# MVIG-SJTU/AlphaPose

Real-Time and Accurate Full-Body Multi-Person Pose Estimation&Tracking System

Repository: https://github.com/MVIG-SJTU/AlphaPose
Canonical: https://ross.abutalabs.com/products/alphapose
Homepage: http://mvig.org/research/alphapose.html
Language: Python
License: NOASSERTION
License Family: other
Topics: pose-estimation, posetracking, tracking, gpu, pytorch, realtime, human-pose-estimation, human-tracking, human-pose-tracking, alpha-pose, alphapose, person-pose-estimation, accurate, crowdpose, full-body, whole-body, skeleton, keypoints, human-computer-interaction, human-joints
Last push: 2024-05-13T14:33:34+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 3151, "days_push": 842, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 8596, forks 2027 (observed 2026-08-28T04:10:23.745624+00:00)

## What it is
AlphaPose is an open-source real-time multi-person full-body pose estimation and tracking system built on PyTorch. It detects human keypoints (including face, hand, and foot) in images and videos and tracks poses across frames with its PoseFlow online tracker.

## Use cases
- estimate human pose keypoints in images and videos
- track multiple people's skeletons across video frames in real time
- detect whole-body keypoints including face, hands, and feet
- analyze human motion for fitness or sports applications
- build gesture or human-computer interaction systems
- run pose estimation on crowded scenes with occlusion

## When to choose
- you need accurate multi-person 2D pose estimation with pretrained models
- you need real-time pose tracking in video on GPU hardware
- you need whole-body keypoints (body, face, hands, feet) in one system

## When to avoid
- you need 3D pose estimation as the primary feature (only limited HybrIK support)
- you need a CPU-only lightweight solution
- you need a maintained library with frequent updates and broad community support

## Facets
- artifact type: library
- maturity: maintenance
- function: computer-vision, image-processing, video-processing, machine-learning, deep-learning
- domain: computer-vision, image-processing, machine-learning, artificial-intelligence
- platform: windows, python
- tags: pose-estimation, pose-tracking, keypoint-detection, human-pose, multi-person, pytorch, realtime, whole-body, skeleton-tracking, poseflow, video, linux, gpu

## Member repositories
- MVIG-SJTU/AlphaPose (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:10:23.745624+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 2026-08-29T17:26:08.531710+00:00, confidence not recorded.
  - readme: https://github.com/MVIG-SJTU/AlphaPose (fetched 2026-08-28T04:10:23.745624+00:00, sha 998ab86d54c4)
  - registry_pypi: https://pypi.org/pypi/alphapose/json (fetched 2026-08-29T08:26:28.547722+00:00, sha 1d0018db08b6)
- Data as of 2026-08-30T08:39:29.467469+00:00.