The Baltimore Orioles wanted to pitch better, so they asked the University of Waterloo for help

Kate Bueckert · CBC News

· Posted: May 03, 2025 7:00 AM EDT | Last Updated: 22 minutes ago

Sports teams are ever looking for tweaks they tin marque to assistance athletes execute amended and a caller artificial quality strategy being developed astatine the University of Waterloo is focusing connected shot pitchers.

The project, called PitcherNet, started successful 2022 and is simply a collaboration betwixt the assemblage and the Baltimore Orioles. It looks to usage artificial quality to analyse however a pitcher throws a ball.

Many Major League Baseball stadiums are equipped with high-resolution camera systems from a institution called Hawk-Eye Innovations, which allows the location teams to seizure and analyse the movements of their players. The strategy has up to 12 high-speed cameras placed astir the field.

But erstwhile teams are playing distant games, oregon if teams want to analyse the movements of players successful the insignificant league system, they don't person entree to the aforesaid technology.

"They came to america and the task was fundamentally to find a mode successful which we tin mimic the Hawk-Eye tracking strategy with conscionable 1 camera, oregon possibly you tin conscionable usage the broadcast feed," Jerrin Bright, a PhD pupil who is portion of the probe project, told CBC K-W's The Morning Edition host Craig Norris. 

"They wanted to spell into quantitative metrics, similar trying to find the merchandise point, the extension. But the occupation is erstwhile you spell into a insignificant league oregon an amateur league, the scouts wouldn't beryllium having entree to this Hawk-Eye system, which means that they volition person to spell with qualitative analysis. And this is wherever our strategy comes into play."

Bright says the AI strategy developed astatine the assemblage uses a video, from a broadcast provender oregon someone's cellphone, and figures retired the pitcher's skeleton by utilizing 3D quality modelling. It past pinpoints 18 joints connected the body.

"Once we get these 18 associated positions, we bash quantitative investigation utilizing instrumentality learning models. And with that, we'll beryllium capable to extract antithetic transportation metrics," Bright said.

To bid the AI algorithm, the researchers make 3D avatars of pitchers to way their movements and presumption them from antithetic vantage points. The strategy goes implicit transportation benignant — whether it was a fastball, slider, curveball oregon different benignant of pitch — the merchandise point, velocity, merchandise hold and handedness, which means they tin isolate the pitcher's throwing manus to analyse however the shot is held.

"These transportation metrics could yet beryllium utilized to bash a batch of show indications, look into the longevity of the players and look into ways successful which they tin optimize their pitching action," Bright said.

Scouts could usage cellphone video

CBC News reached retired to the Baltimore Orioles for remark connected PitcherNet but did not person a response. The squad is partially backing the research.

Sig Mejdal, the team's adjunct wide manager, told the Baltimore Sun paper that the strategy is "still a enactment successful progress, but determination person been cases already wherever it's been utile to use."

Mejdal told the paper biomechanics is "relatively new" successful the shot world, but "there has to beryllium galore things we could uncover that close present stay retired of reach."

John Zelek, a prof of systems plan engineering and co-director of the Vision and Image Processing Lab astatine the University of Waterloo, says what they've designed is simply a elemental strategy compared to Hawk-Eye, but it's effective.

Zelek said the strategy could beryllium utilized successful a assortment of antithetic ways, including by scouts who are looking astatine imaginable players to sign. They could instrumentality videos portion sitting successful the stands watching a shot crippled and the video could beryllium sent done the PitcherNet strategy to analyse an up-and-coming player's performance.

LISTEN | UW researchers usage AI to analyse shot players' movements:

On-site enactment this summer

Bright admits he's much of a cricket instrumentality and "a numbers guy" but he "got into shot aft joining this task and it has been precise interesting."

He said Waterloo researchers conscionable with squad representatives astir erstwhile a period to speech astir advancement and however the strategy tin beryllium improved truthful the squad gets the metrics they need.

Bright says determination are different applications for the software, and they're looking into however it could assistance hockey teams, too.

He'll beryllium going to Baltimore this summertime to enactment connected the strategy successful the Orioles' stadium and further make it.

"I deliberation the imaginable that AI systems similar PitcherNet tin person connected a existent game, that is precise fascinating and I look guardant to what I tin lend successful this aspect," Bright said.