Hugging Face's logo

Datasets:
ShuhuaiRen
/
TimeIT

Languages:
English
ArXiv:
License:
Dataset card Files
xet
Community
ShuhuaiRen commited on
Commit
078dbbd
·
1 Parent(s): caff49b

Upload 2 files

Browse files
Files changed (2) hide show
  1. .gitignore +6 -0
  2. README.md +67 -120
.gitignore ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ __pycache__/
2
+ .DS_Store
3
+ .idea
4
+ .ipynb_checkpoints
5
+ *.ipynb
6
+ *asr.json
README.md CHANGED
@@ -1,5 +1,7 @@
1
  ---
2
- license: mit
 
 
3
  ---
4
 
5
 
@@ -24,43 +26,43 @@ Our dataset compiles diverse tasks of time-sensitive long video understanding, i
24
 
25
  | Task | #Instructions |
26
  |-------------------------------|---------------|
27
- | Dense Video Captioning | |
28
- | Temporal Video Grounding | |
29
- | Video Summarization | |
30
- | Video Highlight Detection | |
31
- | Step Localization | |
32
- | Transcribed Speech Generation | |
33
- | Total | |
34
 
35
  ### Task Statistics
36
 
37
- | Task | Description | #Train | #Val | #Test |
38
- |-------------------------------|----------------------------------------------------------------------------------------------------------------------|---------|---------|---------|
39
- | Dense Video Captioning | detects a series of events in the given video and outputs the corresponding timestamps and descriptions |
40
- | Temporal Video Grounding | predict a timestamp boundary including the start and end time in the video given a natural language query |
41
- | Video Summarization | create a compressed set of frames or clip shots to represent the most informative content of the given video |
42
- | Video Highlight Detection | identify the most exciting, impressive, or emotional moments that may not cover the full scope of the original video |
43
- | Step Localization | segment and describe significant steps in a long untrimmed video |
44
- | Transcribed Speech Generation | predict the speech content and its corresponding start and end timestamps based on visual signals in the video |
45
- | Total | - |
46
 
47
  ### Detailed Dataset Statistics
48
 
49
- | Task | Dataset | #Train | #Val | #Test |
50
- |-------------------------------|------------------------|---------|--------|-------|
51
- | Dense Video Captioning | `ActivityNet Captions` | | | |
52
- | | `ViTT` | 97,765 | 13,965 | 0 |
53
- | | `YouCook2` | 14,575 | 2,487 | 2,489 |
54
- | Temporal Video Grounding | `DiDeMo` | 30,000 | 2,000 | 0 |
55
- | | `QuerYD` | 118,312 | 27,550 | 0 |
56
- | | `HiREST_grounding` | 30,000 | 50,000 | 0 |
57
- | | `Charades-STA` | 30,000 | 5,000 | 5,000 |
58
- | Video Summarization | `TVSum` | 30,000 | 30,000 | 0 |
59
- | | `SumMe` | 13,568 | 1,024 | 1,024 |
60
- | Video Highlight Detection | `QVHighlights` | 9,009 | 5,046 | 0 |
61
- | Step Localization | `COIN` | 30,000 | 2,000 | 0 |
62
- | | `HiREST_step` | 29,372 | 2,000 | 0 |
63
- | Transcribed Speech Generation | `YT-Temporal` | 5,000 | 4,315 | 4,350 |
64
 
65
  ## Dataset Structure
66
 
@@ -106,9 +108,9 @@ dataset = load_dataset("ShuhuaiRen/TimeIT", ds_name)
106
  train_set = dataset["train"]
107
 
108
  for train_instance in train_set:
109
- question = train_instance["QA"][0]['q'] # str
110
- answer = train_instance["QA"][0]['a'] # str
111
- video_path = train_instance["video"] # str
112
  ```
113
 
114
  ### Data Fields
@@ -118,10 +120,9 @@ import datasets
118
 
119
  features = datasets.Features(
120
  {
121
- "instruction": datasets.Value("string"),
122
- "inputs": datasets.Value("string"),
123
- "image_base64_str": [datasets.Value("string")],
124
- "outputs": datasets.Value("string"),
125
  }
126
  )
127
  ```
@@ -134,50 +135,21 @@ features = datasets.Features(
134
 
