THE DETECTION AND ANALYSIS OF THE MOTION TRAJECTORY AND SAFE PUNCHING OF BOXING BASED ON CNN

This study employs MediaPipe posture detection technology for motion capture, focusing on observing the movement of several key points of the right arm. By comparing the data model of the standard action, it can determine whether the current action is within the safe range and give judgment tips. The test includes OpenCV image processing and video analysis, Matplotlib data visualization combined with NumPy numerical calculation to realize action evaluation, providing feedback according to data comparison, convolutional neural network model training, and data management core function modules. Results indicate that the system performs well in judging action normativity, with errors within the normal range; error is basically within the normal range. This could help prevent sports injuries. This portable smart monitoring system uses a regular camera to record movements and then uploads them to a computer for analysis. The advantage is that it is convenient and practical.