PROVO, Utah (ABC4) — BYU researchers created an artificial intelligence algorithm that could possibly change the way football teams prepare for a matchup.
Dr. DJ Lee and his team of BYU master’s and PHD students trained the AI algorithm to accurately determine where a player is located on the field and the offensive formation of the opposing team.
Right now, according to Lee, both collegiate and professional teams spend hours manually analyzing and annotating footage of their opposing team before a game. This algorithm would do that for them.
“They’re doing it manually right now and this algorithm can speed up that process,” said Lee.
Lee and his team did this by first looking at real game footage provided to them by BYU Football.
However, the video given to them only showed a few different angles of the field, and in order to help the AI learn, they needed a wider range of angles. Researchers then purchased Madden NFL, a football video game, and used it to train the AI.
The students played the game for hours, collecting thousands of videos and images, including aerial views of the field and footage from behind the offense.
Lee said the team would adjust the camera in the video to about “thirty degrees above ground and from the back of the offense and that’s perfect because that’s what the football team is analyzing offense formation.”
Despite still being in the early stages of development, the program is over 90% percent accurate when determining the player position and 85% percent accurate on determining formations.
The few inaccuracies that do occur happen when players block other players.
Lee believes this AI algorithm can work with any team sport and hopes to be able to apply it next to basketball, volleyball, and baseball.