HOH Handover Dataset
收藏资源简介:
HOH手递手数据集是一个包含136个物体的大规模数据集,用于加速基于数据的手递手研究、人机手递手实施和人工智能在手递手参数估计方面的研究。该数据集包含多视角RGB和深度数据、骨骼、融合点云、抓取类型和手性标签、物体、给予者和接收者的手部2D和3D分割、给予者和接收者的舒适度评级,以及配对的物体元数据和对齐的3D模型,涵盖了2,720次手递手交互,涉及136个物体和20对给予者-接收者,其中40对包括角色反转,由40名参与者组织。
The HOH Handover Dataset is a large-scale dataset containing 136 objects, designed to accelerate data-driven handover research, human-robot handover implementation, and artificial intelligence research on handover parameter estimation. This dataset includes multi-view RGB and depth data, skeleton data, fused point clouds, grasp type and handedness labels, 2D and 3D segmentations of objects as well as the hands of the giver and receiver, comfort ratings for both the giver and receiver, paired object metadata, and aligned 3D models. It covers 2,720 handover interactions involving 136 objects and 20 giver-receiver pairs, among which 40 pairs include role reversal, and was organized by 40 participants.

- 1HOH: Markerless Multimodal Human-Object-Human Handover Dataset with Large Object Count · 2024年



