Guorrat - The Research and Development of a Real-Time System for Generation of Ball Positions in Football

Abstract

In contemporary football, advanced analysis techniques are increasingly important, enhancing clubs’ capabilities to make strategic in-game adjustments. Central to this advancement is the availability of real-time analytics, which relies heavily on accurate player and ball tracking technologies. While real-time player tracking has become more accessible, automatic and reliable real-time ball tracking remains a significant challenge. This challenge is particularly acute for lower-tier football clubs that often lack the necessary resources and infrastructure. This thesis introduces Guorrat, a novel real-time system designed to generate accurate ball positions throughout live football matches. Developed to serve as a cost-effective alternative to existing real-time ball tracking technologies, Guorrat uniquely combines manual event tagging with real-time player positional data, which is generated by a system developed in parallel with Guorrat. This integration enables the system to produce precise ball location data during live matches, addressing the gap faced by resource-constrained clubs. The system’s effectiveness was evaluated in a series of experiments that demonstrated Guorrat’s capability to provide frequent and precise ball position data in real-time. The results indicate that Guorrat is not only feasible but also a practical alternative for real-time ball tracking in football.