The solution worked and the BYU team's algorithm was able to identify and label 1,000 images and videos from the game. Coupled with end-zone views that showed both the offense and defense, every player could be covered consistently. The camera view that proved most useful was an overhead, bird's-eye-view, vantage point where almost all the players could be seen. "It actually allows you to position the camera view in a variety of different places and gave us the control and consistency our algorithm needed." "We then stumbled upon a solution - the Madden 2020 NFL video game," Lee told ZDNet. So, to save time, the team at BYU decided to come up with a proof of concept, using a game-based solution instead of waiting for the camera angles to sort themselves out. And the defensive players closest to the line of scrimmage are also obstructed by the offensive line.Īlso: This new technology could blow away GPT-4 and everything like itįor a machine-learning algorithm to work effortlessly and automatically, it would have to be able to rely on angles that are consistently the same across all college game footage involving BYU or any of their opponents. Then, there is the quarterback, who stands in front of the center blocking him. When the BYU engineering team started to look at their college's football tapes, it was quickly apparent that there was a big problem: there were no consistent workable camera angles.Īt the college level, camera placement for games tends to be all over the place and not all players on the field are always visible by one camera angle. For machine learning, on the other hand, it's a piece of cake. Game-tape prepping is painstaking work, especially for humans. What's more, it's hard to get the analysis right. If you're also scouring tapes historically - looking at plays from previous years to add more depth to your analysis - then that's enough time to ensure you never see daylight. There are 55 players on each team's roster and 32 teams in the league. That level of analysis involves a lot of hours. They then need to make insightful observations on everything from overall strategies being employed by the opposition coach down to granular details about player movements and tendencies, so that countermeasures can be hatched.Īlso: This machine learning project could help jumpstart self-driving cars again
0 Comments
Leave a Reply. |
Details
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |