The NFL continues to crowdsource new ways to track head and helmet impacts during games from data scientists, and the winner of its artificial intelligence competition for the second year in a row comes from outside the United States.
The NFL and Amazon Web Services will award $100,000 for this year’s competition, with the top prize of $50,000 going to Kipei Matsuda from Osaka, Japan, the league announced Friday.
Working for Matsuda and the rest of the data scientists was to use artificial intelligence to create models that would detect helmet effects from NFL game footage and identify specific players involved in those effects.
Jeff Miller, executive vice president of the NFL, which oversees health and safety, said the league began tracking helmet effects manually a few years ago for a small number of games.
The tedious task of tracking every helmet collision, especially along the line of the scuffle, became difficult to do with more than just a small sample of games as the league tried to collect more data on head impacts.
By sharing game film and information with the data science community, the league hopes to continue to develop better systems that can more efficiently track those effects. The league estimates that Matsuda’s Champ system can detect and track helmet impacts with greater accuracy and is 83 times faster than a manual one.
Miller said, “Of course there were also the number of domestic participants, but the data science community is large and finding solutions with places or communities you wouldn’t normally talk to can be a very useful exercise.” ” “So I think we have proven that this model of working with the global data science community is helpful to us and will continue to be and we will continue to engage with it.”
The first year of competition in 2020 focused on models that detected all helmet effects from NFL game footage. That competition was won by Dimitro Poplawski of Brisbane, Australia., which included approximately 7,800 submissions from 55 countries.
This year’s competition focused more on specific player influences and featured 825 teams and 1,028 competitors from 65 countries and a total of 12,600 submissions.
“It was the most exciting competition I’ve ever experienced,” Matsuda said in a statement. “Detecting 2D images for computer vision is a very common task, but this challenge requires us to consider high dimensional data such as the 3D location of players on the field. NFL videos are also fun to watch, which are very “Important because we need to look at the data frequently during competition. I would be honored if my AI could help improve the safety of NFL players.”
Miller said the league’s goal is to create a “digital athlete” that can become a virtual representation of an NFL player’s actions, movements and impacts during a game and can be used to help predict and prevent injury. could. Future.
“This is novel to us and obviously of great importance to how we think about making the sport safe for athletes,” Miller said, adding, “It will certainly have an impact on training and coaching.” Without a doubt the rules will have an effect. It will certainly have an impact in terms of equipment, and the benefits that we can see from the equipment because now for the first time every time someone gets hit on the head during an NFL game, we will get a very good appreciation, and so, we have many of them. Will look for ways to stop it.”
Priya Ponnapalli, senior manager of Amazon’s Machine Learning Solutions Lab, said the ability of machine learning to analyze past data will help in creating a digital version of players across all positions in the future and analyze its types. Hits they take.
“Machine learning is a very intuitive process and you get a certain level of performance, and in this case we have some precise and comprehensive models,” Ponnapally said. “And as we collect more data, these models keep getting better and better.”
