Mediapipe Pose AI Motion Capture Software for Upper Limb Exercise and Rehabilitative Therapy

Artificial intelligence (AI) based motion capture has revolutionized the field of biomechanics and locomotion analysis by enabling more widespread adoption of the technique. In this realm, the Google based Mediapipe Pose AI motion capture software is a robust platform for close-to-real-time motion capture of many body landmarks stretching from the upper limbs to lower limbs. This preprint reports an attempt to code an in-house Mediapipe Pose based Python motion capture platform useful for upper limb exercise performance analysis, and rehabilitative therapy of upper limbs with dysfunctional muscle control and movement such as in mild stroke patients. Specifically, the software is capable of tracking real-time position and movement of the elbow, wrist, and shoulder joints, and can calculate both the shoulder and elbow joint angle evolution, and angular velocity changes. More importantly, such data are chronicled in both graphs and a frame-by-frame catalogue of joint angle and angular velocity changes, that altogether, serves as useful data for personal evaluation of exercise performance, as well as physiotherapy post-rehab assessment of treatment progress. The software is capable of tracking the full range of motions of the shoulder-elbow movement system, and can be used for a variety of exercise performance tests, as well as for diagnosing and tracking upper limb movement disorders in mild stroke and musculoskeletal dysfunction patients.