IoT-based analysis of tennis player’s serving behavior using image processing

Abstract

In recent years, with the Internet of Things (IoT) and artificial intelligence and the rapid development of technology, various sports sectors have benefited from technological and scientific advancements. The technical and scientific components that have permeated different sports sectors are becoming progressively less important as technology developments and the IoT requires disconnected. To address this problem, this study proposed a based deep learning technique for the tennis sports industry and the implications of smart athletes’ fitness using the support vector machine (SVM) algorithm. The three basic categories of IoT-based smart fitness are player trackers, which include wearable and non-wearable sensors, movement analysis, and player applications. The skills and strategies of tennis players have improved through the use of sports products, and people have come to value physical education more as education has expanded. The development of a smart tennis system is meant to address the slow detection and unpredictable player and coaching staff movements of the traditional game of tennis. The background of the video is rebuilt using a median filter algorithm, and its target is located using an inter-frame difference technique to identify the track’s designated shoulder marker point. Using an enhanced SVM, it creates a tennis service model for evaluating and examining marker point tracks. The experimental results compare the proposed method with different machine learning methods such as decision tree (DT), random forest (RF), deep neural network (DNN), and recurrent neural network (RNN) algorithms. The classification achieved excellent performance by the player motions, and the SVM is better than the other deep learning. The test results proposed scheme the tennis service model achieves a classification accuracy of 97.5% in accurately categorizing different types of service trajectories.