cvg/Hierarchical-Localization

Health v2 · maintenance only

Adoption not part of the score

4194 stars · 768 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

hloc is a modular Python toolbox for state-of-the-art 6-DoF visual localization, combining image retrieval and feature matching (SuperPoint, SuperGlue, LightGlue, DISK) with Structure-from-Motion. It lets researchers and engineers build 3D maps, localize query images, and benchmark custom features on indoor and outdoor datasets.

Use cases

• estimate camera pose from a single image against a 3D map
• run structure-from-motion with SuperPoint and SuperGlue on my own photos
• benchmark my custom local features for visual localization
• localize images captured by a robot or AR device
• reproduce state-of-the-art results on visual localization benchmarks
• build a 3D reconstruction from an image collection

When to choose

• you need accurate 6-DoF camera pose estimation with deep-learned features
• you want a modular pipeline to swap in different feature extractors and matchers
• you need SfM reconstruction plus localization in one toolbox

When to avoid

• you need real-time SLAM rather than offline localization against a reference model
• you want a turnkey end-user application rather than a research toolbox
• your project has no GPU and cannot use COLMAP dependencies

Facets

library · maturity active

computer-vision image-processing machine-learning deep-learning computer-vision machine-learning robotics python windows visual-localization feature-matching structure-from-motion pose-estimation superpoint superglue lightglue image-retrieval 6dof colmap algorithms gpu docker linux macos

Member repositories

For agents

markdown · JSON · MCP: product_card(name="cvg/Hierarchical-Localization")