The SuperGlue model was proposed in SuperGlue: Learning Feature Matching with Graph Neural Networks by Paul-Edouard Sarlin, Daniel DeTone, Tomasz Malisiewicz and Andrew Rabinovich.
This model consists of matching two sets of interest points detected in an image. Paired with the SuperPoint model, it can be used to match two images and estimate the pose between them. This model is useful for tasks such as image matching, homography estimation, etc.
The abstract from the paper is the following:
This paper introduces SuperGlue, a neural network that matches two sets of local features by jointly finding correspondences and rejecting non-matchable points. Assignments are estimated by solving a differentiable optimal transport problem, whose costs are predicted by a graph neural network. We introduce a flexible context aggregation mechanism based on attention, enabling SuperGlue to reason about the underlying 3D scene and feature assignments jointly. Compared to traditional, hand-designed heuristics, our technique learns priors over geometric transformations and regularities of the 3D world through end-to-end training from image pairs. SuperGlue outperforms other learned approaches and achieves state-of-the-art results on the task of pose estimation in challenging real-world indoor and outdoor environments. The proposed method performs matching in real-time on a modern GPU and can be readily integrated into modern SfM or SLAM systems. The code and trained weights are publicly available at this URL.
Here is a quick example of using the model. Since this model is an image matching model, it requires pairs of images to be matched. The raw outputs contain the list of keypoints detected by the keypoint detector as well as the list of matches with their corresponding matching scores.
from transformers import AutoImageProcessor, AutoModel
import torch
from PIL import Image
import requests
url_image1 = "https://raw.githubusercontent.com/magicleap/SuperGluePretrainedNetwork/refs/heads/master/assets/phototourism_sample_images/united_states_capitol_98169888_3347710852.jpg"
image1 = Image.open(requests.get(url_image1, stream=True).raw)
url_image2 = "https://raw.githubusercontent.com/magicleap/SuperGluePretrainedNetwork/refs/heads/master/assets/phototourism_sample_images/united_states_capitol_26757027_6717084061.jpg"
image_2 = Image.open(requests.get(url_image2, stream=True).raw)
images = [image1, image2]
processor = AutoImageProcessor.from_pretrained("magic-leap-community/superglue_outdoor")
model = AutoModel.from_pretrained("magic-leap-community/superglue_outdoor")
inputs = processor(images, return_tensors="pt")
with torch.no_grad():
outputs = model(**inputs)
You can use the post_process_keypoint_matching
method from the SuperGlueImageProcessor
to get the keypoints and matches in a more readable format:
image_sizes = [[(image.height, image.width) for image in images]]
outputs = processor.post_process_keypoint_matching(outputs, image_sizes, threshold=0.2)
for i, output in enumerate(outputs):
print("For the image pair", i)
for keypoint0, keypoint1, matching_score in zip(
output["keypoints0"], output["keypoints1"], output["matching_scores"]
):
print(
f"Keypoint at coordinate {keypoint0.numpy()} in the first image matches with keypoint at coordinate {keypoint1.numpy()} in the second image with a score of {matching_score}."
)
From the outputs, you can visualize the matches between the two images using the following code:
import matplotlib.pyplot as plt
import numpy as np
# Create side by side image
merged_image = np.zeros((max(image1.height, image2.height), image1.width + image2.width, 3))
merged_image[: image1.height, : image1.width] = np.array(image1) / 255.0
merged_image[: image2.height, image1.width :] = np.array(image2) / 255.0
plt.imshow(merged_image)
plt.axis("off")
# Retrieve the keypoints and matches
output = outputs[0]
keypoints0 = output["keypoints0"]
keypoints1 = output["keypoints1"]
matching_scores = output["matching_scores"]
keypoints0_x, keypoints0_y = keypoints0[:, 0].numpy(), keypoints0[:, 1].numpy()
keypoints1_x, keypoints1_y = keypoints1[:, 0].numpy(), keypoints1[:, 1].numpy()
# Plot the matches
for keypoint0_x, keypoint0_y, keypoint1_x, keypoint1_y, matching_score in zip(
keypoints0_x, keypoints0_y, keypoints1_x, keypoints1_y, matching_scores
):
plt.plot(
[keypoint0_x, keypoint1_x + image1.width],
[keypoint0_y, keypoint1_y],
color=plt.get_cmap("RdYlGn")(matching_score.item()),
alpha=0.9,
linewidth=0.5,
)
plt.scatter(keypoint0_x, keypoint0_y, c="black", s=2)
plt.scatter(keypoint1_x + image1.width, keypoint1_y, c="black", s=2)
# Save the plot
plt.savefig("matched_image.png", dpi=300, bbox_inches='tight')
plt.close()
This model was contributed by stevenbucaille. The original code can be found here.
