# mkocabas/VIBE

Official implementation of CVPR2020 paper "VIBE: Video Inference for Human Body Pose and Shape Estimation"

Repository: https://github.com/mkocabas/VIBE
Canonical: https://ross.abutalabs.com/products/mkocabas-vibe
Homepage: https://arxiv.org/abs/1912.05656
Language: Python
License: NOASSERTION
License Family: other
Topics: video-pose-estimation, 3d-pose-estimation, human-pose-estimation, smpl, pytorch, 3d-human-pose, cvpr2020, cvpr-2020, cvpr20, cvpr
Last push: 2023-03-24T23:39:57+00:00

## Health v2 (maintenance only)
Score: 23/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 8, longevity 100
- inputs: {"age_days": 2457, "days_push": 1258, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 3211, forks 579 (observed 2026-08-28T04:07:49.353450+00:00)

## What it is
Official PyTorch implementation of VIBE (CVPR 2020), a video-based method for 3D human body pose and shape estimation that predicts SMPL body model parameters per frame. It supports arbitrary videos with multiple people, GPU/CPU inference, training code, and FBX/glTF export for graphics software.

## Use cases
- estimate 3d human pose and shape from video
- extract smpl body model parameters from video frames
- reconstruct human motion sequences from rgb video
- convert video motion capture to fbx or gltf for blender or unity
- train a temporal 3d pose estimation model from scratch
- run multi-person 3d pose estimation on arbitrary videos

## When to choose
- you need video-based 3d human pose/shape estimation with SMPL output
- you want a research-grade reference implementation with pretrained models and training code
- you need motion data exportable to graphics tools via FBX/glTF

## When to avoid
- you need real-time pose estimation on low-end hardware without a GPU
- you need actively maintained code with recent dependency support
- you only need 2D keypoint detection rather than 3D body model fitting

## Facets
- artifact type: library
- maturity: maintenance
- function: computer-vision, machine-learning, video-processing, deep-learning
- domain: computer-vision, deep-learning, machine-learning
- platform: python
- tags: pose-estimation, smpl, 3d-human-pose, motion-capture, pytorch, cvpr-2020, research-code, video, linux, gpu

## Member repositories
- mkocabas/VIBE (main) score 23

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:07:49.353450+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 2026-08-30T07:24:43.649724+00:00, confidence not recorded.
  - readme: https://github.com/mkocabas/VIBE (fetched 2026-08-28T04:07:49.353450+00:00, sha a63119aeaf76)
  - homepage: https://arxiv.org/abs/1912.05656 (fetched 2026-08-29T09:38:21.622504+00:00, sha ce5b8d017597)
  - site_page: https://info.arxiv.org/about/donate.html (fetched 2026-08-29T09:38:21.631671+00:00, sha cca9c3a11c56)
  - site_page: https://info.arxiv.org/about/ourmembers.html (fetched 2026-08-29T09:38:21.635093+00:00, sha 47cbc55ff1de)
  - site_page: https://info.arxiv.org/about (fetched 2026-08-29T09:38:21.637025+00:00, sha a1f16f915a9a)
  - site_page: https://info.arxiv.org/labs/index.html (fetched 2026-08-29T09:38:21.633441+00:00, sha b14a8d05a0ec)
- Data as of 2026-08-30T08:39:29.467469+00:00.