, we supervise the context transformer using contrastive losses in the cross-modal embedding space. We explore context transformers for video and text modalities. Results consistently demonstrate improved performance on three datasets: YouCook2, EPIC-KITCHENS and a clip-sentence version of ActivityNet Captions. Exhaustive ablation studies and context analysis show the efficacy of the proposed method.

Transformers have achieved the state-of-the-art performance on solving the inverse problem of Snapshot Compressive Imaging (SCI) for video, whose ill-posedness is rooted in the mixed degradation of spatial masking and temporal aliasing. However, previous Transformers lack an insight into the degradation and thus have limited performance and efficiency. In this work, we tailor an efficient reconstruction architecture without temporal aggregation in early layers and Hierarchical Separable Video Transformer (HiSViT) as building block. HiSViT is built by multiple groups of Cross-Scale Separable Multi-head Self-Attention (CSS-MSA) and Gated Self-Modulated Feed-Forward Network (GSM-FFN) with dense connections, each of which is conducted within a separate channel portions at a different scale, for multi-scale interactions and long-range modeling. By separating spatial operations from temporal ones, CSS-MSA introduces an inductive bias of paying more attention within frames instead of between frames while saving computational overheads. GSM-FFN further enhances the locality via gated mechanism and factorized spatial-temporal convolutions. Extensive experiments demonstrate that our method outperforms previous methods by !>!0.5 dB with comparable or fewer parameters and complexity. The source codes and pretrained models are released at https://github.com/pwangcs/HiSViT.

The central building block of convolutional neural networks (CNNs) is the convolution operator, which enables networks to construct informative features by fusing both spatial and channel-wise information within local receptive fields at each layer. A broad range of prior research has investigated the spatial component of this relationship, seeking to strengthen the representational power of a CNN by enhancing the quality of spatial encodings throughout its feature hierarchy. In this work, we focus instead on the channel relationship and propose a novel architectural unit, which we term the "Squeeze-and-Excitation" (SE) block, that adaptively recalibrates channel-wise feature responses by explicitly modelling interdependencies between channels. We show that these blocks can be stacked together to form SENet architectures that generalise extremely effectively across different datasets. We further demonstrate that SE blocks bring significant improvements in performance for existing state-of-the-art CNNs at slight additional computational cost. Squeeze-and-Excitation Networks formed the foundation of our ILSVRC 2017 classification submission which won first place and reduced the top-5 error to 2.251%, surpassing the winning entry of 2016 by a relative improvement of ~25%. Models and code are available at https://github.com/hujie-frank/SENet.

The success of deep learning models has led to their adaptation and adoption by prominent video understanding methods. The majority of these approaches encode features in a joint space-time modality for which the inner workings and learned representations are difficult to visually interpret. We propose LEArned Preconscious Synthesis (LEAPS), an architecture-independent method for synthesizing videos from the internal spatiotemporal representations of models. Using a stimulus video and a target class, we prime a fixed space-time model and iteratively optimize a video initialized with random noise. Additional regularizers are used to improve the feature diversity of the synthesized videos alongside the cross-frame temporal coherence of motions. We quantitatively and qualitatively evaluate the applicability of LEAPS by inverting a range of spatiotemporal convolutional and attention-based architectures trained on Kinetics-400, which to the best of our knowledge has not been previously accomplished.

Neural fields, a category of neural networks trained to represent high-frequency signals, have gained significant attention in recent years due to their impressive performance in modeling complex 3D data, especially large neural signed distance (SDFs) or radiance fields (NeRFs) via a single multi-layer perceptron (MLP). However, despite the power and simplicity of representing signals with an MLP, these methods still face challenges when modeling large and complex temporal signals due to the limited capacity of MLPs. In this paper, we propose an effective approach to address this limitation by incorporating temporal residual layers into neural fields, dubbed ResFields, a novel class of networks specifically designed to effectively represent complex temporal signals. We conduct a comprehensive analysis of the properties of ResFields and propose a matrix factorization technique to reduce the number of trainable parameters and enhance generalization capabilities. Importantly, our formulation seamlessly integrates with existing techniques and consistently improves results across various challenging tasks: 2D video approximation, dynamic shape modeling via temporal SDFs, and dynamic NeRF reconstruction. Lastly, we demonstrate the practical utility of ResFields by showcasing its effectiveness in capturing dynamic 3D scenes from sparse sensory inputs of a lightweight capture system.

We present REM, a framework for segmenting a wide range of concepts in video that can be described through natural language. Our method capitalizes on visual-language representations learned by video diffusion models on Internet-scale datasets. A key insight of our approach is preserving as much of the generative model's original representation as possible, while fine-tuning it on narrow-domain Referral Object Segmentation datasets. As a result, our framework can accurately segment and track rare and unseen objects, despite being trained on object masks from a limited set of categories. Additionally, it can generalize to non-object dynamic concepts, such as waves crashing in the ocean, as demonstrated in our newly introduced benchmark for Referral Video Process Segmentation (Ref-VPS). Our experiments show that REM performs on par with state-of-the-art approaches on in-domain datasets, like Ref-DAVIS, while outperforming them by up to twelve points in terms of region similarity on out-of-domain data, leveraging the power of Internet-scale pre-training.

Integrating CNNs and RNNs to capture spatiotemporal dependencies is a prevalent strategy for spatiotemporal prediction tasks. However, the property of CNNs to learn local spatial information decreases their efficiency in capturing spatiotemporal dependencies, thereby limiting their prediction accuracy. In this paper, we propose a new recurrent cell, SwinLSTM, which integrates Swin Transformer blocks and the simplified LSTM, an extension that replaces the convolutional structure in ConvLSTM with the self-attention mechanism. Furthermore, we construct a network with SwinLSTM cell as the core for spatiotemporal prediction. Without using unique tricks, SwinLSTM outperforms state-of-the-art methods on Moving MNIST, Human3.6m, TaxiBJ, and KTH datasets. In particular, it exhibits a significant improvement in prediction accuracy compared to ConvLSTM. Our competitive experimental results demonstrate that learning global spatial dependencies is more advantageous for models to capture spatiotemporal dependencies. We hope that SwinLSTM can serve as a solid baseline to promote the advancement of spatiotemporal prediction accuracy. The codes are publicly available at https://github.com/SongTang-x/SwinLSTM.

Object detection in point clouds is an important aspect of many robotics applications such as autonomous driving. In this paper we consider the problem of encoding a point cloud into a format appropriate for a downstream detection pipeline. Recent literature suggests two types of encoders; fixed encoders tend to be fast but sacrifice accuracy, while encoders that are learned from data are more accurate, but slower. In this work we propose PointPillars, a novel encoder which utilizes PointNets to learn a representation of point clouds organized in vertical columns (pillars). While the encoded features can be used with any standard 2D convolutional detection architecture, we further propose a lean downstream network. Extensive experimentation shows that PointPillars outperforms previous encoders with respect to both speed and accuracy by a large margin. Despite only using lidar, our full detection pipeline significantly outperforms the state of the art, even among fusion methods, with respect to both the 3D and bird's eye view KITTI benchmarks. This detection performance is achieved while running at 62 Hz: a 2 - 4 fold runtime improvement. A faster version of our method matches the state of the art at 105 Hz. These benchmarks suggest that PointPillars is an appropriate encoding for object detection in point clouds.

Traditional convolutional neural networks have a limited receptive field while transformer-based networks are mediocre in constructing long-term dependency from the perspective of computational complexity. Such the bottleneck poses a significant challenge when processing long video sequences in video analysis tasks. Very recently, the state space models (SSMs) with efficient hardware-aware designs, famous by Mamba, have exhibited impressive achievements in long sequence modeling, which facilitates the development of deep neural networks on many vision tasks. To better capture available cues in video frames, this paper presents a generic Video Vision Mamba-based framework for medical video object segmentation tasks, named Vivim. Our Vivim can effectively compress the long-term spatiotemporal representation into sequences at varying scales by our designed Temporal Mamba Block. Compared to existing video-level Transformer-based methods, our model maintains excellent segmentation results with better speed performance. Extensive experiments on the breast US dataset demonstrate the effectiveness and efficiency of our Vivim. The code for Vivim is available at: https://github.com/scott-yjyang/Vivim.

Recent progress in vision Transformers exhibits great success in various tasks driven by the new spatial modeling mechanism based on dot-product self-attention. In this paper, we show that the key ingredients behind the vision Transformers, namely input-adaptive, long-range and high-order spatial interactions, can also be efficiently implemented with a convolution-based framework. We present the Recursive Gated Convolution (g^nConv) that performs high-order spatial interactions with gated convolutions and recursive designs. The new operation is highly flexible and customizable, which is compatible with various variants of convolution and extends the two-order interactions in self-attention to arbitrary orders without introducing significant extra computation. g^nConv can serve as a plug-and-play module to improve various vision Transformers and convolution-based models. Based on the operation, we construct a new family of generic vision backbones named HorNet. Extensive experiments on ImageNet classification, COCO object detection and ADE20K semantic segmentation show HorNet outperform Swin Transformers and ConvNeXt by a significant margin with similar overall architecture and training configurations. HorNet also shows favorable scalability to more training data and larger model sizes. Apart from the effectiveness in visual encoders, we also show g^nConv can be applied to task-specific decoders and consistently improve dense prediction performance with less computation. Our results demonstrate that g^nConv can be a new basic module for visual modeling that effectively combines the merits of both vision Transformers and CNNs. Code is available at https://github.com/raoyongming/HorNet

In this technical report, we introduce our solution to human-centric spatio-temporal video grounding task. We propose a concise and effective framework named STVGFormer, which models spatiotemporal visual-linguistic dependencies with a static branch and a dynamic branch. The static branch performs cross-modal understanding in a single frame and learns to localize the target object spatially according to intra-frame visual cues like object appearances. The dynamic branch performs cross-modal understanding across multiple frames. It learns to predict the starting and ending time of the target moment according to dynamic visual cues like motions. Both the static and dynamic branches are designed as cross-modal transformers. We further design a novel static-dynamic interaction block to enable the static and dynamic branches to transfer useful and complementary information from each other, which is shown to be effective to improve the prediction on hard cases. Our proposed method achieved 39.6% vIoU and won the first place in the HC-STVG track of the 4th Person in Context Challenge.

We present Omni-RGPT, a multimodal large language model designed to facilitate region-level comprehension for both images and videos. To achieve consistent region representation across spatio-temporal dimensions, we introduce Token Mark, a set of tokens highlighting the target regions within the visual feature space. These tokens are directly embedded into spatial regions using region prompts (e.g., boxes or masks) and simultaneously incorporated into the text prompt to specify the target, establishing a direct connection between visual and text tokens. To further support robust video understanding without requiring tracklets, we introduce an auxiliary task that guides Token Mark by leveraging the consistency of the tokens, enabling stable region interpretation across the video. Additionally, we introduce a large-scale region-level video instruction dataset (RegVID-300k). Omni-RGPT achieves state-of-the-art results on image and video-based commonsense reasoning benchmarks while showing strong performance in captioning and referring expression comprehension tasks.

We study the problem of generating temporal object intrinsics -- temporally evolving sequences of object geometry, reflectance, and texture, such as a blooming rose -- from pre-trained 2D foundation models. Unlike conventional 3D modeling and animation techniques that require extensive manual effort and expertise, we introduce a method that generates such assets with signals distilled from pre-trained 2D diffusion models. To ensure the temporal consistency of object intrinsics, we propose Neural Templates for temporal-state-guided distillation, derived automatically from image features from self-supervised learning. Our method can generate high-quality temporal object intrinsics for several natural phenomena and enable the sampling and controllable rendering of these dynamic objects from any viewpoint, under any environmental lighting conditions, at any time of their lifespan. Project website: https://chen-geng.com/rose4d

This work aims to learn a high-quality text-to-video (T2V) generative model by leveraging a pre-trained text-to-image (T2I) model as a basis. It is a highly desirable yet challenging task to simultaneously a) accomplish the synthesis of visually realistic and temporally coherent videos while b) preserving the strong creative generation nature of the pre-trained T2I model. To this end, we propose LaVie, an integrated video generation framework that operates on cascaded video latent diffusion models, comprising a base T2V model, a temporal interpolation model, and a video super-resolution model. Our key insights are two-fold: 1) We reveal that the incorporation of simple temporal self-attentions, coupled with rotary positional encoding, adequately captures the temporal correlations inherent in video data. 2) Additionally, we validate that the process of joint image-video fine-tuning plays a pivotal role in producing high-quality and creative outcomes. To enhance the performance of LaVie, we contribute a comprehensive and diverse video dataset named Vimeo25M, consisting of 25 million text-video pairs that prioritize quality, diversity, and aesthetic appeal. Extensive experiments demonstrate that LaVie achieves state-of-the-art performance both quantitatively and qualitatively. Furthermore, we showcase the versatility of pre-trained LaVie models in various long video generation and personalized video synthesis applications.

