# open-mmlab/mmhuman3d

OpenMMLab 3D Human Parametric Model Toolbox and Benchmark

Repository: https://github.com/open-mmlab/mmhuman3d
Canonical: https://ross.abutalabs.com/products/mmhuman3d
Homepage: https://mmhuman3d.readthedocs.io/
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2024-11-12T08:32:46+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": 1739, "days_push": 659, "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 1426, forks 156 (observed 2026-08-28T04:04:41.662537+00:00)

## What it is
MMHuman3D is an open-source PyTorch-based toolbox and benchmark for 3D human parametric models (e.g., SMPL, SMPL-X) in computer vision and graphics, part of the OpenMMLab project. It provides modular reimplementations of state-of-the-art methods, a unified HumanData data convention, and versatile rendering/visualization tools.

## Use cases
- reproduce state-of-the-art 3D human pose and mesh recovery methods
- train and evaluate models on human pose datasets with a unified format
- render SMPL/SMPL-X body models with segmentation, depth, and keypoints
- convert and preprocess human motion capture datasets
- run inference demos including webcam real-time estimation

## When to choose
- you need a modular PyTorch framework for 3D human parametric model research
- you want to benchmark multiple HMR methods under one codebase
- you need unified handling of many human pose datasets via HumanData
- you need differentiable or conventional rendering of human body models

## When to avoid
- you need a production-ready human pose estimation service rather than a research toolbox
- you work outside PyTorch or need multi-view mocap (see XRMoCap instead)
- you need actively developed features, as the project appears to be in maintenance mode

## Facets
- artifact type: framework
- maturity: maintenance
- function: machine-learning, deep-learning, computer-vision, image-processing, data-visualization, benchmarking
- domain: computer-vision, machine-learning, deep-learning, graphics, simulation
- platform: python
- tags: 3d-human-pose, smpl, parametric-model, human-mesh-recovery, pytorch, openmmlab, motion-capture, linux, gpu

## Member repositories
- open-mmlab/mmhuman3d (main) score 23

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:41.662537+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 2026-08-30T04:37:27.986105+00:00, confidence not recorded.
  - readme: https://github.com/open-mmlab/mmhuman3d (fetched 2026-08-28T04:04:41.662537+00:00, sha f05a6784b4b9)
  - registry_pypi: https://pypi.org/pypi/mmhuman3d/json (fetched 2026-08-29T11:49:45.029963+00:00, sha 5f309c4461a1)
- Data as of 2026-08-30T08:39:29.467469+00:00.