magicleap/SuperGluePretrainedNetwork
A Graph Neural Network for matching sparse image features between image pairs, built with PyTorch.
Not currently ranked — collecting fresh signals.
star history
SuperGlue is a deep learning model that performs feature matching across image pairs using a Graph Neural Network combined with an Optimal Matching layer. It operates on SuperPoint keypoints and descriptors to identify correspondences between two images. The repository includes pretrained weights for both indoor (ScanNet) and outdoor (MegaDepth) environments, with the model serving as a middle-end component for context aggregation and match filtering in visual localization pipelines.
Frequently asked
- What is magicleap/SuperGluePretrainedNetwork?
- A Graph Neural Network for matching sparse image features between image pairs, built with PyTorch.
- Is SuperGluePretrainedNetwork open source?
- Yes — magicleap/SuperGluePretrainedNetwork is an open-source project tracked on heatdrop.
- What language is SuperGluePretrainedNetwork written in?
- magicleap/SuperGluePretrainedNetwork is primarily written in Python.
- How popular is SuperGluePretrainedNetwork?
- magicleap/SuperGluePretrainedNetwork has 4.1k stars on GitHub.
- Where can I find SuperGluePretrainedNetwork?
- magicleap/SuperGluePretrainedNetwork is on GitHub at https://github.com/magicleap/SuperGluePretrainedNetwork.