Video compression has always been a popular research area, where many traditional and deep video compression methods have been proposed. These methods typically rely on signal prediction theory to enhance compression performance by designing high efficient intra and inter prediction strategies and compressing video frames one by one. In this paper, we propose a novel model-based video compression (MVC) framework that regards scenes as the fundamental units for video sequences. Our proposed MVC directly models the intensity variation of the entire video sequence in one scene, seeking non-redundant representations instead of reducing redundancy through spatio-temporal predictions. To achieve this, we employ implicit neural representation as our basic modeling architecture. To improve the efficiency of video modeling, we first propose context-related spatial positional embedding and frequency domain supervision in spatial context enhancement. For temporal correlation capturing, we design the scene flow constrain mechanism and temporal contrastive loss. Extensive experimental results demonstrate that our method achieves up to a 20\% bitrate reduction compared to the latest video coding standard H.266 and is more efficient in decoding than existing video coding strategies.

We present a method for estimating neural scenes representations of objects given only a single image. The core of our method is the estimation of a geometric scaffold for the object and its use as a guide for the reconstruction of the underlying radiance field. Our formulation is based on a generative process that first maps a latent code to a voxelized shape, and then renders it to an image, with the object appearance being controlled by a second latent code. During inference, we optimize both the latent codes and the networks to fit a test image of a new object. The explicit disentanglement of shape and appearance allows our model to be fine-tuned given a single image. We can then render new views in a geometrically consistent manner and they represent faithfully the input object. Additionally, our method is able to generalize to images outside of the training domain (more realistic renderings and even real photographs). Finally, the inferred geometric scaffold is itself an accurate estimate of the object's 3D shape. We demonstrate in several experiments the effectiveness of our approach in both synthetic and real images.

Category-specific models are provenly valuable methods in 3D single object tracking (SOT) regardless of Siamese or motion-centric paradigms. However, such over-specialized model designs incur redundant parameters, thus limiting the broader applicability of 3D SOT task. This paper first introduces unified models that can simultaneously track objects across all categories using a single network with shared model parameters. Specifically, we propose to explicitly encode distinct attributes associated to different object categories, enabling the model to adapt to cross-category data. We find that the attribute variances of point cloud objects primarily occur from the varying size and shape (e.g., large and square vehicles v.s. small and slender humans). Based on this observation, we design a novel point set representation learning network inheriting transformer architecture, termed AdaFormer, which adaptively encodes the dynamically varying shape and size information from cross-category data in a unified manner. We further incorporate the size and shape prior derived from the known template targets into the model's inputs and learning objective, facilitating the learning of unified representation. Equipped with such designs, we construct two category-unified models SiamCUT and MoCUT.Extensive experiments demonstrate that SiamCUT and MoCUT exhibit strong generalization and training stability. Furthermore, our category-unified models outperform the category-specific counterparts by a significant margin (e.g., on KITTI dataset, 12% and 3% performance gains on the Siamese and motion paradigms). Our code will be available.

This paper addresses the task of space-time video super-resolution (ST-VSR). Existing methods generally suffer from inaccurate motion estimation and motion compensation (MEMC) problems for large motions. Inspired by recent progress in physics-informed neural networks, we model the challenges of MEMC in ST-VSR as a mapping between two continuous function spaces. Specifically, our approach transforms independent low-resolution representations in the coarse-grained continuous function space into refined representations with enriched spatiotemporal details in the fine-grained continuous function space. To achieve efficient and accurate MEMC, we design a Galerkin-type attention function to perform frame alignment and temporal interpolation. Due to the linear complexity of the Galerkin-type attention mechanism, our model avoids patch partitioning and offers global receptive fields, enabling precise estimation of large motions. The experimental results show that the proposed method surpasses state-of-the-art techniques in both fixed-size and continuous space-time video super-resolution tasks.

Understanding the motion states of the surrounding environment is critical for safe autonomous driving. These motion states can be accurately derived from scene flow, which captures the three-dimensional motion field of points. Existing LiDAR scene flow methods extract spatial features from each point cloud and then fuse them channel-wise, resulting in the implicit extraction of spatio-temporal features. Furthermore, they utilize 2D Bird's Eye View and process only two frames, missing crucial spatial information along the Z-axis and the broader temporal context, leading to suboptimal performance. To address these limitations, we propose Flow4D, which temporally fuses multiple point clouds after the 3D intra-voxel feature encoder, enabling more explicit extraction of spatio-temporal features through a 4D voxel network. However, while using 4D convolution improves performance, it significantly increases the computational load. For further efficiency, we introduce the Spatio-Temporal Decomposition Block (STDB), which combines 3D and 1D convolutions instead of using heavy 4D convolution. In addition, Flow4D further improves performance by using five frames to take advantage of richer temporal information. As a result, the proposed method achieves a 45.9% higher performance compared to the state-of-the-art while running in real-time, and won 1st place in the 2024 Argoverse 2 Scene Flow Challenge. The code is available at https://github.com/dgist-cvlab/Flow4D.

To generate accurate videos, algorithms have to understand the spatial and temporal dependencies in the world. Current algorithms enable accurate predictions over short horizons but tend to suffer from temporal inconsistencies. When generated content goes out of view and is later revisited, the model invents different content instead. Despite this severe limitation, no established benchmarks on complex data exist for rigorously evaluating video generation with long temporal dependencies. In this paper, we curate 3 challenging video datasets with long-range dependencies by rendering walks through 3D scenes of procedural mazes, Minecraft worlds, and indoor scans. We perform a comprehensive evaluation of current models and observe their limitations in temporal consistency. Moreover, we introduce the Temporally Consistent Transformer (TECO), a generative model that substantially improves long-term consistency while also reducing sampling time. By compressing its input sequence into fewer embeddings, applying a temporal transformer, and expanding back using a spatial MaskGit, TECO outperforms existing models across many metrics. Videos are available on the website: https://wilson1yan.github.io/teco

Video-language pre-training is crucial for learning powerful multi-modal representation. However, it typically requires a massive amount of computation. In this paper, we develop SMAUG, an efficient pre-training framework for video-language models. The foundation component in SMAUG is masked autoencoders. Different from prior works which only mask textual inputs, our masking strategy considers both visual and textual modalities, providing a better cross-modal alignment and saving more pre-training costs. On top of that, we introduce a space-time token sparsification module, which leverages context information to further select only "important" spatial regions and temporal frames for pre-training. Coupling all these designs allows our method to enjoy both competitive performances on text-to-video retrieval and video question answering tasks, and much less pre-training costs by 1.9X or more. For example, our SMAUG only needs about 50 NVIDIA A6000 GPU hours for pre-training to attain competitive performances on these two video-language tasks across six popular benchmarks.

Existing neural field representations for 3D object reconstruction either (1) utilize object-level representations, but suffer from low-quality details due to conditioning on a global latent code, or (2) are able to perfectly reconstruct the observations, but fail to utilize object-level prior knowledge to infer unobserved regions. We present SimNP, a method to learn category-level self-similarities, which combines the advantages of both worlds by connecting neural point radiance fields with a category-level self-similarity representation. Our contribution is two-fold. (1) We design the first neural point representation on a category level by utilizing the concept of coherent point clouds. The resulting neural point radiance fields store a high level of detail for locally supported object regions. (2) We learn how information is shared between neural points in an unconstrained and unsupervised fashion, which allows to derive unobserved regions of an object during the reconstruction process from given observations. We show that SimNP is able to outperform previous methods in reconstructing symmetric unseen object regions, surpassing methods that build upon category-level or pixel-aligned radiance fields, while providing semantic correspondences between instances

It is a challenging task to learn rich and multi-scale spatiotemporal semantics from high-dimensional videos, due to large local redundancy and complex global dependency between video frames. The recent advances in this research have been mainly driven by 3D convolutional neural networks and vision transformers. Although 3D convolution can efficiently aggregate local context to suppress local redundancy from a small 3D neighborhood, it lacks the capability to capture global dependency because of the limited receptive field. Alternatively, vision transformers can effectively capture long-range dependency by self-attention mechanism, while having the limitation on reducing local redundancy with blind similarity comparison among all the tokens in each layer. Based on these observations, we propose a novel Unified transFormer (UniFormer) which seamlessly integrates merits of 3D convolution and spatiotemporal self-attention in a concise transformer format, and achieves a preferable balance between computation and accuracy. Different from traditional transformers, our relation aggregator can tackle both spatiotemporal redundancy and dependency, by learning local and global token affinity respectively in shallow and deep layers. We conduct extensive experiments on the popular video benchmarks, e.g., Kinetics-400, Kinetics-600, and Something-Something V1&V2. With only ImageNet-1K pretraining, our UniFormer achieves 82.9%/84.8% top-1 accuracy on Kinetics-400/Kinetics-600, while requiring 10x fewer GFLOPs than other state-of-the-art methods. For Something-Something V1 and V2, our UniFormer achieves new state-of-the-art performances of 60.9% and 71.2% top-1 accuracy respectively. Code is available at https://github.com/Sense-X/UniFormer.

In a video, an object usually appears as the trajectory, i.e., it spans over a few spatial but longer temporal patches, that contains abundant spatiotemporal contexts. However, modern Video-Language BERTs (VDL-BERTs) neglect this trajectory characteristic that they usually follow image-language BERTs (IL-BERTs) to deploy the patch-to-word (P2W) attention that may over-exploit trivial spatial contexts and neglect significant temporal contexts. To amend this, we propose a novel TW-BERT to learn Trajectory-Word alignment by a newly designed trajectory-to-word (T2W) attention for solving video-language tasks. Moreover, previous VDL-BERTs usually uniformly sample a few frames into the model while different trajectories have diverse graininess, i.e., some trajectories span longer frames and some span shorter, and using a few frames will lose certain useful temporal contexts. However, simply sampling more frames will also make pre-training infeasible due to the largely increased training burdens. To alleviate the problem, during the fine-tuning stage, we insert a novel Hierarchical Frame-Selector (HFS) module into the video encoder. HFS gradually selects the suitable frames conditioned on the text context for the later cross-modal encoder to learn better trajectory-word alignments. By the proposed T2W attention and HFS, our TW-BERT achieves SOTA performances on text-to-video retrieval tasks, and comparable performances on video question-answering tasks with some VDL-BERTs trained on much more data. The code will be available in the supplementary material.

The ability to integrate context, including perceptual and temporal cues, plays a pivotal role in grounding the meaning of a linguistic utterance. In order to measure to what extent current vision-and-language models master this ability, we devise a new multimodal challenge, Image Retrieval from Contextual Descriptions (ImageCoDe). In particular, models are tasked with retrieving the correct image from a set of 10 minimally contrastive candidates based on a contextual description. As such, each description contains only the details that help distinguish between images. Because of this, descriptions tend to be complex in terms of syntax and discourse and require drawing pragmatic inferences. Images are sourced from both static pictures and video frames. We benchmark several state-of-the-art models, including both cross-encoders such as ViLBERT and bi-encoders such as CLIP, on ImageCoDe. Our results reveal that these models dramatically lag behind human performance: the best variant achieves an accuracy of 20.9 on video frames and 59.4 on static pictures, compared with 90.8 in humans. Furthermore, we experiment with new model variants that are better equipped to incorporate visual and temporal context into their representations, which achieve modest gains. Our hope is that ImageCoDE will foster progress in grounded language understanding by encouraging models to focus on fine-grained visual differences.

Frame quality deterioration is one of the main challenges in the field of video understanding. To compensate for the information loss caused by deteriorated frames, recent approaches exploit transformer-based integration modules to obtain spatio-temporal information. However, these integration modules are heavy and complex. Furthermore, each integration module is specifically tailored for its target task, making it difficult to generalise to multiple tasks. In this paper, we present a neat and unified framework, called Spatio-Temporal Prompting Network (STPN). It can efficiently extract robust and accurate video features by dynamically adjusting the input features in the backbone network. Specifically, STPN predicts several video prompts containing spatio-temporal information of neighbour frames. Then, these video prompts are prepended to the patch embeddings of the current frame as the updated input for video feature extraction. Moreover, STPN is easy to generalise to various video tasks because it does not contain task-specific modules. Without bells and whistles, STPN achieves state-of-the-art performance on three widely-used datasets for different video understanding tasks, i.e., ImageNetVID for video object detection, YouTubeVIS for video instance segmentation, and GOT-10k for visual object tracking. Code is available at https://github.com/guanxiongsun/vfe.pytorch.

