EricGuo5513/HumanML3D resource

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

1522 stars · 143 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

HumanML3D is a large 3D human motion-language dataset with 14,616 motion clips and 44,970 text descriptions, built from HumanAct12 and AMASS. It includes processing scripts, data augmentation via mirroring, and the related KIT-ML dataset.

Use cases

• train text-to-motion generation models
• research on 3d human motion synthesis from text
• benchmark motion-language models
• get annotated motion capture data for deep learning
• build text-conditioned animation systems

When to choose

• you need a large annotated 3D human motion dataset for motion generation research
• you want a standard benchmark for text-to-motion tasks
• you need text descriptions paired with motion clips

When to avoid

• you need raw AMASS data distributed directly (licensing requires regeneration via scripts)
• your project is unrelated to human motion or multimodal learning

Facets

dataset · maturity stable

machine-learning deep-learning data-generation machine-learning computer-vision python cross-platform motion-capture text-to-motion human-motion multimodal motion-generation natural-language-processing datasets

Member repositories

For agents

markdown · JSON · MCP: product_card(name="EricGuo5513/HumanML3D")