JCSE, vol. 20, no. 3, pp.204-215, September, 2026

DOI: http://dx.doi.org/10.5626/JCSE.2026.20.3.204

A Study on Intelligent Score Recognition Methods for Tennis Matches Based on Multimodal Visual Fusion

Hui Feng and Guangfu Zhang
College of Public Courses, Guangdong University of Science and Technology, Dongguan, China School of Physical Education (Tennis Teaching and Research Section), Guangzhou Sport University, Guangzhou, China

Abstract: With the demand for automated refereeing systems in tennis matches on the rise, traditional manual scoring methods struggle to ensure stability and consistency in high-speed situations. To address this problem, this study proposes an intelligent scoring recognition method for tennis matches based on multimodal visual fusion. This method sets up a multi-camera visual acquisition framework. It uses an object detection network to extract the ball’s spatial location and combines it with a trajectory tracking algorithm to generate continuous motion paths. At once, a human pose recognition model is used to obtain the key movement information of the players. Based on this fact, a cross-modal feature fusion model is established to get the joint representation of visual, trajectory, and pose features. Specifically, a network is built to identify shot and landing point events, enabling temporal modeling of key match events. A rule-based inference model is then used to figure out the ball’s landing zone and make scoring decisions. Experimental results indicate that this way effectively improves scoring recognition accuracy in complex match situations and keeps stable real-time processing capabilities, so it provides technical support for the automated realization of intelligent sports refereeing systems.

Keyword: No keyword

Full Paper:   0 Downloads, 3 View