This paper studies the problem of concept-based interpretability of transformer representations for videos. Concretely, we seek to explain the decision-making process of video transformers based on high-level, spatiotemporal concepts that are automatically discovered. Prior research on concept-based interpretability has concentrated solely on image-level tasks. Comparatively, video models deal with the added temporal dimension, increasing complexity and posing challenges in identifying dynamic concepts over time. In this work, we systematically address these challenges by introducing the first Video Transformer Concept Discovery (VTCD) algorithm. To this end, we propose an efficient approach for unsupervised identification of units of video transformer representations - concepts, and ranking their importance to the output of a model. The resulting concepts are highly interpretable, revealing spatio-temporal reasoning mechanisms and object-centric representations in unstructured video models. Performing this analysis jointly over a diverse set of supervised and self-supervised representations, we discover that some of these mechanism are universal in video transformers. Finally, we demonstrate that VTCDcan be used to improve model performance for fine-grained tasks.

Existing object detection frameworks are usually built on a single format of object/part representation, i.e., anchor/proposal rectangle boxes in RetinaNet and Faster R-CNN, center points in FCOS and RepPoints, and corner points in CornerNet. While these different representations usually drive the frameworks to perform well in different aspects, e.g., better classification or finer localization, it is in general difficult to combine these representations in a single framework to make good use of each strength, due to the heterogeneous or non-grid feature extraction by different representations. This paper presents an attention-based decoder module similar as that in Transformer~vaswani2017attention to bridge other representations into a typical object detector built on a single representation format, in an end-to-end fashion. The other representations act as a set of key instances to strengthen the main query representation features in the vanilla detectors. Novel techniques are proposed towards efficient computation of the decoder module, including a key sampling approach and a shared location embedding approach. The proposed module is named bridging visual representations (BVR). It can perform in-place and we demonstrate its broad effectiveness in bridging other representations into prevalent object detection frameworks, including RetinaNet, Faster R-CNN, FCOS and ATSS, where about 1.5sim3.0 AP improvements are achieved. In particular, we improve a state-of-the-art framework with a strong backbone by about 2.0 AP, reaching 52.7 AP on COCO test-dev. The resulting network is named RelationNet++. The code will be available at https://github.com/microsoft/RelationNet2.

As transformers are equivariant to the permutation of input tokens, encoding the positional information of tokens is necessary for many tasks. However, since existing positional encoding schemes have been initially designed for NLP tasks, their suitability for vision tasks, which typically exhibit different structural properties in their data, is questionable. We argue that existing positional encoding schemes are suboptimal for 3D vision tasks, as they do not respect their underlying 3D geometric structure. Based on this hypothesis, we propose a geometry-aware attention mechanism that encodes the geometric structure of tokens as relative transformation determined by the geometric relationship between queries and key-value pairs. By evaluating on multiple novel view synthesis (NVS) datasets in the sparse wide-baseline multi-view setting, we show that our attention, called Geometric Transform Attention (GTA), improves learning efficiency and performance of state-of-the-art transformer-based NVS models without any additional learned parameters and only minor computational overhead.

We present a simple yet powerful neural network that implicitly represents and renders 3D objects and scenes only from 2D observations. The network models 3D geometries as a general radiance field, which takes a set of 2D images with camera poses and intrinsics as input, constructs an internal representation for each point of the 3D space, and then renders the corresponding appearance and geometry of that point viewed from an arbitrary position. The key to our approach is to learn local features for each pixel in 2D images and to then project these features to 3D points, thus yielding general and rich point representations. We additionally integrate an attention mechanism to aggregate pixel features from multiple 2D views, such that visual occlusions are implicitly taken into account. Extensive experiments demonstrate that our method can generate high-quality and realistic novel views for novel objects, unseen categories and challenging real-world scenes.

For a vision-language model (VLM) to understand the physical world, such as cause and effect, a first step is to capture the temporal dynamics of the visual world, for example how the physical states of objects evolve over time (e.g. a whole apple into a sliced apple). Our paper aims to investigate if VLMs pre-trained on web-scale data learn to encode object states, which can be extracted with zero-shot text prompts. We curate an object state recognition dataset ChangeIt-Frames, and evaluate nine open-source VLMs, including models trained with contrastive and generative objectives. We observe that while these state-of-the-art vision-language models can reliably perform object recognition, they consistently fail to accurately distinguish the objects' physical states. Through extensive experiments, we identify three areas for improvements for VLMs to better encode object states, namely the quality of object localization, the architecture to bind concepts to objects, and the objective to learn discriminative visual and language encoders on object states. Data and code are released.

Spatial relationships between objects represent key scene information for humans to understand and interact with the world. To study the capability of current computer vision systems to recognize physically grounded spatial relations, we start by proposing precise relation definitions that permit consistently annotating a benchmark dataset. Despite the apparent simplicity of this task relative to others in the recognition literature, we observe that existing approaches perform poorly on this benchmark. We propose new approaches exploiting the long-range attention capabilities of transformers for this task, and evaluating key design principles. We identify a simple "RelatiViT" architecture and demonstrate that it outperforms all current approaches. To our knowledge, this is the first method to convincingly outperform naive baselines on spatial relation prediction in in-the-wild settings. The code and datasets are available in https://sites.google.com/view/spatial-relation.

With the success of new computational architectures for visual processing, such as convolutional neural networks (CNN) and access to image databases with millions of labeled examples (e.g., ImageNet, Places), the state of the art in computer vision is advancing rapidly. One important factor for continued progress is to understand the representations that are learned by the inner layers of these deep architectures. Here we show that object detectors emerge from training CNNs to perform scene classification. As scenes are composed of objects, the CNN for scene classification automatically discovers meaningful objects detectors, representative of the learned scene categories. With object detectors emerging as a result of learning to recognize scenes, our work demonstrates that the same network can perform both scene recognition and object localization in a single forward-pass, without ever having been explicitly taught the notion of objects.

We propose a unified point cloud video self-supervised learning framework for object-centric and scene-centric data. Previous methods commonly conduct representation learning at the clip or frame level and cannot well capture fine-grained semantics. Instead of contrasting the representations of clips or frames, in this paper, we propose a unified self-supervised framework by conducting contrastive learning at the point level. Moreover, we introduce a new pretext task by achieving semantic alignment of superpoints, which further facilitates the representations to capture semantic cues at multiple scales. In addition, due to the high redundancy in the temporal dimension of dynamic point clouds, directly conducting contrastive learning at the point level usually leads to massive undesired negatives and insufficient modeling of positive representations. To remedy this, we propose a selection strategy to retain proper negatives and make use of high-similarity samples from other instances as positive supplements. Extensive experiments show that our method outperforms supervised counterparts on a wide range of downstream tasks and demonstrates the superior transferability of the learned representations.

Current large multimodal models (LMMs) face significant challenges in processing and comprehending long-duration or high-resolution videos, which is mainly due to the lack of high-quality datasets. To address this issue from a data-centric perspective, we propose VISTA, a simple yet effective Video Spatiotemporal Augmentation framework that synthesizes long-duration and high-resolution video instruction-following pairs from existing video-caption datasets. VISTA spatially and temporally combines videos to create new synthetic videos with extended durations and enhanced resolutions, and subsequently produces question-answer pairs pertaining to these newly synthesized videos. Based on this paradigm, we develop seven video augmentation methods and curate VISTA-400K, a video instruction-following dataset aimed at enhancing long-duration and high-resolution video understanding. Finetuning various video LMMs on our data resulted in an average improvement of 3.3% across four challenging benchmarks for long-video understanding. Furthermore, we introduce the first comprehensive high-resolution video understanding benchmark HRVideoBench, on which our finetuned models achieve a 6.5% performance gain. These results highlight the effectiveness of our framework.

Learning 4D language fields to enable time-sensitive, open-ended language queries in dynamic scenes is essential for many real-world applications. While LangSplat successfully grounds CLIP features into 3D Gaussian representations, achieving precision and efficiency in 3D static scenes, it lacks the ability to handle dynamic 4D fields as CLIP, designed for static image-text tasks, cannot capture temporal dynamics in videos. Real-world environments are inherently dynamic, with object semantics evolving over time. Building a precise 4D language field necessitates obtaining pixel-aligned, object-wise video features, which current vision models struggle to achieve. To address these challenges, we propose 4D LangSplat, which learns 4D language fields to handle time-agnostic or time-sensitive open-vocabulary queries in dynamic scenes efficiently. 4D LangSplat bypasses learning the language field from vision features and instead learns directly from text generated from object-wise video captions via Multimodal Large Language Models (MLLMs). Specifically, we propose a multimodal object-wise video prompting method, consisting of visual and text prompts that guide MLLMs to generate detailed, temporally consistent, high-quality captions for objects throughout a video. These captions are encoded using a Large Language Model into high-quality sentence embeddings, which then serve as pixel-aligned, object-specific feature supervision, facilitating open-vocabulary text queries through shared embedding spaces. Recognizing that objects in 4D scenes exhibit smooth transitions across states, we further propose a status deformable network to model these continuous changes over time effectively. Our results across multiple benchmarks demonstrate that 4D LangSplat attains precise and efficient results for both time-sensitive and time-agnostic open-vocabulary queries.

The recently-developed DETR approach applies the transformer encoder and decoder architecture to object detection and achieves promising performance. In this paper, we handle the critical issue, slow training convergence, and present a conditional cross-attention mechanism for fast DETR training. Our approach is motivated by that the cross-attention in DETR relies highly on the content embeddings for localizing the four extremities and predicting the box, which increases the need for high-quality content embeddings and thus the training difficulty. Our approach, named conditional DETR, learns a conditional spatial query from the decoder embedding for decoder multi-head cross-attention. The benefit is that through the conditional spatial query, each cross-attention head is able to attend to a band containing a distinct region, e.g., one object extremity or a region inside the object box. This narrows down the spatial range for localizing the distinct regions for object classification and box regression, thus relaxing the dependence on the content embeddings and easing the training. Empirical results show that conditional DETR converges 6.7x faster for the backbones R50 and R101 and 10x faster for stronger backbones DC5-R50 and DC5-R101. Code is available at https://github.com/Atten4Vis/ConditionalDETR.

Denoising diffusion probabilistic models that were initially proposed for realistic image generation have recently shown success in various perception tasks (e.g., object detection and image segmentation) and are increasingly gaining attention in computer vision. However, extending such models to multi-frame human pose estimation is non-trivial due to the presence of the additional temporal dimension in videos. More importantly, learning representations that focus on keypoint regions is crucial for accurate localization of human joints. Nevertheless, the adaptation of the diffusion-based methods remains unclear on how to achieve such objective. In this paper, we present DiffPose, a novel diffusion architecture that formulates video-based human pose estimation as a conditional heatmap generation problem. First, to better leverage temporal information, we propose SpatioTemporal Representation Learner which aggregates visual evidences across frames and uses the resulting features in each denoising step as a condition. In addition, we present a mechanism called Lookup-based MultiScale Feature Interaction that determines the correlations between local joints and global contexts across multiple scales. This mechanism generates delicate representations that focus on keypoint regions. Altogether, by extending diffusion models, we show two unique characteristics from DiffPose on pose estimation task: (i) the ability to combine multiple sets of pose estimates to improve prediction accuracy, particularly for challenging joints, and (ii) the ability to adjust the number of iterative steps for feature refinement without retraining the model. DiffPose sets new state-of-the-art results on three benchmarks: PoseTrack2017, PoseTrack2018, and PoseTrack21.

Traffic prediction is a typical spatio-temporal data mining task and has great significance to the public transportation system. Considering the demand for its grand application, we recognize key factors for an ideal spatio-temporal prediction method: efficient, lightweight, and effective. However, the current deep model-based spatio-temporal prediction solutions generally own intricate architectures with cumbersome optimization, which can hardly meet these expectations. To accomplish the above goals, we propose an intuitive and novel framework, MLPST, a pure multi-layer perceptron architecture for traffic prediction. Specifically, we first capture spatial relationships from both local and global receptive fields. Then, temporal dependencies in different intervals are comprehensively considered. Through compact and swift MLP processing, MLPST can well capture the spatial and temporal dependencies while requiring only linear computational complexity, as well as model parameters that are more than an order of magnitude lower than baselines. Extensive experiments validated the superior effectiveness and efficiency of MLPST against advanced baselines, and among models with optimal accuracy, MLPST achieves the best time and space efficiency.

