AddBiomechanics Dataset 1.0

该数据集名为AddBiomechanics，包含了来自273名参与者的超过2400万帧动作数据，涵盖了70多小时的运动场景。这些数据标准化了肌肉骨骼模型以及位置和物理信息，其中包括光学标记位置、地面反作用力矩和力的测量、估计的关节运动学以及估计的关节扭矩。数据集涵盖了九种不同的活动类型，并采用标准化流程进行处理。此外，该数据集还包含了关于受试者的人口统计学信息。其规模之大，涵盖了超过2400万帧来自273名参与者的数据，任务旨在从动作中估计人体动态。

This dataset, named AddBiomechanics, contains over 24 million motion frames from 273 participants, spanning more than 70 hours of recorded movement. The data standardizes musculoskeletal models as well as positional and physical information, including optical marker positions, measurements of ground reaction forces and moments, estimated joint kinematics, and estimated joint torques. The dataset covers nine distinct activity types and was processed using standardized workflows. Additionally, it includes demographic information about the study subjects. Given its considerable scale—encompassing over 24 million motion frames from 273 participants—this dataset is developed for the task of estimating human dynamics from motion data.

搜集汇总

背景与挑战

背景概述

AddBiomechanics Dataset 1.0是一个高质量的人类运动数据集，包含273名参与者的70+小时运动数据，提供详细的物理和运动学信息，适用于大规模机器学习研究。数据集还包括标准化肌肉骨骼模型和多种活动类型的数据，覆盖广泛的年龄和性别分布。