Game analysis in soccer using DyCoN
In the (semi) automatic location identification, which is common in todays elite football, an enormous amount of data is produced. The recognition of every 22 player positions and the ball position can be detected with an accuracy of 25 frames per second. The goal is finding tactical patterns out of the enormous amount of data and possibly showing the weakness of the opponent. The neural network DyCoN, developed in Mainz, helps to identify these patterns and processes in the tactical game. The DFG-funded project will be carried out in cooperation with the German Sport University (DSHS) in Cologne.