Many Capital District Youth Soccer League (CDYSL) teams struggle to keep up with the financial demands associated with paying for three referees for each match throughout a typical season. Amid a broader shortage of referees across many levels of youth soccer, games are increasingly officiated by a single referee who must perform the responsibilities that are typically distributed among three officials on the field. This shift places additional pressure on referees while also raising questions about how officiating quality can be maintained under these constraints. This project will evaluate the potential of computer vision in combination with Message Queuing Telemetry Transport (MQTT) to assist with officiating by making determinations on whether a soccer ball has exited the field of play towards the end line of a soccer field. The proposed implementation integrates cameras, a single-board computer, and a microcontroller that communicates using MQTT with the single-board computer. The computer vision will rely primarily on Kalman filtering, a technique used to aid in the event of occlusion events and in predictive tracking of the targeted ball, and traditional image processing techniques to aid reduce visual noise, extract vital features, and relay the ball's position with respect to the end line. In implementing the aforementioned techniques and protocols, the created device will achieve a minimum accuracy of 80%, make and relay decisions to the referee within 6 seconds, have a minimum continuous-use battery life of 2.5 hours, and be designed and deployed under a cost of $300. By automating this aspect of officiating, the project will provide an accurate, cost-efficient, and functional supplementation to aid both referees and youth soccer leagues in maintaining high standards of play.
Acknowledgements: Prof. Momota