Large-scale image-language pretrained models, e.g., CLIP, have demonstrated remarkable proficiency in acquiring general multi-modal knowledge through web-scale image-text data. Despite the impressive performance of image-language models on various image tasks, how to effectively expand them on general video understanding remains an area of ongoing exploration. In this paper, we investigate the image-to-video transferring from the perspective of the model and the data, unveiling two key obstacles impeding the adaptation of image-language models: non-generalizable temporal modeling and partially misaligned video-text data. To address these challenges, we propose Spatial-Temporal Auxiliary Network with Mutual-guided alignment module (Mug-STAN), a simple yet effective framework extending image-text model to diverse video tasks and video-text data.Specifically, STAN adopts a branch structure with decomposed spatial-temporal modules to enable generalizable temporal modeling, while Mug suppresses misalignment by introducing token-wise feature aggregation of either modality from the other. Extensive experimental results verify Mug-STAN significantly improves adaptation of language-image pretrained models such as CLIP and CoCa at both video-text post-pretraining and finetuning stages. With our solution, state-of-the-art zero-shot and finetuning results on various downstream datasets, including MSR-VTT, DiDeMo, LSMDC, Kinetics-400, Something-Something-2, HMDB-51, UCF- 101, and AVA, are achieved. Moreover, by integrating pretrained Mug-STAN with the emerging multimodal dialogue model, we can realize zero-shot video chatting. Codes are available at https://github.com/farewellthree/STAN

Understanding complex scenes at multiple levels of abstraction remains a formidable challenge in computer vision. To address this, we introduce Nested Neural Feature Fields (N2F2), a novel approach that employs hierarchical supervision to learn a single feature field, wherein different dimensions within the same high-dimensional feature encode scene properties at varying granularities. Our method allows for a flexible definition of hierarchies, tailored to either the physical dimensions or semantics or both, thereby enabling a comprehensive and nuanced understanding of scenes. We leverage a 2D class-agnostic segmentation model to provide semantically meaningful pixel groupings at arbitrary scales in the image space, and query the CLIP vision-encoder to obtain language-aligned embeddings for each of these segments. Our proposed hierarchical supervision method then assigns different nested dimensions of the feature field to distill the CLIP embeddings using deferred volumetric rendering at varying physical scales, creating a coarse-to-fine representation. Extensive experiments show that our approach outperforms the state-of-the-art feature field distillation methods on tasks such as open-vocabulary 3D segmentation and localization, demonstrating the effectiveness of the learned nested feature field.

A central objective in computer vision is to design models with appropriate 2-D inductive bias. Desiderata for 2D inductive bias include two-dimensional position awareness, dynamic spatial locality, and translation and permutation invariance. To address these goals, we leverage an expressive variation of the multidimensional State Space Model (SSM). Our approach introduces efficient parameterization, accelerated computation, and a suitable normalization scheme. Empirically, we observe that incorporating our layer at the beginning of each transformer block of Vision Transformers (ViT) significantly enhances performance for multiple ViT backbones and across datasets. The new layer is effective even with a negligible amount of additional parameters and inference time. Ablation studies and visualizations demonstrate that the layer has a strong 2-D inductive bias. For example, vision transformers equipped with our layer exhibit effective performance even without positional encoding

Arbitrary-scale video super-resolution (AVSR) aims to enhance the resolution of video frames, potentially at various scaling factors, which presents several challenges regarding spatial detail reproduction, temporal consistency, and computational complexity. In this paper, we first describe a strong baseline for AVSR by putting together three variants of elementary building blocks: 1) a flow-guided recurrent unit that aggregates spatiotemporal information from previous frames, 2) a flow-refined cross-attention unit that selects spatiotemporal information from future frames, and 3) a hyper-upsampling unit that generates scaleaware and content-independent upsampling kernels. We then introduce ST-AVSR by equipping our baseline with a multi-scale structural and textural prior computed from the pre-trained VGG network. This prior has proven effective in discriminating structure and texture across different locations and scales, which is beneficial for AVSR. Comprehensive experiments show that ST-AVSR significantly improves super-resolution quality, generalization ability, and inference speed over the state-of-theart. The code is available at https://github.com/shangwei5/ST-AVSR.

We present a video generation model that accurately reproduces object motion, changes in camera viewpoint, and new content that arises over time. Existing video generation methods often fail to produce new content as a function of time while maintaining consistencies expected in real environments, such as plausible dynamics and object persistence. A common failure case is for content to never change due to over-reliance on inductive biases to provide temporal consistency, such as a single latent code that dictates content for the entire video. On the other extreme, without long-term consistency, generated videos may morph unrealistically between different scenes. To address these limitations, we prioritize the time axis by redesigning the temporal latent representation and learning long-term consistency from data by training on longer videos. To this end, we leverage a two-phase training strategy, where we separately train using longer videos at a low resolution and shorter videos at a high resolution. To evaluate the capabilities of our model, we introduce two new benchmark datasets with explicit focus on long-term temporal dynamics.

Video amodal segmentation is a particularly challenging task in computer vision, which requires to deduce the full shape of an object from the visible parts of it. Recently, some studies have achieved promising performance by using motion flow to integrate information across frames under a self-supervised setting. However, motion flow has a clear limitation by the two factors of moving cameras and object deformation. This paper presents a rethinking to previous works. We particularly leverage the supervised signals with object-centric representation in real-world scenarios. The underlying idea is the supervision signal of the specific object and the features from different views can mutually benefit the deduction of the full mask in any specific frame. We thus propose an Efficient object-centric Representation amodal Segmentation (EoRaS). Specially, beyond solely relying on supervision signals, we design a translation module to project image features into the Bird's-Eye View (BEV), which introduces 3D information to improve current feature quality. Furthermore, we propose a multi-view fusion layer based temporal module which is equipped with a set of object slots and interacts with features from different views by attention mechanism to fulfill sufficient object representation completion. As a result, the full mask of the object can be decoded from image features updated by object slots. Extensive experiments on both real-world and synthetic benchmarks demonstrate the superiority of our proposed method, achieving state-of-the-art performance. Our code will be released at https://github.com/kfan21/EoRaS.

We present pure-transformer based models for video classification, drawing upon the recent success of such models in image classification. Our model extracts spatio-temporal tokens from the input video, which are then encoded by a series of transformer layers. In order to handle the long sequences of tokens encountered in video, we propose several, efficient variants of our model which factorise the spatial- and temporal-dimensions of the input. Although transformer-based models are known to only be effective when large training datasets are available, we show how we can effectively regularise the model during training and leverage pretrained image models to be able to train on comparatively small datasets. We conduct thorough ablation studies, and achieve state-of-the-art results on multiple video classification benchmarks including Kinetics 400 and 600, Epic Kitchens, Something-Something v2 and Moments in Time, outperforming prior methods based on deep 3D convolutional networks. To facilitate further research, we release code at https://github.com/google-research/scenic/tree/main/scenic/projects/vivit

We introduce an unsupervised feature learning approach that embeds 3D shape information into a single-view image representation. The main idea is a self-supervised training objective that, given only a single 2D image, requires all unseen views of the object to be predictable from learned features. We implement this idea as an encoder-decoder convolutional neural network. The network maps an input image of an unknown category and unknown viewpoint to a latent space, from which a deconvolutional decoder can best "lift" the image to its complete viewgrid showing the object from all viewing angles. Our class-agnostic training procedure encourages the representation to capture fundamental shape primitives and semantic regularities in a data-driven manner---without manual semantic labels. Our results on two widely-used shape datasets show 1) our approach successfully learns to perform "mental rotation" even for objects unseen during training, and 2) the learned latent space is a powerful representation for object recognition, outperforming several existing unsupervised feature learning methods.

Camera traps, unmanned observation devices, and deep learning-based image recognition systems have greatly reduced human effort in collecting and analyzing wildlife images. However, data collected via above apparatus exhibits 1) long-tailed and 2) open-ended distribution problems. To tackle the open-set long-tailed recognition problem, we propose the Temporal Flow Mask Attention Network that comprises three key building blocks: 1) an optical flow module, 2) an attention residual module, and 3) a meta-embedding classifier. We extract temporal features of sequential frames using the optical flow module and learn informative representation using attention residual blocks. Moreover, we show that applying the meta-embedding technique boosts the performance of the method in open-set long-tailed recognition. We apply this method on a Korean Demilitarized Zone (DMZ) dataset. We conduct extensive experiments, and quantitative and qualitative analyses to prove that our method effectively tackles the open-set long-tailed recognition problem while being robust to unknown classes.

Spatio-temporal grounding describes the task of localizing events in space and time, e.g., in video data, based on verbal descriptions only. Models for this task are usually trained with human-annotated sentences and bounding box supervision. This work addresses this task from a multimodal supervision perspective, proposing a framework for spatio-temporal action grounding trained on loose video and subtitle supervision only, without human annotation. To this end, we combine local representation learning, which focuses on leveraging fine-grained spatial information, with a global representation encoding that captures higher-level representations and incorporates both in a joint approach. To evaluate this challenging task in a real-life setting, a new benchmark dataset is proposed providing dense spatio-temporal grounding annotations in long, untrimmed, multi-action instructional videos for over 5K events. We evaluate the proposed approach and other methods on the proposed and standard downstream tasks showing that our method improves over current baselines in various settings, including spatial, temporal, and untrimmed multi-action spatio-temporal grounding.

Object-centric learning (OCL) aspires general and compositional understanding of scenes by representing a scene as a collection of object-centric representations. OCL has also been extended to multi-view image and video datasets to apply various data-driven inductive biases by utilizing geometric or temporal information in the multi-image data. Single-view images carry less information about how to disentangle a given scene than videos or multi-view images do. Hence, owing to the difficulty of applying inductive biases, OCL for single-view images remains challenging, resulting in inconsistent learning of object-centric representation. To this end, we introduce a novel OCL framework for single-view images, SLot Attention via SHepherding (SLASH), which consists of two simple-yet-effective modules on top of Slot Attention. The new modules, Attention Refining Kernel (ARK) and Intermediate Point Predictor and Encoder (IPPE), respectively, prevent slots from being distracted by the background noise and indicate locations for slots to focus on to facilitate learning of object-centric representation. We also propose a weak semi-supervision approach for OCL, whilst our proposed framework can be used without any assistant annotation during the inference. Experiments show that our proposed method enables consistent learning of object-centric representation and achieves strong performance across four datasets. Code is available at https://github.com/object-understanding/SLASH.

Learning structured representations of the visual world in terms of objects promises to significantly improve the generalization abilities of current machine learning models. While recent efforts to this end have shown promising empirical progress, a theoretical account of when unsupervised object-centric representation learning is possible is still lacking. Consequently, understanding the reasons for the success of existing object-centric methods as well as designing new theoretically grounded methods remains challenging. In the present work, we analyze when object-centric representations can provably be learned without supervision. To this end, we first introduce two assumptions on the generative process for scenes comprised of several objects, which we call compositionality and irreducibility. Under this generative process, we prove that the ground-truth object representations can be identified by an invertible and compositional inference model, even in the presence of dependencies between objects. We empirically validate our results through experiments on synthetic data. Finally, we provide evidence that our theory holds predictive power for existing object-centric models by showing a close correspondence between models' compositionality and invertibility and their empirical identifiability.

We investigate the CLIP image encoder by analyzing how individual model components affect the final representation. We decompose the image representation as a sum across individual image patches, model layers, and attention heads, and use CLIP's text representation to interpret the summands. Interpreting the attention heads, we characterize each head's role by automatically finding text representations that span its output space, which reveals property-specific roles for many heads (e.g. location or shape). Next, interpreting the image patches, we uncover an emergent spatial localization within CLIP. Finally, we use this understanding to remove spurious features from CLIP and to create a strong zero-shot image segmenter. Our results indicate that a scalable understanding of transformer models is attainable and can be used to repair and improve models.

We introduce a novel method for pre-training of large-scale vision encoders. Building on recent advancements in autoregressive pre-training of vision models, we extend this framework to a multimodal setting, i.e., images and text. In this paper, we present AIMV2, a family of generalist vision encoders characterized by a straightforward pre-training process, scalability, and remarkable performance across a range of downstream tasks. This is achieved by pairing the vision encoder with a multimodal decoder that autoregressively generates raw image patches and text tokens. Our encoders excel not only in multimodal evaluations but also in vision benchmarks such as localization, grounding, and classification. Notably, our AIMV2-3B encoder achieves 89.5% accuracy on ImageNet-1k with a frozen trunk. Furthermore, AIMV2 consistently outperforms state-of-the-art contrastive models (e.g., CLIP, SigLIP) in multimodal image understanding across diverse settings.

Our world offers a never-ending stream of visual stimuli, yet today's vision systems only accurately recognize patterns within a few seconds. These systems understand the present, but fail to contextualize it in past or future events. In this paper, we study long-form video understanding. We introduce a framework for modeling long-form videos and develop evaluation protocols on large-scale datasets. We show that existing state-of-the-art short-term models are limited for long-form tasks. A novel object-centric transformer-based video recognition architecture performs significantly better on 7 diverse tasks. It also outperforms comparable state-of-the-art on the AVA dataset.

The extraction of modular object-centric representations for downstream tasks is an emerging area of research. Learning grounded representations of objects that are guaranteed to be stable and invariant promises robust performance across different tasks and environments. Slot Attention (SA) learns object-centric representations by assigning objects to slots, but presupposes a single distribution from which all slots are randomly initialised. This results in an inability to learn specialized slots which bind to specific object types and remain invariant to identity-preserving changes in object appearance. To address this, we present \textsc{Conditional Slot Attention} (CoSA) using a novel concept of Grounded Slot Dictionary (GSD) inspired by vector quantization. Our proposed GSD comprises (i) canonical object-level property vectors and (ii) parametric Gaussian distributions, which define a prior over the slots. We demonstrate the benefits of our method in multiple downstream tasks such as scene generation, composition, and task adaptation, whilst remaining competitive with SA in popular object discovery benchmarks.