135
  ### Source Data
136
 
137
- | Task | Dataset [Citation] | Source |
138
- |---------------------------|----------------------------------|------------------------------------------------------------------------------------|
139
- | Image Captioning | `coco` [1] | [Source](https://cocodataset.org/#home) |
140
- | | `textcap` [2] | [Source](https://textvqa.org/textcaps/) |
141
- | | `image-paragraph-captioning` [3] | [Source](https://cs.stanford.edu/people/ranjaykrishna/im2p/index.html) |
142
- | Classification | `coco-goi` [1] | [Source](https://cocodataset.org/#home) |
143
- | | `coco-text` [4] | [Source](https://bgshih.github.io/cocotext/) |
144
- | | `imagenet` [5] | [Source](https://www.image-net.org/) |
145
- | | `coco-itm` [1] | [Source](https://cocodataset.org/#home) |
146
- | | `snli-ve` [6] | [Source](https://github.com/necla-ml/SNLI-VE) |
147
- | | `mocheg` [7] | [Source](https://github.com/VT-NLP/Mocheg) |
148
- | | `iqa` [8] | [Source](https://github.com/icbcbicc/IQA-Dataset) |
149
- | Visual Question Answering | `vqa-v2` [9] | [Source](https://visualqa.org/) |
150
- | | `shapes` [10] | [Source](https://github.com/ronghanghu/n2nmn) |
151
- | | `docvqa` [11] | [Source](https://www.docvqa.org/) |
152
- | | `ocr-vqa` [12] | [Source](https://ocr-vqa.github.io/) |
153
- | | `st-vqa` [13] | [Source](https://rrc.cvc.uab.es/?ch=11) |
154
- | | `text-vqa` [14] | [Source](https://textvqa.org/) |
155
- | | `gqa` [15] | [Source](https://cs.stanford.edu/people/dorarad/gqa/about.html) |
156
- | Knowledgeable Visual QA | `okvqa` [16] | [Source](https://okvqa.allenai.org/) |
157
- | | `a-okvqa` [17] | [Source](https://allenai.org/project/a-okvqa/home) |
158
- | | `science-qa` [18] | [Source](https://scienceqa.github.io/) |
159
- | | `viquae` [19] | [Source](https://github.com/PaulLerner/ViQuAE) |
160
- | Reasoning | `clevr` [20] | [Source](https://cs.stanford.edu/people/jcjohns/clevr/) |
161
- | | `nlvr` [21] | [Source](https://lil.nlp.cornell.edu/nlvr/) |
162
- | | `vcr` [22] | [Source](https://visualcommonsense.com/) |
163
- | | `visual-mrc` [23] | [Source](https://github.com/nttmdlab-nlp/VisualMRC) |
164
- | | `winoground` [24] | [Source](https://huggingface.co/datasets/facebook/winoground) |
165
- | Generation | `vist` [25] | [Source](https://visionandlanguage.net/VIST/) |
166
- | | `visual-dialog` [26] | [Source](https://visualdialog.org/) |
167
- | | `multi30k` [27] | [Source](https://github.com/multi30k/dataset) |
168
- | Chinese | `fm-iqa` [28] | [Source](https://paperswithcode.com/dataset/fm-iqa) |
169
- | | `coco-cn` [29] | [Source](https://github.com/li-xirong/coco-cn) |
170
- | | `flickr8k-cn` [30] | [Source](https://github.com/li-xirong/flickr8kcn) |
171
- | | `chinese-food` [31] | [Source](https://sites.google.com/view/chinesefoodnet) |
172
- | | `mmchat` [32] | [Source](https://github.com/silverriver/MMChat) |
173
- | Video | `ss` [33] | [Source](https://developer.qualcomm.com/software/ai-datasets/something-something) |
174
- | | `ivqa` [34] | [Source](https://antoyang.github.io/just-ask.html) |
175
- | | `msvd-qa` [35] | [Source](https://paperswithcode.com/dataset/msvd) |
176
- | | `activitynet-qa` [36] | [Source](https://github.com/MILVLG/activitynet-qa) |
177
- | | `msrvtt` [35] | [Source](https://paperswithcode.com/dataset/msr-vtt) |
178
- | | `msrvtt-qa` [37] | [Source](https://paperswithcode.com/sota/visual-question-answering-on-msrvtt-qa-1) |
179
-
180
-
181
 
182
  ### Annotations
183
 
@@ -225,40 +197,15 @@ designed to enable the development of general-purpose video agents.
225
 
