zju3dv/EfficientLoFTR

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

1042 stars · 112 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

Efficient LoFTR is a PyTorch implementation of a semi-dense local feature matching model that matches keypoints between image pairs with sparse-like speed, published at CVPR 2024. It provides inference and training code, pretrained indoor/outdoor models, and is also available via Hugging Face Transformers.

Use cases

• match keypoints between two images
• estimate camera pose from image pairs
• feature matching for 3D reconstruction
• image registration for UAV localization
• feature matching for autonomous driving
• run semi-dense matching in real time on GPU

When to choose

• you need fast semi-dense feature matching between image pairs
• you want pretrained indoor/outdoor matching models with training code
• you need a matcher for SfM, SLAM, or visual localization pipelines

When to avoid

• you need CPU-only real-time matching
• you need a simple classical detector like SIFT/ORB without deep learning dependencies

Facets

library · maturity active

machine-learning deep-learning computer-vision image-processing computer-vision machine-learning deep-learning robotics python cross-platform feature-matching loftr cvpr-2024 pose-estimation image-registration 3d-reconstruction pytorch gpu linux

Member repositories

For agents

markdown · JSON · MCP: product_card(name="zju3dv/EfficientLoFTR")