With the impressive progress in diffusion-based text-to-image generation, extending such powerful generative ability to text-to-video raises enormous attention. Existing methods either require large-scale text-video pairs and a large number of training resources or learn motions that are precisely aligned with template videos. It is non-trivial to balance a trade-off between the degree of generation freedom and the resource costs for video generation. In our study, we present a few-shot-based tuning framework, LAMP, which enables text-to-image diffusion model Learn A specific Motion Pattern with 8~16 videos on a single GPU. Specifically, we design a first-frame-conditioned pipeline that uses an off-the-shelf text-to-image model for content generation so that our tuned video diffusion model mainly focuses on motion learning. The well-developed text-to-image techniques can provide visually pleasing and diverse content as generation conditions, which highly improves video quality and generation freedom. To capture the features of temporal dimension, we expand the pretrained 2D convolution layers of the T2I model to our novel temporal-spatial motion learning layers and modify the attention blocks to the temporal level. Additionally, we develop an effective inference trick, shared-noise sampling, which can improve the stability of videos with computational costs. Our method can also be flexibly applied to other tasks, e.g. real-world image animation and video editing. Extensive experiments demonstrate that LAMP can effectively learn the motion pattern on limited data and generate high-quality videos. The code and models are available at https://rq-wu.github.io/projects/LAMP.

The recent advent of Large Language Models (LLMs) has ushered sophisticated reasoning capabilities into the realm of video through Video Large Language Models (VideoLLMs). However, VideoLLMs currently rely on a single vision encoder for all of their visual processing, which limits the amount and type of visual information that can be conveyed to the LLM. Our method, MERV, Multi-Encoder Representation of Videos, instead leverages multiple frozen visual encoders to create a unified representation of a video, providing the VideoLLM with a comprehensive set of specialized visual knowledge. Spatio-temporally aligning the features from each encoder allows us to tackle a wider range of open-ended and multiple-choice video understanding questions and outperform prior state-of-the-art works. MERV is up to 3.7% better in accuracy than Video-LLaVA across the standard suite video understanding benchmarks, while also having a better Video-ChatGPT score. We also improve upon SeViLA, the previous best on zero-shot Perception Test accuracy, by 2.2%. MERV introduces minimal extra parameters and trains faster than equivalent single-encoder methods while parallelizing the visual processing. Finally, we provide qualitative evidence that MERV successfully captures domain knowledge from each of its encoders. Our results offer promising directions in utilizing multiple vision encoders for comprehensive video understanding.

Image-to-text tasks, such as open-ended image captioning and controllable image description, have received extensive attention for decades. Here, we further advance this line of work by presenting Visual Spatial Description (VSD), a new perspective for image-to-text toward spatial semantics. Given an image and two objects inside it, VSD aims to produce one description focusing on the spatial perspective between the two objects. Accordingly, we manually annotate a dataset to facilitate the investigation of the newly-introduced task and build several benchmark encoder-decoder models by using VL-BART and VL-T5 as backbones. In addition, we investigate pipeline and joint end-to-end architectures for incorporating visual spatial relationship classification (VSRC) information into our model. Finally, we conduct experiments on our benchmark dataset to evaluate all our models. Results show that our models are impressive, providing accurate and human-like spatial-oriented text descriptions. Meanwhile, VSRC has great potential for VSD, and the joint end-to-end architecture is the better choice for their integration. We make the dataset and codes public for research purposes.

The spatio-temporal complexity of video data presents significant challenges in tasks such as compression, generation, and inpainting. We present four key contributions to address the challenges of spatiotemporal video processing. First, we introduce the 3D Mobile Inverted Vector-Quantization Variational Autoencoder (3D-MBQ-VAE), which combines Variational Autoencoders (VAEs) with masked token modeling to enhance spatiotemporal video compression. The model achieves superior temporal consistency and state-of-the-art (SOTA) reconstruction quality by employing a novel training strategy with full frame masking. Second, we present MotionAura, a text-to-video generation framework that utilizes vector-quantized diffusion models to discretize the latent space and capture complex motion dynamics, producing temporally coherent videos aligned with text prompts. Third, we propose a spectral transformer-based denoising network that processes video data in the frequency domain using the Fourier Transform. This method effectively captures global context and long-range dependencies for high-quality video generation and denoising. Lastly, we introduce a downstream task of Sketch Guided Video Inpainting. This task leverages Low-Rank Adaptation (LoRA) for parameter-efficient fine-tuning. Our models achieve SOTA performance on a range of benchmarks. Our work offers robust frameworks for spatiotemporal modeling and user-driven video content manipulation. We will release the code, datasets, and models in open-source.

We present CARTO, a novel approach for reconstructing multiple articulated objects from a single stereo RGB observation. We use implicit object-centric representations and learn a single geometry and articulation decoder for multiple object categories. Despite training on multiple categories, our decoder achieves a comparable reconstruction accuracy to methods that train bespoke decoders separately for each category. Combined with our stereo image encoder we infer the 3D shape, 6D pose, size, joint type, and the joint state of multiple unknown objects in a single forward pass. Our method achieves a 20.4% absolute improvement in mAP 3D IOU50 for novel instances when compared to a two-stage pipeline. Inference time is fast and can run on a NVIDIA TITAN XP GPU at 1 HZ for eight or less objects present. While only trained on simulated data, CARTO transfers to real-world object instances. Code and evaluation data is available at: http://carto.cs.uni-freiburg.de

We introduce a self-supervised method for learning visual correspondence from unlabeled video. The main idea is to use cycle-consistency in time as free supervisory signal for learning visual representations from scratch. At training time, our model learns a feature map representation to be useful for performing cycle-consistent tracking. At test time, we use the acquired representation to find nearest neighbors across space and time. We demonstrate the generalizability of the representation -- without finetuning -- across a range of visual correspondence tasks, including video object segmentation, keypoint tracking, and optical flow. Our approach outperforms previous self-supervised methods and performs competitively with strongly supervised methods.

We build upon time-series classification by leveraging the capabilities of Vision Language Models (VLMs). We find that VLMs produce competitive results after two or less epochs of fine-tuning. We develop a novel approach that incorporates graphical data representations as images in conjunction with numerical data. This approach is rooted in the hypothesis that graphical representations can provide additional contextual information that numerical data alone may not capture. Additionally, providing a graphical representation can circumvent issues such as limited context length faced by LLMs. To further advance this work, we implemented a scalable end-to-end pipeline for training on different scenarios, allowing us to isolate the most effective strategies for transferring learning capabilities from LLMs to Time Series Classification (TSC) tasks. Our approach works with univariate and multivariate time-series data. In addition, we conduct extensive and practical experiments to show how this approach works for time-series classification and generative labels.

We present a new data-driven video inpainting method for recovering missing regions of video frames. A novel deep learning architecture is proposed which contains two sub-networks: a temporal structure inference network and a spatial detail recovering network. The temporal structure inference network is built upon a 3D fully convolutional architecture: it only learns to complete a low-resolution video volume given the expensive computational cost of 3D convolution. The low resolution result provides temporal guidance to the spatial detail recovering network, which performs image-based inpainting with a 2D fully convolutional network to produce recovered video frames in their original resolution. Such two-step network design ensures both the spatial quality of each frame and the temporal coherence across frames. Our method jointly trains both sub-networks in an end-to-end manner. We provide qualitative and quantitative evaluation on three datasets, demonstrating that our method outperforms previous learning-based video inpainting methods.

Vision-language models (VLMs) are typically composed of a vision encoder, e.g. CLIP, and a language model (LM) that interprets the encoded features to solve downstream tasks. Despite remarkable progress, VLMs are subject to several shortcomings due to the limited capabilities of vision encoders, e.g. "blindness" to certain image features, visual hallucination, etc. To address these issues, we study broadening the visual encoding capabilities of VLMs. We first comprehensively benchmark several vision encoders with different inductive biases for solving VLM tasks. We observe that there is no single encoding configuration that consistently achieves top performance across different tasks, and encoders with different biases can perform surprisingly similarly. Motivated by this, we introduce a method, named BRAVE, that consolidates features from multiple frozen encoders into a more versatile representation that can be directly fed as the input to a frozen LM. BRAVE achieves state-of-the-art performance on a broad range of captioning and VQA benchmarks and significantly reduces the aforementioned issues of VLMs, while requiring a smaller number of trainable parameters than existing methods and having a more compressed representation. Our results highlight the potential of incorporating different visual biases for a more broad and contextualized visual understanding of VLMs.

Recent advances in zero-shot image recognition suggest that vision-language models learn generic visual representations with a high degree of semantic information that may be arbitrarily probed with natural language phrases. Understanding an image, however, is not just about understanding what content resides within an image, but importantly, where that content resides. In this work we examine how well vision-language models are able to understand where objects reside within an image and group together visually related parts of the imagery. We demonstrate how contemporary vision and language representation learning models based on contrastive losses and large web-based data capture limited object localization information. We propose a minimal set of modifications that results in models that uniquely learn both semantic and spatial information. We measure this performance in terms of zero-shot image recognition, unsupervised bottom-up and top-down semantic segmentations, as well as robustness analyses. We find that the resulting model achieves state-of-the-art results in terms of unsupervised segmentation, and demonstrate that the learned representations are uniquely robust to spurious correlations in datasets designed to probe the causal behavior of vision models.

This paper introduces a novel method for self-supervised video representation learning via feature prediction. In contrast to the previous methods that focus on future feature prediction, we argue that a supervisory signal arising from unobserved past frames is complementary to one that originates from the future frames. The rationale behind our method is to encourage the network to explore the temporal structure of videos by distinguishing between future and past given present observations. We train our model in a contrastive learning framework, where joint encoding of future and past provides us with a comprehensive set of temporal hard negatives via swapping. We empirically show that utilizing both signals enriches the learned representations for the downstream task of action recognition. It outperforms independent prediction of future and past.

Self-supervised learning of image representations by predicting future frames is a promising direction but still remains a challenge. This is because of the under-determined nature of frame prediction; multiple potential futures can arise from a single current frame. To tackle this challenge, in this paper, we revisit the idea of stochastic video generation that learns to capture uncertainty in frame prediction and explore its effectiveness for representation learning. Specifically, we design a framework that trains a stochastic frame prediction model to learn temporal information between frames. Moreover, to learn dense information within each frame, we introduce an auxiliary masked image modeling objective along with a shared decoder architecture. We find this architecture allows for combining both objectives in a synergistic and compute-efficient manner. We demonstrate the effectiveness of our framework on a variety of tasks from video label propagation and vision-based robot learning domains, such as video segmentation, pose tracking, vision-based robotic locomotion, and manipulation tasks. Code is available on the project webpage: https://sites.google.com/view/2024rsp.

We present an architecture and a training recipe that adapts pre-trained open-world image models to localization in videos. Understanding the open visual world (without being constrained by fixed label spaces) is crucial for many real-world vision tasks. Contrastive pre-training on large image-text datasets has recently led to significant improvements for image-level tasks. For more structured tasks involving object localization applying pre-trained models is more challenging. This is particularly true for video tasks, where task-specific data is limited. We show successful transfer of open-world models by building on the OWL-ViT open-vocabulary detection model and adapting it to video by adding a transformer decoder. The decoder propagates object representations recurrently through time by using the output tokens for one frame as the object queries for the next. Our model is end-to-end trainable on video data and enjoys improved temporal consistency compared to tracking-by-detection baselines, while retaining the open-world capabilities of the backbone detector. We evaluate our model on the challenging TAO-OW benchmark and demonstrate that open-world capabilities, learned from large-scale image-text pre-training, can be transferred successfully to open-world localization across diverse videos.

Convolutional Neural Networks (CNNs) and Vision Transformers (ViTs) stand as the two most popular foundation models for visual representation learning. While CNNs exhibit remarkable scalability with linear complexity w.r.t. image resolution, ViTs surpass them in fitting capabilities despite contending with quadratic complexity. A closer inspection reveals that ViTs achieve superior visual modeling performance through the incorporation of global receptive fields and dynamic weights. This observation motivates us to propose a novel architecture that inherits these components while enhancing computational efficiency. To this end, we draw inspiration from the recently introduced state space model and propose the Visual State Space Model (VMamba), which achieves linear complexity without sacrificing global receptive fields. To address the encountered direction-sensitive issue, we introduce the Cross-Scan Module (CSM) to traverse the spatial domain and convert any non-causal visual image into order patch sequences. Extensive experimental results substantiate that VMamba not only demonstrates promising capabilities across various visual perception tasks, but also exhibits more pronounced advantages over established benchmarks as the image resolution increases. Source code has been available at https://github.com/MzeroMiko/VMamba.

