DynTherapy: Physical Therapy Exercises Dataset

Published: 4 July 2024| Version 1 | DOI: 10.17632/hghdm99rwg.1
Mo'men Abdallah,


This dataset contains 33 key points collected using MediaPipe's pose estimation AI model during the performance of various physical therapy exercises. The dataset includes a diverse range of exercises, each segmented into key positions such as Start and End to capture essential phases for classification. Each exercise was performed multiple times, and the key points, representing joints such as the shoulder, elbow, and wrist, were recorded under various camera settings and lighting conditions to provide a comprehensive representation of poses. Classes: knee_raises_down knee_raises_up shoulder_press_up shoulder_press_down shoulder_flexion_up lateral_raises_down lateral_raises_up lateral_leg_raises_up lateral_leg_raises_down glute_bridge_down glute_bridge_up leg_raises_up leg_raises_down



Jordan University of Science and Technology


Physical Therapy