
University of Waterloo, Orioles partner on AI project to track pitcher biomechanics
Staying competitive in the American League East means staying on the cutting edge of technology.
That’s why the Baltimore Orioles have partnered with researchers at the University of Waterloo to develop artificial intelligence technology, called PitcherNet, that will help the Major League Baseball team track the biomechanics of pitchers at every level of the organization, from the collegiate ranks and through its minor-league system.
PitcherNet takes the main camera feeds fans are used to seeing — either from behind the pitcher or behind the batter — and combines it with video footage shot on a smartphone by a scout or team official. Using AI, PitcherNet then extrapolates the pitcher’s entire throwing motion, recreating the level of detail that the 12-camera Hawk-Eye system is capable of producing.
“What we do is we try to extract 2D information, 2D joint position information, and then extrapolate using 3D avatars,” said Waterloo PhD student Jerrin Bright. “The 3D avatars are like blobs encompassing a human itself and we use those blobs to basically find the 3D human pose, which can be represented with the shape of the person, so that’s the idea behind it.”