AI Analysis: The Future of Football Strategy?
In the weeks leading up to the Super Bowl, once it’s decided who’s going to be playing in the big game, both teams will spend endless hours watching and analyzing footage of their opponents’ games in hopes of getting a better understanding of their plays, positions, and formations, in order to develop a winning strategy.
But analyzing and annotating game footage is a tedious and time-consuming task, and finding ways to expedite the process could give a team a unique advantage. That’s why researchers at Brigham Young University leveraged deep learning and computer vision to create an algorithm that can identify and track the movements of players in order to determine formations and plays. Footage from BYU’s own football team was first used as training data, but it was later augmented with in-game footage from the Madden 2020 video game which was able to provide more diverse angles.
With the accuracy of player detection currently hitting 90%, and an 85% success rate with determining formations, the AI tool isn’t quite ready to be adopted by NFL teams just yet, but it won’t be long before a team of video analysts is replaced by a single laptop on the hunt for game-winning strategies.
