Physical Action Dataset

Thousands of people around the globe are limited in their daily activities due to physical or mental disability, which hinders in their daily tasks. All such individuals can get a boost in their lives by introduction of computer aided assistive applications that help them by classifying the muscle movements. Designing of such systems gets further uplifts by availability of non-invasive sensors, like Surface Electromyography (sEMG), which can be mapped onto the routine activities using machine learning. In this dataset, sEMG signature against routine activities such as typing, resting, lifting and exercise (push up) has been acquired. The sEMG data is coupled with information from a 9-DoF inertial measurement unit (IMU) in the form of accelerometer, gyro and derived orientation signals. Fusion of sEMG with IMU data can lead us to get better classification results especially while segregating abnormal data from normal or routine actions. This data can be of particular value to the researchers working on computer-aided systems for subjects with physical or mental disabilities.全球范围内，数以千计的人群因生理或心理残疾，日常活动受到限制，日常任务的完成也因此受阻。此类人群可通过引入可对肌肉运动进行分类的计算机辅助助力应用，改善自身生活状况。此类系统的设计可因非侵入式传感器的普及而进一步优化，例如表面肌电信号（Surface Electromyography, sEMG），这类传感器可通过机器学习方法与日常活动建立关联。本数据集采集了打字、静息、抬举及运动（俯卧撑）等日常活动对应的sEMG信号特征。该sEMG数据还耦合了来自9自由度惯性测量单元（inertial measurement unit, IMU）的信息，包括加速度计、陀螺仪及派生姿态信号。将sEMG与IMU数据融合，可获得更优异的分类效果，尤其在区分异常数据与正常或日常动作数据时效果更为突出。该数据集对于致力于为生理或心理残障人士研发计算机辅助系统的研究人员而言具有重要价值。

搜集汇总

背景与挑战

背景概述

该数据集专注于采集日常物理活动（如打字、休息、举重和俯卧撑）的表面肌电图（sEMG）信号，并结合9自由度惯性测量单元（IMU）的传感器数据，以融合多模态信息。其目标是支持计算机辅助应用开发，通过机器学习分类肌肉运动，帮助改善身体或精神残疾人士的日常活动能力，并提升异常动作与常规动作的区分精度。