Almost every idea or product at the 2019 MIT Sloan Sports Analytics Conference could be traced back to a “what if…” conversation between coworkers.
STATS, which has been involved in player tracking in the NBA dating back to 2009 and professional soccer since 1999, has found an answer to the question of, “What if we could track players through a broadcast video feed, instead of in-house cameras?”
Through AutoStats, the first patented AI-powered technology to capture sports tracking data via broadcast video, STATS has unlocked the potential to capture tracking data from any game, past or present.
The “It’s impossible to compare players from different generations” argument may now be obsolete, as Jason Concepcion of The Ringer found out and highlighted on his NBA Desktop show.
Dr. Patrick Lucey, vice president of artificial intelligence (AI) at STATS, was a member of the “Unlocking Potential: The Next Generation of Tracking Data” panel, in which he spoke on the power and implications of AutoStats to the future of tracking data.
“At STATS, we have been collecting tracking data since 1999. Tracking data is not a new thing. It’s 20 years old. But tracking data is limited because you have to have computer vision in each venue. That really limits the scale,” Lucey said during his opening remarks. “…If you have video, we can get tracking data. And we can collect really detailed tracking data.
“Using this technology, we can actually digitize it, query it, do analysis.”
Sujoy Ganguly, head of computer vision at STATS, presented “Going Beyond Tracking Data: Using Body-Pose for Next Generation Analytics,” to shine a brighter light on AutoStats and discuss how using OpenPose – a product manufactured under license from Carnegie Mellon University – to track players directly from broadcast video will expand the availability of tracking data, as well as deepen the quality of data by providing human pose estimation.
“[Tracking data] is too valuable not to use,” Chris Capuano, a former MLB pitcher, Fellow at MIT’s Sloan School, and member of the Unlocking Potential: The Next Generation of Tracking Data panel stated at one point during the discussion.
The panel went on to discuss a number of topics relating to player tracking, including latest developments, data ownership, ethics, and the potential impact on fan experience. For the full panel discussion, click here.
On February 25, STATS and the Orlando Magic announced an exclusive deal that allows the Magic to use AutoStats data to help analyze collegiate players in an effort to improve evaluation and decisions for the NBA draft.
Learn more about AutoStats at https://www.stats.com/auto-stats/