Computer Vision Based Goal Decision System for Football Match: A Prototype Study

Abstract

The need for autonomous systems increases day by day in order to be able to appeal to the referee's decisions in soccer. The main reason for this increase in demand is the fact that referees can make critical mistakes that could have an effect on important matches. Therefore, in addition to referees in games, assistant referee systems with a computer vision system have been engaged in soccer games. It is aimed to provide fair competition environment by reducing mistakes with such systems. In this study, a computer vision system, GolKaSis, was designed for the determination of the goal event in football matches. In the GolKaSis, firstly, the images taken from the videos obtained from the cameras placed in a region close to the tower in the designed football field prototype are separated as negative and positive. The ones that are positive from these images are those showing the goal event, the ones that are negative are the video images where the goal event is not coming to fruition. In the developed computer vision system, the matching of the positive video with the video images taken in real time is determined by providing Haar Cascade Classifier. It has been seen that the computerized vision system proposed in the test procedures on the designed prototype correctly determined the goal event with 91% success rate.

Keywords

• Football match
• Goal-line detection Computer vision system Haar Cascade classifier
• Goal-line detection
• Computer vision system
• Haar Cascade classifier