( keypoint_detector_config: SuperPointConfig = None hidden_size: int = 256 keypoint_encoder_sizes: typing.List[int] = None gnn_layers_types: typing.List[str] = None num_attention_heads: int = 4 sinkhorn_iterations: int = 100 matching_threshold: float = 0.0 initializer_range: float = 0.02 **kwargs )
Parameters
Union[AutoConfig, dict]
, optional, defaults to SuperPointConfig
) —
The config object or dictionary of the keypoint detector. int
, optional, defaults to 256) —
The dimension of the descriptors. List[int]
, optional, defaults to [32, 64, 128, 256]
) —
The sizes of the keypoint encoder layers. List[str]
, optional, defaults to ['self', 'cross', 'self', 'cross', 'self', 'cross', 'self', 'cross', 'self', 'cross', 'self', 'cross', 'self', 'cross', 'self', 'cross', 'self', 'cross']
) —
The types of the GNN layers. Must be either ‘self’ or ‘cross’. int
, optional, defaults to 4) —
The number of heads in the GNN layers. int
, optional, defaults to 100) —
The number of Sinkhorn iterations. float
, optional, defaults to 0.0) —
The matching threshold. float
, optional, defaults to 0.02) —
The standard deviation of the truncated_normal_initializer for initializing all weight matrices. This is the configuration class to store the configuration of a SuperGlueModel
. It is used to instantiate a
SuperGlue model according to the specified arguments, defining the model architecture. Instantiating a
configuration with the defaults will yield a similar configuration to that of the SuperGlue
magic-leap-community/superglue_indoor architecture.
Configuration objects inherit from PretrainedConfig and can be used to control the model outputs. Read the documentation from PretrainedConfig for more information.
Examples:
>>> from transformers import SuperGlueConfig, SuperGlueModel
>>> # Initializing a SuperGlue superglue style configuration
>>> configuration = SuperGlueConfig()
>>> # Initializing a model from the superglue style configuration
>>> model = SuperGlueModel(configuration)
>>> # Accessing the model configuration
>>> configuration = model.config
( do_resize: bool = True size: typing.Dict[str, int] = None resample: Resampling = <Resampling.BILINEAR: 2> do_rescale: bool = True rescale_factor: float = 0.00392156862745098 do_grayscale: bool = True **kwargs )
Parameters
bool
, optional, defaults to True
) —
Controls whether to resize the image’s (height, width) dimensions to the specified size
. Can be overriden
by do_resize
in the preprocess
method. Dict[str, int]
optional, defaults to {"height" -- 480, "width": 640}
):
Resolution of the output image after resize
is applied. Only has an effect if do_resize
is set to
True
. Can be overriden by size
in the preprocess
method. PILImageResampling
, optional, defaults to Resampling.BILINEAR
) —
Resampling filter to use if resizing the image. Can be overriden by resample
in the preprocess
method. bool
, optional, defaults to True
) —
Whether to rescale the image by the specified scale rescale_factor
. Can be overriden by do_rescale
in
the preprocess
method. int
or float
, optional, defaults to 1/255
) —
Scale factor to use if rescaling the image. Can be overriden by rescale_factor
in the preprocess
method. bool
, optional, defaults to True
) —
Whether to convert the image to grayscale. Can be overriden by do_grayscale
in the preprocess
method. Constructs a SuperGlue image processor.
( outputs: KeypointMatchingOutput target_sizes: typing.Union[transformers.utils.generic.TensorType, typing.List[typing.Tuple]] threshold: float = 0.0 ) → List[Dict]
Parameters
KeypointMatchingOutput
) —
Raw outputs of the model. torch.Tensor
or List[Tuple[Tuple[int, int]]]
, optional) —
Tensor of shape (batch_size, 2, 2)
or list of tuples of tuples (Tuple[int, int]
) containing the
target size (height, width)
of each image in the batch. This must be the original image size (before
any processing). float
, optional, defaults to 0.0) —
Threshold to filter out the matches with low scores. Returns
List[Dict]
A list of dictionaries, each dictionary containing the keypoints in the first and second image of the pair, the matching scores and the matching indices.
Converts the raw output of KeypointMatchingOutput
into lists of keypoints, scores and descriptors
with coordinates absolute to the original image sizes.
