Markerless Motion Capture System to Detect Upper Limb Movement during Rehabilitation using Video Games

This paper presents a method that analyzes the frequency and duration of upper limb movement during rehabilitation using video games. In particular, an algorithm based on markerless motion capture employs a video recorded from behind the user to segment and classify elbow flexion and extension and shoulder abduction and rotation. The proposed method was evaluated in a study involving participants during burn rehabilitation. Within the implemented protocol, participants used two video games, namely Beatsaber when using an Oculus Quest virtual reality headset and Wii Sports when using a Nintendo Wii. The results demonstrate the feasibility to employ markerless motion capture in the clinical environment. Limitations of this method were also highlighted.