Complete Inertial Pose Dataset

Complete Inertial Pose Dataset 由微机电系统中心(CMEMS)和米尼奥大学联合创建，包含两个子数据集，分别使用低成本和高精度磁力计、角速度计和重力(MARG)传感器采集。数据集总计约350万样本，涵盖从原始测量到完整人体姿态的全过程，包括传感器到节段校准、多传感器融合和骨骼运动学等。通过21名和10名参与者执行6种不同类型的动作序列来收集数据，动作涵盖从校准到日常活动和随机运动，具有高度的变异性和复杂的动态特性。数据集旨在解决穿戴技术在姿态监测中的应用问题，特别是在低成本系统中的挑战，可用于评估、基准测试和开发新的算法，适用于经典或数据驱动的惯性姿态估计算法、人类运动理解和预测以及工业或康复环境中的工效学评估。

The Complete Inertial Pose Dataset was jointly developed by the Center for Micro-Electro-Mechanical Systems (CMEMS) and the University of Minho. It comprises two sub-datasets, which were collected using low-cost and high-precision magnetometer, gyroscope and gravity (MARG) sensors respectively. The entire dataset totals approximately 3.5 million samples, covering the full pipeline from raw sensor measurements to complete human body pose, encompassing sensor-to-segment calibration, multi-sensor fusion, and skeletal kinematics. Data was collected from 21 and 10 participants who executed 6 distinct types of motion sequences. The covered motion scenarios span calibration tasks, daily activities and random movements, featuring high variability and complex dynamic characteristics. This dataset aims to address the application challenges of wearable technology in pose monitoring, especially for low-cost systems. It can be used for the evaluation, benchmarking and development of new algorithms, applicable to classical or data-driven inertial pose estimation algorithms, human motion understanding and prediction, as well as ergonomic assessments in industrial or rehabilitation environments.