Biomechanics analysis of real-time tennis batting images using Internet of Things and deep learning

This work aims to integrate the Deep Learning (DL) technology with the tennis players’ batting angle selection and provide an intelligent basis for the players’ biomechanical analysis. This work collects the image data of tennis players using the acquisition circuit of sensor images based on the Internet of Things (IoT) Modbus. The General Adversarial Network (GAN) optimized by feature mapping is used to optimize the tennis players' video image, the VICON system is adopted to analyze the joint movement indicators for different batting angles, and data are statistically analyzed. The results show that the performance of the proposed DL GAN algorithm is about 4.5db and 0.143 higher than other algorithms in Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity (SSIM) directions, respectively. The knee movement speed under the two batting angles is 2.59 ± 0.07 m/s and 2.21 ± 0.065 m/s, respectively, with significant differences (P < 0.05); there is a significant difference in the speed of the right ankle and right hip at the swing stage of forehand top-spin and forehand flat (P < 0.05); there are significant differences in the speed of the right ankle, right knee, and right hip in two forehand battings (P < 0.05); there are significant differences in the trunk torsion angle and speed at the swing stage and the end of swing (P < 0.05). The difficulties and challenges of this work are that the image identification network is prone to overfitting, and the global average pooling layer replaces the fully connected layer of the network, which reduces the parameters of the model and shortens the image recognition time. Meanwhile, it shows that the method proposed, based on the biomechanical analysis of tennis players' batting images, can effectively collect their batting video images and improve the image definition, from one-time biomechanical analysis to image acquisition and then to quality optimization, which is practical and efficient.