AscendMotion
收藏资源简介:
AscendMotion数据集是由厦门大学等多个机构共同收集的一个大规模、标注详尽的攀岩运动多模态数据集。该数据集包含22名熟练攀岩者在12种不同岩壁上的运动,共有344分钟标注数据和441分钟未标注数据,总计412000帧,涵盖RGB图像、LiDAR点云和IMU动作捕捉系统同步记录的数据。数据集旨在为攀岩运动的人类网格恢复研究提供挑战,特别是在全球坐标系中的运动捕捉。
AscendMotion is a large-scale, comprehensively annotated multi-modal climbing motion dataset jointly collected by Xiamen University and multiple other institutions. This dataset contains motions from 22 skilled climbers on 12 distinct rock walls, with 344 minutes of annotated data and 441 minutes of unannotated data, totaling 412,000 frames. It covers synchronously recorded data including RGB images, LiDAR point clouds, and data from IMU-based motion capture systems. The dataset is designed to pose challenges for research on human mesh recovery in climbing motions, particularly motion capture within the global coordinate system.




