A Support Training System for Table Tennis Using OpenCV and MediaPipe

Monitoring and supervising the training process in table tennis plays an important role, not just in reducing the risk of common injuries in shoulder, elbow, and knee joints, but also in developing and maintaining proper forms and movements, particularly among beginner and amateur players. This study aims to design such systems, employing a dual-camera setup positioned beside the players’ dominant hand and behind their backs, each situated 3.5 m away from the midpoint of the table’s baseline. The videos collected are processed through the system, and with the application of OpenCV and MediaPipe, the required angles are displayed on the screen for the “ready” and “finish” positions. The precision of these parameters has been calibrated, demonstrating R-squared scores ranging from 93% to almost 96%. The experimental process involved five individuals performing forehand strokes at varying angle values, yielding promising results with an error range of 1 to 5 \(^\circ \) compared to the actual measurements. Future enhancements to the system will prioritize the analysis of backhand movements, as well as enhance the frame capture accuracy, even in less ideal scenarios.