HO-Cap
收藏资源简介:
HO-Cap数据集由德克萨斯大学达拉斯分校创建,专注于研究手与物体交互的3D重建和姿态跟踪。该数据集包含70个视频,总计699,000帧的RGB-D数据,涉及9名参与者和64种物体。数据集的创建过程采用了多视角RGB-D摄像机和HoloLens头戴设备,通过半自动方法进行标注,显著减少了人工标注时间。HO-Cap数据集的应用领域包括人机交互、虚拟现实/增强现实以及机器人学习,旨在解决手与物体交互的识别问题,为相关研究提供丰富的实验数据。
The HO-Cap dataset, created by The University of Texas at Dallas, focuses on 3D reconstruction and pose tracking for hand-object interactions. It contains 70 videos totaling 699,000 RGB-D frames, involving 9 participants and 64 types of objects. The dataset was constructed using multi-view RGB-D cameras and Microsoft HoloLens head-mounted devices, with semi-automatic annotation methods that significantly reduce manual annotation time. The HO-Cap dataset has applications in human-computer interaction, virtual reality/augmented reality, and robotic learning, aiming to address the recognition of hand-object interactions and provide abundant experimental data for relevant research.




