MODELLING OF THE QUALIFICATION & IMPROVEMENT OF TENNIS STANCE FOR PLAYER PERFORMANCE IMPROVEMENT USING 2D ANALYSIS OF VIDEOS TAKEN FROM A MOBILE CAMERA (PART A)

DOI: https://doie.org/10.0118/Jbse.2025328537

Abhilash Manu*, Dr. D. Ganesh, Dr. Aravinda H.S, Mayur Gowda R., N.R. Swedith Reddy

Keywords:

Tennis Performance Optimization, 2D Video Analysis, Recurrent Neural Networks, Pose Estimation, Biomechanical Modeling, Unsupervised Clustering, k-Means, Joint Angles, Adaptive Acceptance Areas, Motion Artifacts, High-SNR Imaging, Cloud Computing, Pa

Abstract:

This study outlines a novel framework for enhancing tennis performance through the optimization of biomechanical postures during specific tennis shots, utilizing advanced 2D video analysis and stabilization techniques coupled with Recurrent Neural Networks (RNNs). Through precise pose estimation algorithms, skeletal key-points are extracted to compute joint angles via vector dot product formulae. These keypoints allow for detailed biomechanical analysis and the classification of movement patterns using unsupervised clustering methods such as k-means. The analysis is refined through adaptive acceptance areas determined by a blend of distance metrics, enhancing the accuracy of pose alignment evaluations. Challenges such as motion artifacts, variable lighting conditions, and low signal-to-noise ratios (SNR) are addressed through the use of high-SNR imaging devices and optimized camera calibration. This approach ensures high-quality data capture essential for reliable computational analysis. The study leverages cloud platforms for processing, maintaining strict confidentiality of data while harnessing scalable computational resources. This integration facilitates robust kinematic analysis through part affinity fields and TensorFlow Lite, enabling real-time feedback on player movements and biomechanical alignment. The research makes significant contributions by integrating advanced computational algorithms and tailored hardware solutions to surpass traditional video analysis limitations. Through the detailed kinematic analysis of player movements and the innovative use of clustering algorithms, the study provides a comprehensive method for enhancing tennis performance. This methodology not only refines current coaching practices but also sets new standards in sports performance analysis, ultimately aiming to revolutionize tennis training through data-driven insights and technological advancement. The modelling results shows the effectiveness of the method developed, which could be used for a host of science & engineering applications. 

PDF