How AI reads a match: from video to pass maps
For a computer to “see” a match, images first have to be turned into data. Tracking systems detect the players and the ball in the video and calculate their position many times per second.
With that information you can draw heat maps, passing networks (who passes to whom) or the distance between a team's lines. You can also detect patterns: where an opponent attacks most or how they react when they lose the ball.
But AI has limits. It can make mistakes identifying players, especially with low-quality cameras. And a map does not explain intentions: it does not know whether a player moved to pull a defender away or was simply tired.
The best way to use it is as a help for the coach: a summary that points to where to look, which is then confirmed with video and experience.