Diffusion Transformers (DiTs) dominate video generation but their high computational cost severely limits real-world applicability, usually requiring tens of minutes to generate a few seconds of video even on high-performance GPUs. This inefficiency primarily arises from the quadratic computational complexity of 3D Full Attention with respect to the context length. In this paper, we propose a training-free framework termed Sparse VideoGen (SVG) that leverages the inherent sparsity in 3D Full Attention to boost inference efficiency. We reveal that the attention heads can be dynamically classified into two groups depending on distinct sparse patterns: (1) Spatial Head, where only spatially-related tokens within each frame dominate the attention output, and (2) Temporal Head, where only temporally-related tokens across different frames dominate. Based on this insight, SVG proposes an online profiling strategy to capture the dynamic sparse patterns and predicts the type of attention head. Combined with a novel hardware-efficient tensor layout transformation and customized kernel implementations, SVG achieves up to 2.28x and 2.33x end-to-end speedup on CogVideoX-v1.5 and HunyuanVideo, respectively, while preserving generation quality.

Despite encouraging progress in 3D scene understanding, it remains challenging to develop an effective Large Multi-modal Model (LMM) that is capable of understanding and reasoning in complex 3D environments. Most previous methods typically encode 3D point and 2D image features separately, neglecting interactions between 2D semantics and 3D object properties, as well as the spatial relationships within the 3D environment. This limitation not only hinders comprehensive representations of 3D scene, but also compromises training and inference efficiency. To address these challenges, we propose a unified Instance-aware 3D Large Multi-modal Model (Inst3D-LMM) to deal with multiple 3D scene understanding tasks simultaneously. To obtain the fine-grained instance-level visual tokens, we first introduce a novel Multi-view Cross-Modal Fusion (MCMF) module to inject the multi-view 2D semantics into their corresponding 3D geometric features. For scene-level relation-aware tokens, we further present a 3D Instance Spatial Relation (3D-ISR) module to capture the intricate pairwise spatial relationships among objects. Additionally, we perform end-to-end multi-task instruction tuning simultaneously without the subsequent task-specific fine-tuning. Extensive experiments demonstrate that our approach outperforms the state-of-the-art methods across 3D scene understanding, reasoning and grounding tasks. Source code is available at https://github.com/hanxunyu/Inst3D-LMM

In this paper, we propose PETRv2, a unified framework for 3D perception from multi-view images. Based on PETR, PETRv2 explores the effectiveness of temporal modeling, which utilizes the temporal information of previous frames to boost 3D object detection. More specifically, we extend the 3D position embedding (3D PE) in PETR for temporal modeling. The 3D PE achieves the temporal alignment on object position of different frames. A feature-guided position encoder is further introduced to improve the data adaptability of 3D PE. To support for multi-task learning (e.g., BEV segmentation and 3D lane detection), PETRv2 provides a simple yet effective solution by introducing task-specific queries, which are initialized under different spaces. PETRv2 achieves state-of-the-art performance on 3D object detection, BEV segmentation and 3D lane detection. Detailed robustness analysis is also conducted on PETR framework. We hope PETRv2 can serve as a strong baseline for 3D perception. Code is available at https://github.com/megvii-research/PETR.

We present a self-supervised learning approach for optical flow. Our method distills reliable flow estimations from non-occluded pixels, and uses these predictions as ground truth to learn optical flow for hallucinated occlusions. We further design a simple CNN to utilize temporal information from multiple frames for better flow estimation. These two principles lead to an approach that yields the best performance for unsupervised optical flow learning on the challenging benchmarks including MPI Sintel, KITTI 2012 and 2015. More notably, our self-supervised pre-trained model provides an excellent initialization for supervised fine-tuning. Our fine-tuned models achieve state-of-the-art results on all three datasets. At the time of writing, we achieve EPE=4.26 on the Sintel benchmark, outperforming all submitted methods.

In this technical report, we describe the Guided-Attention mechanism based solution for the short-term anticipation (STA) challenge for the EGO4D challenge. It combines the object detections, and the spatiotemporal features extracted from video clips, enhancing the motion and contextual information, and further decoding the object-centric and motion-centric information to address the problem of STA in egocentric videos. For the challenge, we build our model on top of StillFast with Guided Attention applied on fast network. Our model obtains better performance on the validation set and also achieves state-of-the-art (SOTA) results on the challenge test set for EGO4D Short-Term Object Interaction Anticipation Challenge.

Developing deep learning models that effectively learn object-centric representations, akin to human cognition, remains a challenging task. Existing approaches facilitate object discovery by representing objects as fixed-size vectors, called ``slots'' or ``object files''. While these approaches have shown promise in certain scenarios, they still exhibit certain limitations. First, they rely on architectural priors which can be unreliable and usually require meticulous engineering to identify the correct objects. Second, there has been a notable gap in investigating the practical utility of these representations in downstream tasks. To address the first limitation, we introduce a method that explicitly optimizes the constraint that each object in a scene should be associated with a distinct slot. We formalize this constraint by introducing consistency objectives which are cyclic in nature. By integrating these consistency objectives into various existing slot-based object-centric methods, we showcase substantial improvements in object-discovery performance. These enhancements consistently hold true across both synthetic and real-world scenes, underscoring the effectiveness and adaptability of the proposed approach. To tackle the second limitation, we apply the learned object-centric representations from the proposed method to two downstream reinforcement learning tasks, demonstrating considerable performance enhancements compared to conventional slot-based and monolithic representation learning methods. Our results suggest that the proposed approach not only improves object discovery, but also provides richer features for downstream tasks.

Learning object-centric representations of complex scenes is a promising step towards enabling efficient abstract reasoning from low-level perceptual features. Yet, most deep learning approaches learn distributed representations that do not capture the compositional properties of natural scenes. In this paper, we present the Slot Attention module, an architectural component that interfaces with perceptual representations such as the output of a convolutional neural network and produces a set of task-dependent abstract representations which we call slots. These slots are exchangeable and can bind to any object in the input by specializing through a competitive procedure over multiple rounds of attention. We empirically demonstrate that Slot Attention can extract object-centric representations that enable generalization to unseen compositions when trained on unsupervised object discovery and supervised property prediction tasks.

Scaling has not yet been convincingly demonstrated for pure self-supervised learning from video. However, prior work has focused evaluations on semantic-related tasks x2013 action classification, ImageNet classification, etc. In this paper we focus on evaluating self-supervised learning on non-semantic vision tasks that are more spatial (3D) and temporal (+1D = 4D), such as camera pose estimation, point and object tracking, and depth estimation. We show that by learning from very large video datasets, masked auto-encoding (MAE) with transformer video models actually scales, consistently improving performance on these 4D tasks, as model size increases from 20M all the way to the largest by far reported self-supervised video model x2013 22B parameters. Rigorous apples-to-apples comparison with many recent image and video models demonstrates the benefits of scaling 4D representations.

A central goal of visual recognition is to understand objects and scenes from a single image. 2D recognition has witnessed tremendous progress thanks to large-scale learning and general-purpose representations. Comparatively, 3D poses new challenges stemming from occlusions not depicted in the image. Prior works try to overcome these by inferring from multiple views or rely on scarce CAD models and category-specific priors which hinder scaling to novel settings. In this work, we explore single-view 3D reconstruction by learning generalizable representations inspired by advances in self-supervised learning. We introduce a simple framework that operates on 3D points of single objects or whole scenes coupled with category-agnostic large-scale training from diverse RGB-D videos. Our model, Multiview Compressive Coding (MCC), learns to compress the input appearance and geometry to predict the 3D structure by querying a 3D-aware decoder. MCC's generality and efficiency allow it to learn from large-scale and diverse data sources with strong generalization to novel objects imagined by DALLcdotE 2 or captured in-the-wild with an iPhone.

We propose a conditional positional encoding (CPE) scheme for vision Transformers. Unlike previous fixed or learnable positional encodings, which are pre-defined and independent of input tokens, CPE is dynamically generated and conditioned on the local neighborhood of the input tokens. As a result, CPE can easily generalize to the input sequences that are longer than what the model has ever seen during training. Besides, CPE can keep the desired translation-invariance in the image classification task, resulting in improved performance. We implement CPE with a simple Position Encoding Generator (PEG) to get seamlessly incorporated into the current Transformer framework. Built on PEG, we present Conditional Position encoding Vision Transformer (CPVT). We demonstrate that CPVT has visually similar attention maps compared to those with learned positional encodings and delivers outperforming results. Our code is available at https://github.com/Meituan-AutoML/CPVT .

Recently, there has been a growing trend toward feature-based approaches for Online Action Detection (OAD). However, these approaches have limitations due to their fixed backbone design, which ignores the potential capability of a trainable backbone. In this paper, we propose the first end-to-end OAD model, termed E2E-LOAD, designed to address the major challenge of OAD, namely, long-term understanding and efficient online reasoning. Specifically, our proposed approach adopts an initial spatial model that is shared by all frames and maintains a long sequence cache for inference at a low computational cost. We also advocate an asymmetric spatial-temporal model for long-form and short-form modeling effectively. Furthermore, we propose a novel and efficient inference mechanism that accelerates heavy spatial-temporal exploration. Extensive ablation studies and experiments demonstrate the effectiveness and efficiency of our proposed method. Notably, we achieve 17.3 (+12.6) FPS for end-to-end OAD with 72.4%~(+1.2%), 90.3%~(+0.7%), and 48.1%~(+26.0%) mAP on THMOUS14, TVSeries, and HDD, respectively, which is 3x faster than previous approaches. The source code will be made publicly available.

Visual search is important in our daily life. The efficient allocation of visual attention is critical to effectively complete visual search tasks. Prior research has predominantly modelled the spatial allocation of visual attention in images at the pixel level, e.g. using a saliency map. However, emerging evidence shows that visual attention is guided by objects rather than pixel intensities. This paper introduces the Object-level Attention Transformer (OAT), which predicts human scanpaths as they search for a target object within a cluttered scene of distractors. OAT uses an encoder-decoder architecture. The encoder captures information about the position and appearance of the objects within an image and about the target. The decoder predicts the gaze scanpath as a sequence of object fixations, by integrating output features from both the encoder and decoder. We also propose a new positional encoding that better reflects spatial relationships between objects. We evaluated OAT on the Amazon book cover dataset and a new dataset for visual search that we collected. OAT's predicted gaze scanpaths align more closely with human gaze patterns, compared to predictions by algorithms based on spatial attention on both established metrics and a novel behavioural-based metric. Our results demonstrate the generalization ability of OAT, as it accurately predicts human scanpaths for unseen layouts and target objects.

Visual scenes are composed of visual concepts and have the property of combinatorial explosion. An important reason for humans to efficiently learn from diverse visual scenes is the ability of compositional perception, and it is desirable for artificial intelligence to have similar abilities. Compositional scene representation learning is a task that enables such abilities. In recent years, various methods have been proposed to apply deep neural networks, which have been proven to be advantageous in representation learning, to learn compositional scene representations via reconstruction, advancing this research direction into the deep learning era. Learning via reconstruction is advantageous because it may utilize massive unlabeled data and avoid costly and laborious data annotation. In this survey, we first outline the current progress on reconstruction-based compositional scene representation learning with deep neural networks, including development history and categorizations of existing methods from the perspectives of the modeling of visual scenes and the inference of scene representations; then provide benchmarks, including an open source toolbox to reproduce the benchmark experiments, of representative methods that consider the most extensively studied problem setting and form the foundation for other methods; and finally discuss the limitations of existing methods and future directions of this research topic.

We present YOLO, a new approach to object detection. Prior work on object detection repurposes classifiers to perform detection. Instead, we frame object detection as a regression problem to spatially separated bounding boxes and associated class probabilities. A single neural network predicts bounding boxes and class probabilities directly from full images in one evaluation. Since the whole detection pipeline is a single network, it can be optimized end-to-end directly on detection performance. Our unified architecture is extremely fast. Our base YOLO model processes images in real-time at 45 frames per second. A smaller version of the network, Fast YOLO, processes an astounding 155 frames per second while still achieving double the mAP of other real-time detectors. Compared to state-of-the-art detection systems, YOLO makes more localization errors but is far less likely to predict false detections where nothing exists. Finally, YOLO learns very general representations of objects. It outperforms all other detection methods, including DPM and R-CNN, by a wide margin when generalizing from natural images to artwork on both the Picasso Dataset and the People-Art Dataset.