( images do_resize: bool = None size: typing.Dict[str, int] = None resample: Resampling = None do_rescale: bool = None rescale_factor: float = None do_grayscale: bool = None return_tensors: typing.Union[str, transformers.utils.generic.TensorType, NoneType] = None data_format: ChannelDimension = <ChannelDimension.FIRST: 'channels_first'> input_data_format: typing.Union[str, transformers.image_utils.ChannelDimension, NoneType] = None **kwargs )
Parameters
ImageInput
) —
Image pairs to preprocess. Expects either a list of 2 images or a list of list of 2 images list with
pixel values ranging from 0 to 255. If passing in images with pixel values between 0 and 1, set
do_rescale=False
. bool
, optional, defaults to self.do_resize
) —
Whether to resize the image. Dict[str, int]
, optional, defaults to self.size
) —
Size of the output image after resize
has been applied. If size["shortest_edge"]
>= 384, the image
is resized to (size["shortest_edge"], size["shortest_edge"])
. Otherwise, the smaller edge of the
image will be matched to int(size["shortest_edge"]/ crop_pct)
, after which the image is cropped to
(size["shortest_edge"], size["shortest_edge"])
. Only has an effect if do_resize
is set to True
. PILImageResampling
, optional, defaults to self.resample
) —
Resampling filter to use if resizing the image. This can be one of PILImageResampling
, filters. Only
has an effect if do_resize
is set to True
. bool
, optional, defaults to self.do_rescale
) —
Whether to rescale the image values between [0 - 1]. float
, optional, defaults to self.rescale_factor
) —
Rescale factor to rescale the image by if do_rescale
is set to True
. bool
, optional, defaults to self.do_grayscale
) —
Whether to convert the image to grayscale. str
or TensorType
, optional) —
The type of tensors to return. Can be one of:np.ndarray
.TensorType.TENSORFLOW
or 'tf'
: Return a batch of type tf.Tensor
.TensorType.PYTORCH
or 'pt'
: Return a batch of type torch.Tensor
.TensorType.NUMPY
or 'np'
: Return a batch of type np.ndarray
.TensorType.JAX
or 'jax'
: Return a batch of type jax.numpy.ndarray
.ChannelDimension
or str
, optional, defaults to ChannelDimension.FIRST
) —
The channel dimension format for the output image. Can be one of:"channels_first"
or ChannelDimension.FIRST
: image in (num_channels, height, width) format."channels_last"
or ChannelDimension.LAST
: image in (height, width, num_channels) format.ChannelDimension
or str
, optional) —
The channel dimension format for the input image. If unset, the channel dimension format is inferred
from the input image. Can be one of:"channels_first"
or ChannelDimension.FIRST
: image in (num_channels, height, width) format."channels_last"
or ChannelDimension.LAST
: image in (height, width, num_channels) format."none"
or ChannelDimension.NONE
: image in (height, width) format.Preprocess an image or batch of images.
( image: ndarray size: typing.Dict[str, int] data_format: typing.Union[str, transformers.image_utils.ChannelDimension, NoneType] = None input_data_format: typing.Union[str, transformers.image_utils.ChannelDimension, NoneType] = None **kwargs )
Parameters
np.ndarray
) —
Image to resize. Dict[str, int]
) —
Dictionary of the form {"height": int, "width": int}
, specifying the size of the output image. ChannelDimension
or str
, optional) —
The channel dimension format of the output image. If not provided, it will be inferred from the input
image. Can be one of:"channels_first"
or ChannelDimension.FIRST
: image in (num_channels, height, width) format."channels_last"
or ChannelDimension.LAST
: image in (height, width, num_channels) format."none"
or ChannelDimension.NONE
: image in (height, width) format.ChannelDimension
or str
, optional) —
The channel dimension format for the input image. If unset, the channel dimension format is inferred
from the input image. Can be one of:"channels_first"
or ChannelDimension.FIRST
: image in (num_channels, height, width) format."channels_last"
or ChannelDimension.LAST
: image in (height, width, num_channels) format."none"
or ChannelDimension.NONE
: image in (height, width) format.Resize an image.
( config: SuperGlueConfig )
Parameters
SuperGlue model taking images as inputs and outputting the matching of them. This model is a PyTorch torch.nn.Module subclass. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and behavior.
SuperGlue feature matching middle-end
Given two sets of keypoints and locations, we determine the correspondences by:
The correspondence ids use -1 to indicate non-matching points.
Paul-Edouard Sarlin, Daniel DeTone, Tomasz Malisiewicz, and Andrew Rabinovich. SuperGlue: Learning Feature Matching with Graph Neural Networks. In CVPR, 2020. https://arxiv.org/abs/1911.11763
( pixel_values: FloatTensor labels: typing.Optional[torch.LongTensor] = None output_attentions: typing.Optional[bool] = None output_hidden_states: typing.Optional[bool] = None return_dict: typing.Optional[bool] = None )
Parameters
torch.FloatTensor
of shape (batch_size, num_channels, height, width)
) —
Pixel values. Pixel values can be obtained using SuperGlueImageProcessor. See
SuperGlueImageProcessor.call() for details. bool
, optional) —
Whether or not to return the attentions tensors. See attentions
under returned tensors for more detail. bool
, optional) —
Whether or not to return the hidden states of all layers. See hidden_states
under returned tensors for
more detail. bool
, optional) —
Whether or not to return a ModelOutput instead of a plain tuple. from transformers import AutoImageProcessor, AutoModel import torch from PIL import Image import requests
The SuperGlueForKeypointMatching forward method, overrides the __call__
special method.
Although the recipe for forward pass needs to be defined within this function, one should call the Module
instance afterwards instead of this since the former takes care of running the pre and post processing steps while
the latter silently ignores them.