infinite-dataset-hub/MovementPatterns

MovementPatterns数据集是为了人类活动识别任务而设计的，它包含了来自加速度计和陀螺仪的时间序列数据。这些数据来源于可穿戴设备，每个记录表示不同活动的短暂动作片段。活动类别根据运动模式分为走路、跑步、坐着或跳跃等。数据集经过数据清洗、精确标记，并且每种活动都有平衡的表示。标签是人工标注的，代表人类观察者所感知的活动类别。

The MovementPatterns dataset is designed for the task of human activity recognition using data from inertial sensors. It contains time-series data from accelerometers and gyroscopes, typically found in wearable devices. Each record represents a short snippet of movement from different activities, categorized into classes like walking, running, sitting, or jumping based on observed movement patterns. The dataset is cleaned, accurately labeled, and balanced in terms of activity representation. The labels are human-annotated, representing the activity class as perceived by human observers.