# magicleap/SuperGluePretrainedNetwork

SuperGlue: Learning Feature Matching with Graph Neural Networks (CVPR 2020, Oral)

Repository: https://github.com/magicleap/SuperGluePretrainedNetwork
Canonical: https://ross.abutalabs.com/products/supergluepretrainednetwork
Language: Python
License: NOASSERTION
License Family: other
Topics: deep-learning, feature-matching, pose-estimation, graph-neural-networks
Last push: 2024-08-30T23:50:47+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 2360, "days_push": 733, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 4072, forks 758 (observed 2026-08-28T04:08:34.227045+00:00)

## What it is
SuperGlue is a PyTorch implementation of a graph neural network with an optimal matching layer that matches sparse image features between two images, released with pretrained indoor and outdoor weights. It runs on top of SuperPoint keypoints and descriptors and includes demo and evaluation scripts for image pairs and live video streams.

## Use cases
- match features between two images
- estimate relative pose from image pairs
- run feature matching on webcam or video streams
- build visual localization or SfM pipelines
- match SuperPoint keypoints with a learned matcher

## When to choose
- you need state-of-the-art sparse feature matching for localization, SfM, or pose estimation
- you want pretrained indoor and outdoor matching models ready to run in PyTorch

## When to avoid
- you need dense pixel-level matching rather than sparse keypoint matching
- you cannot use a GPU or PyTorch in your environment
- you need a permissively licensed library for commercial products (license is non-standard)

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, computer-vision, image-processing
- domain: computer-vision, deep-learning, machine-learning, artificial-intelligence
- platform: python, cross-platform
- tags: feature-matching, graph-neural-networks, superpoint, pose-estimation, pytorch, pretrained-weights, visual-localization, slam, gpu

## Member repositories
- magicleap/SuperGluePretrainedNetwork (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:08:34.227045+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T18:23:29.953001+00:00, confidence not recorded.
  - readme: https://github.com/magicleap/SuperGluePretrainedNetwork (fetched 2026-08-28T04:08:34.227045+00:00, sha 7f86c6df3f13)
- Data as of 2026-08-30T08:39:29.467469+00:00.