GRAB
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/otaheri/grab
下载链接
链接失效反馈官方服务：
资源简介：
该数据集记录了全身人体抓取小物体的场景，同时我们也研究了该数据集在跨数据集泛化能力方面的表现。该数据集包含1,334个视频，旨在进行三维人体与物体交互预测的任务。
This dataset records scenarios of full-body human grasping of small objects, and we also investigated its performance in cross-dataset generalization. Comprising 1,334 videos, it is designed for the task of 3D human-object interaction prediction.
