Body- and Pose-tracking for Augmented Reality Applications in the Field of Sports

This paper presents the concept for a system that utilizes a single RGB camera to track and evaluate a human body. The system aims to support office workers in improving their posture, by doing simple exercises. This should reduce the risk of health issues like back, shoulder and neck pain which will lead to an increase in the overall productivity. By creating an augmented reality mirror this system will give instructions for doing simple exercises. During this workout the user will be evaluated and the user will be notified about wrong stances. This paper discusses related works on body tracking systems, including systems that utilize a single camera as well as those systems that utilize RGBD cameras. It also presents a basic system design and ideas for pose comparison. Additionally a prototype that is based on the MediaPipe pose library for JavaScript has been created, which will be discussed and evaluated in the course of this paper. This paper concludes by highlighting possible future developments, such as additional features and improved usability. This whole work demonstrates the feasibility of using body tracking for sports applications.