Accurately analyzing core strength training postures is the technical cornerstone for improving foot flexibility of badminton players. In this paper, a badminton core strength training posture intelligent analysis system is designed based on deep learning algorithm. Firstly, a human motion data acquisition subsystem was constructed based on IMU sensors to obtain high-precision motion data. Then in order to accurately identify the human motion patterns, a motion recognition model based on symmetric coding, time scale coding and structural coding is constructed, and the model is trained based on a deep learning algorithm (Adam optimizer). Finally, a comparative analysis of the training effect through the compliance and violation indexes of training movements such as pull-ups, sit-ups and push-ups is proposed. The mAP and GFLOPs/V values of the constructed motion recognition model are 93.65% and 25.77, respectively, and the model performance is greater than that of mainstream recognition models. The evaluation effect of the posture intelligent analysis system on the training movement meets the practical use requirements, and the foot flexibility of badminton players is improved after the application of the system, and the p-value is less than 0.05. The deepening application of deep learning technology in the field of sports brings a broader vision for the long-term development of sports.
