← all datasets

1M-HUGs

Emerging
1papers using it
2026first seen

The '1M-HUGs' dataset is an egocentric collection of human grasps comprising 1 million frames and 6,707 object instances, used to evaluate the performance of a flow-matching model for generating diverse human-like grasps for robotic applications.

Papers using 1M-HUGs (1)

1M-HUGs β€” datasets β€” robotics