226
  ## References
227
 
228
- - [1] Microsoft COCO: Common Objects in Context
229
- - [2] TextCaps: a dataset for image captioning with reading comprehension
230
- - [3] A Hierarchical Approach for Generating Descriptive Image Paragraphs
231
- - [4] COCO-Text: Dataset and benchmark for text detection and recognition in natural images
232
- - [5] Imagenet large scale visual recognition challenge
233
- - [6] E-ViL: A Dataset and Benchmark for Natural Language Explanations in Vision-Language Tasks
234
- - [7] End-to-End Multimodal Fact-Checking and Explanation Generation: A Challenging Dataset and Models
235
- - [8] Quantifying visual image quality: A Bayesian view
236
- - [9] Making the V in VQA Matter: Elevating the Role of Image Understanding in Visual Question Answering
237
- - [10] Neural Module Networks
238
- - [11] DocVQA: A dataset for vqa on document images
239
- - [12] OCR-VQA: Visual Question Answering by Reading Text in Images
240
- - [13] Scene Text Visual Question Answering
241
- - [14] Towards VQA Models That Can Read
242
- - [15] GQA: A new dataset for real-world visual reasoning and compositional question answering
243
- - [16] OK-VQA: A Visual Question Answering Benchmark Requiring External Knowledge
244
- - [17] A-OKVQA: A Benchmark for Visual Question Answering using World Knowledge
245
- - [18] Learn to Explain: Multimodal Reasoning via Thought Chains for Science Question Answering
246
- - [19] ViQuAE: a dataset for knowledge-based visual question answering about named entities
247
- - [20] CLEVR: A diagnostic dataset for compositional language and elementary visual reasoning
248
- - [21] A Corpus of Natural Language for Visual Reasoning
249
- - [22] From recognition to cognition: Visual Commonsense Reasoning
250
- - [23] VisualMRC: Machine reading comprehension on document images
251
- - [24] WinoGround: Probing vision and language models for visio-linguistic compositionality
252
- - [25] Visual Storytelling
253
- - [26] Visual Dialog
254
- - [27] Multi30k: Multilingual english-german image descriptions
255
- - [28] Are You Talking to a Machine? Dataset and Methods for Multilingual Image Question
256
- - [29] COCO-CN for cross-lingual image tagging, captioning, and retrieval
257
- - [30] Adding Chinese Captions to Images
258
- - [31] ChineseFoodNet: A large-scale image dataset for chinese food recognition
259
- - [32] MMChat: Multi-Modal Chat Dataset on Social Media
260
- - [33] The "Something Something" Video Database for Learning and Evaluating Visual Common Sense
261
- - [34] Just Ask: Learning to answer questions from millions of narrated videos
262
- - [35] Video Question Answering via Gradually Refined Attention over Appearance and Motion
263
- - [36] ActivityNet-qa: A dataset for understanding complex web videos via question answering
264
- - [37] MSR-VTT: A large video description dataset for bridging video and language
 
1
  ---
2
+ license: cc-by-4.0
3
+ language:
4
+ - en
5
  ---
6
 
7
 
 
26
 
27
  | Task | #Instructions |
28
  |-------------------------------|---------------|
29
+ | Dense Video Captioning | 6 |
30
+ | Temporal Video Grounding | 6 |
31
+ | Video Summarization | 6 |
32
+ | Video Highlight Detection | 6 |
33
+ | Step Localization | 6 |
34
+ | Transcribed Speech Generation | 6 |
35
+ | Total | 36 |
36
 
37
  ### Task Statistics
38
 
39
+ | Task | Description | #Train |
40
+ |-------------------------------|----------------------------------------------------------------------------------------------------------------------|---------|
41
+ | Dense Video Captioning | detects a series of events in the given video and outputs the corresponding timestamps and descriptions | 16,342 |
42
+ | Temporal Video Grounding | predict a timestamp boundary including the start and end time in the video given a natural language query | 60,471 |
43
+ | Video Summarization | create a compressed set of frames or clip shots to represent the most informative content of the given video | 75 |
44
+ | Video Highlight Detection | identify the most exciting, impressive, or emotional moments that may not cover the full scope of the original video | 6,858 |
45
+ | Step Localization | segment and describe significant steps in a long untrimmed video | 9,488 |
46
+ | Transcribed Speech Generation | predict the speech content and its corresponding start and end timestamps based on visual signals in the video | 31,627 |
47
+ | Total | - | 124861 |
48
 
