EricGuo5513/HumanML3D
HumanML3D: A large and diverse 3d human motion-language dataset.
This dataset offers a rich collection of 3D human motion clips paired with natural language descriptions, covering a wide array of activities from daily tasks to sports and acrobatics. It takes existing 3D motion data and descriptive text, then processes and links them to create a large dataset of motion-text pairs. Researchers and developers working on AI models for human animation, virtual reality, or human-computer interaction will find this useful for training and evaluating their systems.
1,408 stars. No commits in the last 6 months.
Use this if you need a diverse and extensive dataset to train AI models that generate realistic 3D human motion from text descriptions, or vice versa.
Not ideal if you are looking for a dataset with highly specific, niche human movements or require raw sensor data rather than processed 3D joint positions.
Stars
1,408
Forks
131
Language
Python
License
MIT
Category
Last pushed
Aug 18, 2024
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/EricGuo5513/HumanML3D"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
talmolab/sleap
A deep learning framework for multi-animal pose tracking.
kennymckormick/pyskl
A toolbox for skeleton-based action recognition.
open-mmlab/mmaction2
OpenMMLab's Next Generation Video Understanding Toolbox and Benchmark
jgraving/DeepPoseKit
a toolkit for pose estimation using deep learning
DenisTome/Lifting-from-the-Deep-release
Implementation of "Lifting from the Deep: Convolutional 3D Pose Estimation from a Single Image"