# verlab/accelerated_features

Implementation of XFeat (CVPR 2024). Do you need robust and fast local feature extraction? You are in the right place!

Repository: https://github.com/verlab/accelerated_features
Canonical: https://ross.abutalabs.com/products/accelerated_features
Homepage: https://www.verlab.dcc.ufmg.br/descriptors/xfeat_cvpr24
Language: Jupyter Notebook
License: Apache-2.0
License Family: permissive
Topics: descriptors, image-matching, image-registration, keypoints, local-features, real-time, lightweight
Last push: 2025-01-15T16:22:03+00:00

## Health v2 (maintenance only)
Score: 16/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 1, release rhythm 8, longevity 62
- inputs: {"age_days": 870, "days_push": 595, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1716, forks 233 (observed 2026-08-28T04:05:26.372347+00:00)

## What it is
XFeat is a lightweight, fast learned keypoint detector and descriptor for local feature extraction and image matching, supporting both sparse and semi-dense matching. It runs in real-time on CPUs without specialized hardware, making it suitable for resource-limited devices.

## Use cases
- extract keypoints and descriptors from images
- match features between two images in real time
- register images on CPU without GPU
- build visual odometry or SLAM frontends
- augmented reality tracking on resource-limited devices
- visual localization and pose estimation
- replace SIFT with a faster learned alternative

## When to choose
- you need fast, robust local feature matching on CPU or edge devices
- you want sparse or semi-dense matching with a single lightweight model
- you need real-time performance for AR, navigation, or robotics

## When to avoid
- you need state-of-the-art accuracy regardless of compute cost
- your pipeline depends on a different feature ecosystem like SuperPoint+SuperGlue with heavy GPU matching

## Facets
- artifact type: library
- maturity: active
- function: computer-vision, image-processing, machine-learning
- domain: computer-vision, image-processing, robotics
- platform: python, cross-platform
- tags: keypoint-detection, image-matching, image-registration, local-features, descriptors, real-time, lightweight, cvpr-2024, torch-hub, sparse-matching, semi-dense-matching, cpu, gpu

## Member repositories
- verlab/accelerated_features (main) score 16

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:26.372347+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 2026-08-30T03:34:16.888061+00:00, confidence not recorded.
  - readme: https://github.com/verlab/accelerated_features (fetched 2026-08-28T04:05:26.372347+00:00, sha d7520ae9342a)
  - homepage: https://www.verlab.dcc.ufmg.br/descriptors/xfeat_cvpr24 (fetched 2026-08-29T11:10:55.489376+00:00, sha a9b5b0a23db5)
  - site_page: https://www.verlab.dcc.ufmg.br/about-us (fetched 2026-08-29T11:10:55.498708+00:00, sha 6bb4b1a6b6f6)
- Data as of 2026-08-30T08:39:29.467469+00:00.