49
  ### Detailed Dataset Statistics
50
 
51
+ | Task | Dataset | #Train |
52
+ |-------------------------------|------------------------|--------|
53
+ | Dense Video Captioning | `ActivityNet Captions` | 10,009 |
54
+ | | `ViTT` | 5,141 |
55
+ | | `YouCook2` | 1,192 |
56
+ | Temporal Video Grounding | `DiDeMo` | 33,002 |
57
+ | | `QuerYD` | 14,602 |
58
+ | | `HiREST_grounding` | 459 |
59
+ | | `Charades-STA` | 12,408 |
60
+ | Video Summarization | `TVSum` | 50 |
61
+ | | `SumMe` | 25 |
62
+ | Video Highlight Detection | `QVHighlights` | 6,858 |
63
+ | Step Localization | `COIN` | 9,029 |
64
+ | | `HiREST_step` | 459 |
65
+ | Transcribed Speech Generation | `YT-Temporal` | 31,627 |
66
 
67
  ## Dataset Structure
68
 
 
108
  train_set = dataset["train"]
109
 
110
  for train_instance in train_set:
111
+ question = train_instance["question"] # str
112
+ answer = train_instance["answer"] # str
113
+ video_path = train_instance["video_path"] # str
114
  ```
115
 
116
  ### Data Fields
 
120
 
121
  features = datasets.Features(
122
  {
123
+ "video_path": datasets.Value("string"),
124
+ "question": datasets.Value("string"),
125
+ "answer": datasets.Value("string"),
 
126
  }
127
  )
128
  ```
 
135
 
136
  ### Source Data
137
 
138
+ | Task | Dataset [Citation] | Source |
139
+ |-------------------------------|----------------------------|--------------------------------------------------------------------------------|
140
+ | Dense Video Captioning | `ActivityNet Captions` [1] | [Source](https://cs.stanford.edu/people/ranjaykrishna/densevid/) |
141
+ | | `ViTT` [2] | [Source](https://github.com/google-research-datasets/Video-Timeline-Tags-ViTT) |
142
+ | | `YouCook2` [3] | [Source](http://youcook2.eecs.umich.edu/) |
143
+ | Temporal Video Grounding | `DiDeMo` [4] | [Source](https://github.com/LisaAnne/TemporalLanguageRelease) |
144
+ | | `QuerYD` [5] | [Source](https://www.robots.ox.ac.uk/~vgg/data/queryd/) |
145
+ | | `HiREST_grounding` [6] | [Source](https://hirest-cvpr2023.github.io/) |
146
+ | | `Charades-STA` [7] | [Source](https://github.com/jiyanggao/TALL) |
147
+ | Video Summarization | `TVSum` [8] | [Source](https://github.com/yalesong/tvsum) |
148
+ | | `SumMe` [9] | [Source](http://classif.ai/dataset/ethz-cvl-video-summe/) |
149
+ | Video Highlight Detection | `QVHighlights` [10] | [Source](https://github.com/jayleicn/moment_detr/tree/main/data) |
150
+ | Step Localization | `COIN` [11] | [Source](https://coin-dataset.github.io/) |
151
+ | | `HiREST_step` [6] | [Source](https://hirest-cvpr2023.github.io/) |
152
+ | Transcribed Speech Generation | `YT-Temporal` [12] | [Source](https://rowanzellers.com/merlot/#data) |
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
153
 
154
  ### Annotations
155
 
 
197
 
198
  ## References
199
 
200
+ - [1] Dense-Captioning Events in Videos
201
+ - [2] Multimodal Pretraining for Dense Video Captioning
202
+ - [3] Towards Automatic Learning of Procedures from Web Instructional Videos
203
+ - [4] Localizing Moments in Video with Natural Language
204
+ - [5] QuerYD: A video dataset with high-quality text and audio narrations
205
+ - [6] Hierarchical Video-Moment Retrieval and Step-Captioning
206
+ - [7] TALL: Temporal Activity Localization via Language Query
207
+ - [8] TVSum: Summarizing Web Videos Using Titles
208
+ - [9] Creating Summaries from User Videos
209
+ - [10] QVHighlights: Detecting Moments and Highlights in Videos via Natural Language Queries
210
+ - [11] COIN: A Large-scale Dataset for Comprehensive Instructional Video Analysis
211
+ - [12] MERLOT: Multimodal Neural Script Knowledge Models