In this paper, we introduce a novel visual representation learning which relies on a handful of adaptively learned tokens, and which is applicable to both image and video understanding tasks. Instead of relying on hand-designed splitting strategies to obtain visual tokens and processing a large number of densely sampled patches for attention, our approach learns to mine important tokens in visual data. This results in efficiently and effectively finding a few important visual tokens and enables modeling of pairwise attention between such tokens, over a longer temporal horizon for videos, or the spatial content in images. Our experiments demonstrate strong performance on several challenging benchmarks for both image and video recognition tasks. Importantly, due to our tokens being adaptive, we accomplish competitive results at significantly reduced compute amount. We obtain comparable results to the state-of-the-arts on ImageNet while being computationally more efficient. We also confirm the effectiveness of the approach on multiple video datasets, including Kinetics-400, Kinetics-600, Charades, and AViD. The code is available at: https://github.com/google-research/scenic/tree/main/scenic/projects/token_learner

We present VideoGPT: a conceptually simple architecture for scaling likelihood based generative modeling to natural videos. VideoGPT uses VQ-VAE that learns downsampled discrete latent representations of a raw video by employing 3D convolutions and axial self-attention. A simple GPT-like architecture is then used to autoregressively model the discrete latents using spatio-temporal position encodings. Despite the simplicity in formulation and ease of training, our architecture is able to generate samples competitive with state-of-the-art GAN models for video generation on the BAIR Robot dataset, and generate high fidelity natural videos from UCF-101 and Tumbler GIF Dataset (TGIF). We hope our proposed architecture serves as a reproducible reference for a minimalistic implementation of transformer based video generation models. Samples and code are available at https://wilson1yan.github.io/videogpt/index.html

Image-based visual-language (I-VL) pre-training has shown great success for learning joint visual-textual representations from large-scale web data, revealing remarkable ability for zero-shot generalisation. This paper presents a simple but strong baseline to efficiently adapt the pre-trained I-VL model, and exploit its powerful ability for resource-hungry video understanding tasks, with minimal training. Specifically, we propose to optimise a few random vectors, termed as continuous prompt vectors, that convert video-related tasks into the same format as the pre-training objectives. In addition, to bridge the gap between static images and videos, temporal information is encoded with lightweight Transformers stacking on top of frame-wise visual features. Experimentally, we conduct extensive ablation studies to analyse the critical components. On 10 public benchmarks of action recognition, action localisation, and text-video retrieval, across closed-set, few-shot, and zero-shot scenarios, we achieve competitive or state-of-the-art performance to existing methods, despite optimising significantly fewer parameters.

In this work, we introduce Dual Attention Vision Transformers (DaViT), a simple yet effective vision transformer architecture that is able to capture global context while maintaining computational efficiency. We propose approaching the problem from an orthogonal angle: exploiting self-attention mechanisms with both "spatial tokens" and "channel tokens". With spatial tokens, the spatial dimension defines the token scope, and the channel dimension defines the token feature dimension. With channel tokens, we have the inverse: the channel dimension defines the token scope, and the spatial dimension defines the token feature dimension. We further group tokens along the sequence direction for both spatial and channel tokens to maintain the linear complexity of the entire model. We show that these two self-attentions complement each other: (i) since each channel token contains an abstract representation of the entire image, the channel attention naturally captures global interactions and representations by taking all spatial positions into account when computing attention scores between channels; (ii) the spatial attention refines the local representations by performing fine-grained interactions across spatial locations, which in turn helps the global information modeling in channel attention. Extensive experiments show our DaViT achieves state-of-the-art performance on four different tasks with efficient computations. Without extra data, DaViT-Tiny, DaViT-Small, and DaViT-Base achieve 82.8%, 84.2%, and 84.6% top-1 accuracy on ImageNet-1K with 28.3M, 49.7M, and 87.9M parameters, respectively. When we further scale up DaViT with 1.5B weakly supervised image and text pairs, DaViT-Gaint reaches 90.4% top-1 accuracy on ImageNet-1K. Code is available at https://github.com/dingmyu/davit.

Building animatable 3D models is challenging due to the need for 3D scans, laborious registration, and manual rigging, which are difficult to scale to arbitrary categories. Recently, differentiable rendering provides a pathway to obtain high-quality 3D models from monocular videos, but these are limited to rigid categories or single instances. We present RAC that builds category 3D models from monocular videos while disentangling variations over instances and motion over time. Three key ideas are introduced to solve this problem: (1) specializing a skeleton to instances via optimization, (2) a method for latent space regularization that encourages shared structure across a category while maintaining instance details, and (3) using 3D background models to disentangle objects from the background. We show that 3D models of humans, cats, and dogs can be learned from 50-100 internet videos.

Recent studies have achieved outstanding success in explaining 2D image recognition ConvNets. On the other hand, due to the computation cost and complexity of video data, the explanation of 3D video recognition ConvNets is relatively less studied. In this paper, we present a 3D ACE (Automatic Concept-based Explanation) framework for interpreting 3D ConvNets. In our approach: (1) videos are represented using high-level supervoxels, which is straightforward for human to understand; and (2) the interpreting framework estimates a score for each voxel, which reflects its importance in the decision procedure. Experiments show that our method can discover spatial-temporal concepts of different importance-levels, and thus can explore the influence of the concepts on a target task, such as action classification, in-depth. The codes are publicly available.

We propose WaveMix -- a novel neural architecture for computer vision that is resource-efficient yet generalizable and scalable. WaveMix networks achieve comparable or better accuracy than the state-of-the-art convolutional neural networks, vision transformers, and token mixers for several tasks, establishing new benchmarks for segmentation on Cityscapes; and for classification on Places-365, five EMNIST datasets, and iNAT-mini. Remarkably, WaveMix architectures require fewer parameters to achieve these benchmarks compared to the previous state-of-the-art. Moreover, when controlled for the number of parameters, WaveMix requires lesser GPU RAM, which translates to savings in time, cost, and energy. To achieve these gains we used multi-level two-dimensional discrete wavelet transform (2D-DWT) in WaveMix blocks, which has the following advantages: (1) It reorganizes spatial information based on three strong image priors -- scale-invariance, shift-invariance, and sparseness of edges, (2) in a lossless manner without adding parameters, (3) while also reducing the spatial sizes of feature maps, which reduces the memory and time required for forward and backward passes, and (4) expanding the receptive field faster than convolutions do. The whole architecture is a stack of self-similar and resolution-preserving WaveMix blocks, which allows architectural flexibility for various tasks and levels of resource availability. Our code and trained models are publicly available.

Remote sensing imagery is dense with objects and contextual visual information. There is a recent trend to combine paired satellite images and text captions for pretraining performant encoders for downstream tasks. However, while contrastive image-text methods like CLIP enable vision-language alignment and zero-shot classification ability, vision-only downstream performance tends to degrade compared to image-only pretraining, such as MAE. In this paper, we propose FLAVARS, a pretraining method that combines the best of both contrastive learning and masked modeling, along with geospatial alignment via contrastive location encoding. We find that FLAVARS significantly outperforms a baseline of SkyCLIP for vision-only tasks such as KNN classification and semantic segmentation, +6\% mIOU on SpaceNet1, while retaining the ability to perform zero-shot classification, unlike MAE pretrained methods.

We present a model for temporally precise action spotting in videos, which uses a dense set of detection anchors, predicting a detection confidence and corresponding fine-grained temporal displacement for each anchor. We experiment with two trunk architectures, both of which are able to incorporate large temporal contexts while preserving the smaller-scale features required for precise localization: a one-dimensional version of a u-net, and a Transformer encoder (TE). We also suggest best practices for training models of this kind, by applying Sharpness-Aware Minimization (SAM) and mixup data augmentation. We achieve a new state-of-the-art on SoccerNet-v2, the largest soccer video dataset of its kind, with marked improvements in temporal localization. Additionally, our ablations show: the importance of predicting the temporal displacements; the trade-offs between the u-net and TE trunks; and the benefits of training with SAM and mixup.

We propose an approach to self-supervised representation learning based on maximizing mutual information between features extracted from multiple views of a shared context. For example, one could produce multiple views of a local spatio-temporal context by observing it from different locations (e.g., camera positions within a scene), and via different modalities (e.g., tactile, auditory, or visual). Or, an ImageNet image could provide a context from which one produces multiple views by repeatedly applying data augmentation. Maximizing mutual information between features extracted from these views requires capturing information about high-level factors whose influence spans multiple views -- e.g., presence of certain objects or occurrence of certain events. Following our proposed approach, we develop a model which learns image representations that significantly outperform prior methods on the tasks we consider. Most notably, using self-supervised learning, our model learns representations which achieve 68.1% accuracy on ImageNet using standard linear evaluation. This beats prior results by over 12% and concurrent results by 7%. When we extend our model to use mixture-based representations, segmentation behaviour emerges as a natural side-effect. Our code is available online: https://github.com/Philip-Bachman/amdim-public.

How to effectively explore spatial-temporal features is important for video colorization. Instead of stacking multiple frames along the temporal dimension or recurrently propagating estimated features that will accumulate errors or cannot explore information from far-apart frames, we develop a memory-based feature propagation module that can establish reliable connections with features from far-apart frames and alleviate the influence of inaccurately estimated features. To extract better features from each frame for the above-mentioned feature propagation, we explore the features from large-pretrained visual models to guide the feature estimation of each frame so that the estimated features can model complex scenarios. In addition, we note that adjacent frames usually contain similar contents. To explore this property for better spatial and temporal feature utilization, we develop a local attention module to aggregate the features from adjacent frames in a spatial-temporal neighborhood. We formulate our memory-based feature propagation module, large-pretrained visual model guided feature estimation module, and local attention module into an end-to-end trainable network (named ColorMNet) and show that it performs favorably against state-of-the-art methods on both the benchmark datasets and real-world scenarios. The source code and pre-trained models will be available at https://github.com/yyang181/colormnet.

Existing video object segmentation (VOS) benchmarks focus on short-term videos which just last about 3-5 seconds and where objects are visible most of the time. These videos are poorly representative of practical applications, and the absence of long-term datasets restricts further investigation of VOS on the application in realistic scenarios. So, in this paper, we present a new benchmark dataset named LVOS, which consists of 220 videos with a total duration of 421 minutes. To the best of our knowledge, LVOS is the first densely annotated long-term VOS dataset. The videos in our LVOS last 1.59 minutes on average, which is 20 times longer than videos in existing VOS datasets. Each video includes various attributes, especially challenges deriving from the wild, such as long-term reappearing and cross-temporal similar objeccts.Based on LVOS, we assess existing video object segmentation algorithms and propose a Diverse Dynamic Memory network (DDMemory) that consists of three complementary memory banks to exploit temporal information adequately. The experimental results demonstrate the strength and weaknesses of prior methods, pointing promising directions for further study. Data and code are available at https://lingyihongfd.github.io/lvos.github.io/.

We introduce Human-like Video Models (HVM-1), large-scale video models pretrained with nearly 5000 hours of curated human-like video data (mostly egocentric, temporally extended, continuous video recordings), using the spatiotemporal masked autoencoder (ST-MAE) algorithm. We release two 633M parameter models trained at spatial resolutions of 224x224 and 448x448 pixels. We evaluate the performance of these models in downstream few-shot video and image recognition tasks and compare them against a model pretrained with 1330 hours of short action-oriented video clips from YouTube (Kinetics-700). HVM-1 models perform competitively against the Kinetics-700 pretrained model in downstream evaluations despite substantial qualitative differences between the spatiotemporal characteristics of the corresponding pretraining datasets. HVM-1 models also learn more accurate and more robust object representations compared to models pretrained with the image-based MAE algorithm on the same data, demonstrating the potential benefits of learning to predict temporal regularities in natural videos for learning better object representations.

Although DALL-E has shown an impressive ability of composition-based systematic generalization in image generation, it requires the dataset of text-image pairs and the compositionality is provided by the text. In contrast, object-centric representation models like the Slot Attention model learn composable representations without the text prompt. However, unlike DALL-E its ability to systematically generalize for zero-shot generation is significantly limited. In this paper, we propose a simple but novel slot-based autoencoding architecture, called SLATE, for combining the best of both worlds: learning object-centric representations that allows systematic generalization in zero-shot image generation without text. As such, this model can also be seen as an illiterate DALL-E model. Unlike the pixel-mixture decoders of existing object-centric representation models, we propose to use the Image GPT decoder conditioned on the slots for capturing complex interactions among the slots and pixels. In experiments, we show that this simple and easy-to-implement architecture not requiring a text prompt achieves significant improvement in in-distribution and out-of-distribution (zero-shot) image generation and qualitatively comparable or better slot-attention structure than the models based on mixture decoders.

High dynamic range (HDR) imaging aims to retrieve information from multiple low-dynamic range inputs to generate realistic output. The essence is to leverage the contextual information, including both dynamic and static semantics, for better image generation. Existing methods often focus on the spatial misalignment across input frames caused by the foreground and/or camera motion. However, there is no research on jointly leveraging the dynamic and static context in a simultaneous manner. To delve into this problem, we propose a novel alignment-free network with a Semantics Consistent Transformer (SCTNet) with both spatial and channel attention modules in the network. The spatial attention aims to deal with the intra-image correlation to model the dynamic motion, while the channel attention enables the inter-image intertwining to enhance the semantic consistency across frames. Aside from this, we introduce a novel realistic HDR dataset with more variations in foreground objects, environmental factors, and larger motions. Extensive comparisons on both conventional datasets and ours validate the effectiveness of our method, achieving the best trade-off on the performance and the computational cost.

