shubham-goel/4D-Humans (HMR 2.0)
Source: https://github.com/shubham-goel/4d-humans
Structured fields
{
"archived": false,
"checked_at": "2026-10-17T00:00:00.000Z",
"usefulness": "HMR 2.0 transformer-based per-frame 3D human mesh recovery plus PHALP-based tracking of multiple people through video, outputting 3D pose/shape tracklets; common first step for markerless mocap.",
"source_type": "repository",
"availability": "available",
"dependencies": [
"PHALP",
"SMPL model files",
"Python 3.10 + PyTorch"
],
"next_check_at": "2026-10-31T00:00:00.000Z",
"upstream_updated_at": "2026-02-07T16:46:00.000Z"
}Evidence claims
- [upstream_claim] README: runs ViTDet and HMR2.0 on images and video; track.py builds on PHALP and accepts a video, frame folder or YouTube link; output is a rendered video and a .pkl of tracklets with 3D pose and shape. source
- [upstream_claim] Neutral SMPL model (basicModel_neutral_lbs_10_207_0_v1.0.0.pkl) must be downloaded after registering at the SMPL site; training was done on 8 A100 GPUs over 7 days. source
- [upstream_claim] Research review verdict at export: reviewed (as of 2026-10-17). Recorded by the research pipeline, not a hub review outcome; evidence.review stays unreviewed until an independent reviewer records one.
Rights
| component | license | hosting | redistribution | modification | training | Evidence and notes |
|---|---|---|---|---|---|---|
| code | MIT | permitted | permitted | permitted | unknown | Repository lists MIT. |
| other | unknown | restricted | prohibited | restricted | restricted | Requires SMPL; see smpl-body-model-license for non-commercial-only terms. |
| model_weights | unknown | unknown | unknown | unknown | unknown | HMR2.0 checkpoints auto-download to ~/.cache/4DHumans; their license was not read. |