In NBA lingo, Garth Case has what the players call “court vision.”
Inside the blur of the game, he sees scoring opportunities before they develop. And within the sport’s deluge of data, he spots innovative possibilities rising among the terabytes.
Case, the NBA’s vice president of IT, helped develop a new, AI-powered tool that automatically sorts and curates the massive stream of game photos generated by 30 teams across 82 regular season games and the playoffs.
NBA Photo Sorter, which Case demoed earlier this month at Microsoft Build in Seattle, replaces the tedious method league photographers have long used to manually tag each game image with the names of players, teams, celebrities and more.
Built with Cognitive Search – a new AI-driven, content-understanding capability available from Microsoft – the platform saves NBA employees hundreds of hours of work while increasing content accuracy and unlocking new business scenarios. But, Case says, it is a first step that only hints at the tech’s enormous potential.
Transform caught up with Case after Build to ask him a bit more about his vision.
TRANSFORM: How much time does NBA Photo Sorter save the league’s photo department?
GARTH CASE: Say we had an event today. Photographers give us all their photos. Then another team would sit there, combing through all of those photos, determining who is in them and creating folders for those photos. Time would mount up.
Photos from one event could take up to four weeks to categorize. We can now reduce that four weeks down to about one day. When the system is properly trained, we can whip through it in a few hours.
But there’s something bigger behind this whole thing. There is this effort to bring all of our content together. Everything from videos to photos to tons of articles that get published. There’s a lot of information generated through every game.
No comments:
Post a Comment