CZU-MHAD
收藏资源简介:
CZU-MHAD是由江苏数字孪生技术工程研究中心创建的综合性多模态人类动作数据集,包含22种动作和三种模态的时间同步数据。数据集通过Kinect v2相机收集深度视频和骨骼位置,以及通过10个穿戴式传感器收集惯性信号,旨在更准确地描述人体主要运动关节的信息。创建过程中,数据集通过Raspberry PI与MPU9250传感器交互,实现数据的读取、保存和上传。该数据集适用于研究人体动作识别、姿态估计、疾病诊断等领域,特别在动作协调性研究中具有重要价值。
CZU-MHAD is a comprehensive multimodal human action dataset created by the Jiangsu Engineering Research Center for Digital Twin Technology. It contains time-synchronized data of 22 actions across three modalities. The dataset collects depth videos and skeleton positions via a Kinect v2 camera, as well as inertial signals through 10 wearable sensors, aiming to more accurately describe the kinematic information of major human joints. During its creation, the dataset interacted with MPU9250 sensors via a Raspberry PI to realize data reading, storage and uploading. This dataset is applicable to research fields such as human action recognition, pose estimation, disease diagnosis and other related areas, and is particularly valuable in the study of motor coordination.