We propose Spatio-temporal Crop Aggregation for video representation LEarning (SCALE), a novel method that enjoys high scalability at both training and inference time. Our model builds long-range video features by learning from sets of video clip-level features extracted with a pre-trained backbone. To train the model, we propose a self-supervised objective consisting of masked clip feature prediction. We apply sparsity to both the input, by extracting a random set of video clips, and to the loss function, by only reconstructing the sparse inputs. Moreover, we use dimensionality reduction by working in the latent space of a pre-trained backbone applied to single video clips. These techniques make our method not only extremely efficient to train but also highly effective in transfer learning. We demonstrate that our video representation yields state-of-the-art performance with linear, non-linear, and KNN probing on common action classification and video understanding datasets.

Referring video object segmentation aims to segment a referent throughout a video sequence according to a natural language expression. It requires aligning the natural language expression with the objects' motions and their dynamic associations at the global video level but segmenting objects at the frame level. To achieve this goal, we propose to simultaneously maintain a global referent token and a sequence of object queries, where the former is responsible for capturing video-level referent according to the language expression, while the latter serves to better locate and segment objects with each frame. Furthermore, to explicitly capture object motions and spatial-temporal cross-modal reasoning over objects, we propose a novel temporal collection-distribution mechanism for interacting between the global referent token and object queries. Specifically, the temporal collection mechanism collects global information for the referent token from object queries to the temporal motions to the language expression. In turn, the temporal distribution first distributes the referent token to the referent sequence across all frames and then performs efficient cross-frame reasoning between the referent sequence and object queries in every frame. Experimental results show that our method outperforms state-of-the-art methods on all benchmarks consistently and significantly.

We present L4GM, the first 4D Large Reconstruction Model that produces animated objects from a single-view video input -- in a single feed-forward pass that takes only a second. Key to our success is a novel dataset of multiview videos containing curated, rendered animated objects from Objaverse. This dataset depicts 44K diverse objects with 110K animations rendered in 48 viewpoints, resulting in 12M videos with a total of 300M frames. We keep our L4GM simple for scalability and build directly on top of LGM, a pretrained 3D Large Reconstruction Model that outputs 3D Gaussian ellipsoids from multiview image input. L4GM outputs a per-frame 3D Gaussian Splatting representation from video frames sampled at a low fps and then upsamples the representation to a higher fps to achieve temporal smoothness. We add temporal self-attention layers to the base LGM to help it learn consistency across time, and utilize a per-timestep multiview rendering loss to train the model. The representation is upsampled to a higher framerate by training an interpolation model which produces intermediate 3D Gaussian representations. We showcase that L4GM that is only trained on synthetic data generalizes extremely well on in-the-wild videos, producing high quality animated 3D assets.

We extend neural radiance fields (NeRFs) to dynamic large-scale urban scenes. Prior work tends to reconstruct single video clips of short durations (up to 10 seconds). Two reasons are that such methods (a) tend to scale linearly with the number of moving objects and input videos because a separate model is built for each and (b) tend to require supervision via 3D bounding boxes and panoptic labels, obtained manually or via category-specific models. As a step towards truly open-world reconstructions of dynamic cities, we introduce two key innovations: (a) we factorize the scene into three separate hash table data structures to efficiently encode static, dynamic, and far-field radiance fields, and (b) we make use of unlabeled target signals consisting of RGB images, sparse LiDAR, off-the-shelf self-supervised 2D descriptors, and most importantly, 2D optical flow. Operationalizing such inputs via photometric, geometric, and feature-metric reconstruction losses enables SUDS to decompose dynamic scenes into the static background, individual objects, and their motions. When combined with our multi-branch table representation, such reconstructions can be scaled to tens of thousands of objects across 1.2 million frames from 1700 videos spanning geospatial footprints of hundreds of kilometers, (to our knowledge) the largest dynamic NeRF built to date. We present qualitative initial results on a variety of tasks enabled by our representations, including novel-view synthesis of dynamic urban scenes, unsupervised 3D instance segmentation, and unsupervised 3D cuboid detection. To compare to prior work, we also evaluate on KITTI and Virtual KITTI 2, surpassing state-of-the-art methods that rely on ground truth 3D bounding box annotations while being 10x quicker to train.

Spatially dense self-supervised learning is a rapidly growing problem domain with promising applications for unsupervised segmentation and pretraining for dense downstream tasks. Despite the abundance of temporal data in the form of videos, this information-rich source has been largely overlooked. Our paper aims to address this gap by proposing a novel approach that incorporates temporal consistency in dense self-supervised learning. While methods designed solely for images face difficulties in achieving even the same performance on videos, our method improves not only the representation quality for videos-but also images. Our approach, which we call time-tuning, starts from image-pretrained models and fine-tunes them with a novel self-supervised temporal-alignment clustering loss on unlabeled videos. This effectively facilitates the transfer of high-level information from videos to image representations. Time-tuning improves the state-of-the-art by 8-10% for unsupervised semantic segmentation on videos and matches it for images. We believe this method paves the way for further self-supervised scaling by leveraging the abundant availability of videos. The implementation can be found here : https://github.com/SMSD75/Timetuning

Modeling and re-rendering dynamic 3D scenes is a challenging task in 3D vision. Prior approaches build on NeRF and rely on implicit representations. This is slow since it requires many MLP evaluations, constraining real-world applications. We show that dynamic 3D scenes can be explicitly represented by six planes of learned features, leading to an elegant solution we call HexPlane. A HexPlane computes features for points in spacetime by fusing vectors extracted from each plane, which is highly efficient. Pairing a HexPlane with a tiny MLP to regress output colors and training via volume rendering gives impressive results for novel view synthesis on dynamic scenes, matching the image quality of prior work but reducing training time by more than 100times. Extensive ablations confirm our HexPlane design and show that it is robust to different feature fusion mechanisms, coordinate systems, and decoding mechanisms. HexPlane is a simple and effective solution for representing 4D volumes, and we hope they can broadly contribute to modeling spacetime for dynamic 3D scenes.

We introduce Ev-TTA, a simple, effective test-time adaptation algorithm for event-based object recognition. While event cameras are proposed to provide measurements of scenes with fast motions or drastic illumination changes, many existing event-based recognition algorithms suffer from performance deterioration under extreme conditions due to significant domain shifts. Ev-TTA mitigates the severe domain gaps by fine-tuning the pre-trained classifiers during the test phase using loss functions inspired by the spatio-temporal characteristics of events. Since the event data is a temporal stream of measurements, our loss function enforces similar predictions for adjacent events to quickly adapt to the changed environment online. Also, we utilize the spatial correlations between two polarities of events to handle noise under extreme illumination, where different polarities of events exhibit distinctive noise distributions. Ev-TTA demonstrates a large amount of performance gain on a wide range of event-based object recognition tasks without extensive additional training. Our formulation can be successfully applied regardless of input representations and further extended into regression tasks. We expect Ev-TTA to provide the key technique to deploy event-based vision algorithms in challenging real-world applications where significant domain shift is inevitable.

Complex 3D scene understanding has gained increasing attention, with scene encoding strategies playing a crucial role in this success. However, the optimal scene encoding strategies for various scenarios remain unclear, particularly compared to their image-based counterparts. To address this issue, we present a comprehensive study that probes various visual encoding models for 3D scene understanding, identifying the strengths and limitations of each model across different scenarios. Our evaluation spans seven vision foundation encoders, including image-based, video-based, and 3D foundation models. We evaluate these models in four tasks: Vision-Language Scene Reasoning, Visual Grounding, Segmentation, and Registration, each focusing on different aspects of scene understanding. Our evaluations yield key findings: DINOv2 demonstrates superior performance, video models excel in object-level tasks, diffusion models benefit geometric tasks, and language-pretrained models show unexpected limitations in language-related tasks. These insights challenge some conventional understandings, provide novel perspectives on leveraging visual foundation models, and highlight the need for more flexible encoder selection in future vision-language and scene-understanding tasks.

Depth Anything has achieved remarkable success in monocular depth estimation with strong generalization ability. However, it suffers from temporal inconsistency in videos, hindering its practical applications. Various methods have been proposed to alleviate this issue by leveraging video generation models or introducing priors from optical flow and camera poses. Nonetheless, these methods are only applicable to short videos (< 10 seconds) and require a trade-off between quality and computational efficiency. We propose Video Depth Anything for high-quality, consistent depth estimation in super-long videos (over several minutes) without sacrificing efficiency. We base our model on Depth Anything V2 and replace its head with an efficient spatial-temporal head. We design a straightforward yet effective temporal consistency loss by constraining the temporal depth gradient, eliminating the need for additional geometric priors. The model is trained on a joint dataset of video depth and unlabeled images, similar to Depth Anything V2. Moreover, a novel key-frame-based strategy is developed for long video inference. Experiments show that our model can be applied to arbitrarily long videos without compromising quality, consistency, or generalization ability. Comprehensive evaluations on multiple video benchmarks demonstrate that our approach sets a new state-of-the-art in zero-shot video depth estimation. We offer models of different scales to support a range of scenarios, with our smallest model capable of real-time performance at 30 FPS.

Occlusions between consecutive frames have long posed a significant challenge in optical flow estimation. The inherent ambiguity introduced by occlusions directly violates the brightness constancy constraint and considerably hinders pixel-to-pixel matching. To address this issue, multi-frame optical flow methods leverage adjacent frames to mitigate the local ambiguity. Nevertheless, prior multi-frame methods predominantly adopt recursive flow estimation, resulting in a considerable computational overlap. In contrast, we propose a streamlined in-batch framework that eliminates the need for extensive redundant recursive computations while concurrently developing effective spatio-temporal modeling approaches under in-batch estimation constraints. Specifically, we present a Streamlined In-batch Multi-frame (SIM) pipeline tailored to video input, attaining a similar level of time efficiency to two-frame networks. Furthermore, we introduce an efficient Integrative Spatio-temporal Coherence (ISC) modeling method for effective spatio-temporal modeling during the encoding phase, which introduces no additional parameter overhead. Additionally, we devise a Global Temporal Regressor (GTR) that effectively explores temporal relations during decoding. Benefiting from the efficient SIM pipeline and effective modules, StreamFlow not only excels in terms of performance on the challenging KITTI and Sintel datasets, with particular improvement in occluded areas but also attains a remarkable 63.82% enhancement in speed compared with previous multi-frame methods. The code will be available soon at https://github.com/littlespray/StreamFlow.

We investigate the internal representations of vision-language models (VLMs) to address hallucinations, a persistent challenge despite advances in model size and training. We project VLMs' internal image representations to their language vocabulary and observe more confident output probabilities on real objects than hallucinated objects. We additionally use these output probabilities to spatially localize real objects. Building on this approach, we introduce a knowledge erasure algorithm that removes hallucinations by linearly orthogonalizing image features with respect to hallucinated object features. We show that targeted edits to a model's latent representations can reduce hallucinations by up to 25.7% on the COCO2014 dataset while preserving performance. Our findings demonstrate how a deeper understanding of VLMs' latent representations can enhance reliability and enable novel capabilities, such as zero-shot segmentation.

Masked Image Modeling (MIM) arises as a promising option for Vision Transformers among various self-supervised learning (SSL) methods. The essence of MIM lies in token-wise masked patch predictions, with targets patchified from images; or generated by pre-trained tokenizers or models. We argue targets from the pre-trained models usually exhibit spatial inconsistency, which makes it excessively challenging for the model to follow to learn more discriminative representations. To mitigate the issue, we introduce a novel self-supervision signal based on Dynamic Token Morphing (DTM), which dynamically aggregates contextually related tokens. DTM can be generally applied to various SSL frameworks, yet we propose a simple MIM that employs DTM to effectively improve the performance barely introducing extra training costs. Our experiments on ImageNet-1K and ADE20K evidently demonstrate the superiority of our methods. Furthermore, the comparative evaluation of iNaturalist and Fine-grained Visual Classification datasets further validates the transferability of our method on various downstream tasks. Our code will be released publicly.

We capitalize on large amounts of unlabeled video in order to learn a model of scene dynamics for both video recognition tasks (e.g. action classification) and video generation tasks (e.g. future prediction). We propose a generative adversarial network for video with a spatio-temporal convolutional architecture that untangles the scene's foreground from the background. Experiments suggest this model can generate tiny videos up to a second at full frame rate better than simple baselines, and we show its utility at predicting plausible futures of static images. Moreover, experiments and visualizations show the model internally learns useful features for recognizing actions with minimal supervision, suggesting scene dynamics are a promising signal for representation learning. We believe generative video models can impact many applications in video understanding and simulation.