# facebookresearch/VideoPose3D

Efficient 3D human pose estimation in video using 2D keypoint trajectories

Repository: https://github.com/facebookresearch/VideoPose3D
Canonical: https://ross.abutalabs.com/products/videopose3d
Language: Python
License: NOASSERTION
License Family: other
Archived: true
Last push: 2022-12-10T16:22:15+00:00

## Health v2 (maintenance only)
Score: 10/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 2876, "days_push": 1362, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, archived, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 4052, forks 785 (observed 2026-08-28T04:08:33.834197+00:00)

## What it is
A PyTorch implementation of CVPR 2019 research on 3D human pose estimation in video using temporal convolutions over 2D keypoint trajectories. It includes pretrained models for Human3.6M and HumanEva-I, training scripts, and visualization tools.

## Use cases
- estimate 3D human pose from video
- lift 2D keypoints to 3D poses
- reproduce CVPR 2019 pose estimation baselines
- train a temporal convolution model on Human3.6M
- visualize 3D skeleton predictions from video
- run semi-supervised 3D pose training

## When to choose
- you need research-grade 3D human pose estimation from 2D detections
- you want pretrained models benchmarked on Human3.6M
- you are building on temporal convolution approaches to pose lifting

## When to avoid
- you need real-time production pose estimation with active support
- you want a maintained library with recent updates
- you need pose estimation for multi-person crowded scenes out of the box

## Facets
- artifact type: library
- maturity: maintenance
- function: computer-vision, machine-learning, deep-learning
- domain: computer-vision, machine-learning, deep-learning
- platform: python
- tags: pose-estimation, 3d-pose, temporal-convolutions, pytorch, research-code, human3.6m, gpu

## Member repositories
- facebookresearch/VideoPose3D (main) score 10

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:08:33.834197+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 2026-08-29T18:23:40.098406+00:00, confidence not recorded.
  - readme: https://github.com/facebookresearch/VideoPose3D (fetched 2026-08-28T04:08:33.834197+00:00, sha 70ffe3721ad8)
- Data as of 2026-08-30T08:39:29.467469+00:00.