cvg/LightGlue

LightGlue: Local Feature Matching at Light Speed (ICCV 2023)

57
/ 100
Established

LightGlue helps computer vision engineers quickly and accurately find corresponding points across two different images. You provide two images along with their extracted "keypoints" (distinctive points) and "descriptors" (mathematical descriptions of those keypoints). The output is a list of which keypoints in the first image match keypoints in the second image. This is ideal for those building systems that need to understand how different views of a scene relate to each other.

4,412 stars.

Use this if you need to rapidly and precisely identify matching features between image pairs for applications like 3D reconstruction, augmented reality, or object tracking.

Not ideal if your application doesn't require finding sparse, local feature correspondences, or if you're not already working with extracted keypoints and descriptors.

computer-vision image-matching 3d-reconstruction augmented-reality robotics-navigation
No Package No Dependents
Maintenance 10 / 25
Adoption 10 / 25
Maturity 16 / 25
Community 21 / 25

How are scores calculated?

Stars

4,412

Forks

515

Language

Python

License

Apache-2.0

Last pushed

Feb 18, 2026

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/computer-vision/cvg/LightGlue"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.