A Triaxial Accelerometry and Physical Activity Intensity Dataset for Chinese Primary and Secondary School Students

This study presents a comprehensive dataset of triaxial accelerometer data and physical activity intensity (metabolic equivalents, METs) for diverse physical activities among 125 Chinese primary and secondary school students. The dataset incorporates 13 representative physical activities (PA): slow walking, shooting baskets, brisk walking, and radio calisthenics, soccer passing drills, badminton, soccer slalom, basketball slalom, and jogging, jump rope, single-leg hops on both legs, and sprinting. Of 14,625 theoretically expected accelerometer records from 125 participants wearing nine sensors across 13 activities, 2,912 records (19.9%) were missing, with concentrations in the lower limbs (38% from the right mid-lateral aspect of the femur, 17% from the left distal lateral aspect of the fibula, and 16% from the right distal lateral aspect of the fibula).The collected data comprised two synchronized modalities: first, raw inertial signals (including triaxial acceleration with a dynamic range of ± 16 g, angular velocity with a full-scale range of ± 2000 deg/s, Euler angles, and magnetic field intensity) sampled at 100 Hz from nine body locations; second, time-aligned metabolic intensity labels derived from synchronized breath-by-breath indirect calorimetry. Combined with detailed PA and grade-level labels, this dataset aims to support the development of action recognition algorithms for minors, cross-site sensor fusion research, exercise intensity estimation models, and studies on stage-specific movement variations.