Dataset · OriginFlow-AI
Hand-Object Interaction: Video, Poses, and Force
OriginFlow-AI/origindata-preview-sample
A robotics multimodal dataset featuring hand motion recordings and applied force in real-world tasks.
What's inside
A preview sample of 7,717 episodes (100.62 hours) covering 1,170 real-world tasks across 28 domains. It includes stereo RGB video, time-series hand pose and pressure force data, and semantic annotations calibrated for embodied AI. Data is provided in the LeRobot v3.0 format.
What you can do with it
- Train multimodal models for robotic arm and manipulator control
- Simulate and predict physical hand-object interactions
- Tasks
- Robotics
- Size
- 10M–100M rows
- Downloads this month
- 6.4K
- Likes
- 25
- Access
- on request on Hugging Face
Models trained on it
How to load it
from datasets import load_dataset
ds = load_dataset("OriginFlow-AI/origindata-preview-sample")