Pose Estimation and Joint Angle Detection Using Mediapipe Machine Learning Solution

Health is one of the central aspects of life and innovative ways for its improvement are constantly being studied. Artificial intelligence has an extensive application and its contribution to health and medicine is widely recognized. In this paper, the application of machine learning algorithms in the field of health care is presented. A model for physical activity injury prevention based on the MediaPipe solution for body pose tracking has been developed. The solutions for pose estimation and detection of joint angles and angles relative to the horizontal are integrated into a comprehensive system that detects all key body landmarks and angles during the movement of the observed person. In addition, one of the goals of this research is to develop a flexible system with the ability to process a variety of inputs in terms of video content and format. The system is trained and tested on video inputs and can process front, left, and right perspectives. In the processing phase, a graph of posture and angle estimation is generated. The graph represents detected joints and the corresponding angles that vary depending on the observed perspective. The input is integrated with the graph and thus provides valuable information about body posture and alignment. The results provide support to professionals in physical activity monitoring and injury prevention.