Related Experiment Video
Updated: Aug 6, 2026
06:51
Physical Activity Measurement in Children Accepting Table Tennis Training
Published on: July 27, 2022
A Triaxial Accelerometry and Physical Activity Intensity Dataset for Chinese Primary and Secondary School Students
Chunyu Zhao1, Lei Jiang2, Khoa Anh Nguyen Le1,3
1Institute of Artificial Intelligence in Sports, Capital University of Physical Education and Sports, Beijing, China.
Scientific Data
|July 20, 2026
Summary
This study created a dataset of physical activity (PA) intensity and accelerometer data for 125 Chinese students across 13 activities. This data supports developing algorithms for minors' action recognition and exercise intensity estimation.
Area of Science:
- Biomedical Engineering
- Sports Science
- Data Science
Background:
- Accurate physical activity (PA) intensity measurement is crucial for understanding children's health.
- Existing datasets often lack comprehensive data covering diverse activities and synchronized physiological measures.
- Developing reliable algorithms for PA recognition in minors requires robust, multi-modal data.
Purpose of the Study:
- To present a comprehensive dataset of triaxial accelerometer data and physical activity intensity (METs) for diverse PA among Chinese primary and secondary school students.
- To provide synchronized inertial sensor data and indirect calorimetry-derived metabolic intensity labels.
- To facilitate research in action recognition, sensor fusion, and exercise intensity estimation for minors.
Main Methods:
- Collected triaxial accelerometer and inertial sensor data (acceleration, angular velocity, Euler angles, magnetic field) at 100 Hz from nine body locations on 125 students.
- Recorded 13 distinct physical activities, including walking, running, and sports drills.
- Synchronized sensor data with breath-by-breath indirect calorimetry to determine metabolic intensity (METs).
Main Results:
- A dataset comprising 14,625 theoretically expected records, with 2,912 (19.9%) missing, primarily from lower limb sensors.
- Rich data including raw inertial signals and validated metabolic intensity labels.
- Detailed labels for PA type and grade level.
Conclusions:
- The dataset provides a valuable resource for developing and validating action recognition algorithms for minors.
- It supports cross-site sensor fusion research and the creation of accurate exercise intensity estimation models.
- The data enables studies on age-specific movement variations in physical activity intensity.
