facebookresearch/silk

[ICCV 2025] Keypoint detection and description learned from image pairs only - no depth, no pose, no artificial augmentation required.

UKPGAN: A General Self-Supervised Keypoint Detector (CVPR2022)

Training library for local feature detection and matching

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

PyTorch code and models for the DINOv2 self-supervised learning method.

Code for "Efficient LoFTR: Semi-Dense Local Feature Matching with Sparse-Like Speed", CVPR 2024

A powerful yet practical General Keypoint Detection Transformer (GKDT) model, and a large-scale high-quality dataset MegaKPT that unifies 29 public datasets with over 1.3 million object instances [ECCV 2026]

Code release for CVPR'24 submission 'OmniGlue'

Repository of MatchFormer