josefheidler/har_adults_2021-harth

HARTH数据集是一个包含31名健康成年人右大腿和腰部双传感器加速度计数据的集合，用于识别12种活动类型，如行走、跑步、爬楼梯、骑自行车等。数据以50Hz的频率采集，总样本数为8.3M，标注覆盖率为99.9%。该数据集旨在为大规模人群健康研究提供标注训练资源。

HARTH dataset is a collection of dual-sensor accelerometry data (right thigh and lower back) from 31 healthy adults, designed for recognizing 12 activity types including walking, running, stair climbing, cycling, and more. Data is sampled at 50 Hz, totaling 8.3 M samples with a labeling coverage of 99.9%. The dataset serves as a labeled training resource for large-scale population health research.