Amateur hockey given professional viewing experience courtesy of machine vision startup co-founded by Vector researcher

March 7, 2022

By Ian Gormely
March 3, 2022

A new startup co-founded by Vector researcher James Elder aims to bring a curated, professional-level viewing experience to amateur hockey. AttentiveVision uses machine vision models to follow the flow of play, automatically adjusting camera angles as the location of play on the ice changes. 

Professor and York Research Chair in Human and Computer Vision at York University as well as a Vector Faculty Affiliate, Elder has spent most of his career studying human perception, or how we see the world, and trying to figure out how computers might mimic that

Humans have the ability to direct our eyes to wherever the brain thinks we need more information, something that computer vision systems designed to perform specific tasks in controlled environments like medical imaging or assembly lines, can’t replicate. “But as we move to more general forms of AI,” says Elder, “we’d like to have computer vision systems that can handle less structured and controlled problems in the wild.” Developed at Elder’s Laboratory, attentive machine vision sensors that can better mimic the visual behaviour of humans go a long way to reaching that goal.    

AttentiveVision’s Attentive Puck Tracking demo.

Among the potential applications, they identified an immediate need in the world of amateur hockey. “We thought there was an opportunity to apply some of our attentive sensing technology to amateur sports, to give them something that approximates that professional level of curation, something that could be affordable for your local rink.”

Elder and his team developed and trained an AI model on footage from York University’s varsity men’s and women’s hockey teams. The model learned to infer the location of play from “the positions, speed, and appearance of the players,” rather than the puck. Their system is currently accurate to within roughly four meters of the puck’s location. “The rink is 60 meters long, so that turns out to be sufficient for camera planning purposes.” 

While professional hockey leagues and teams have invested millions into complex video systems, amateur organizations can’t afford the expensive equipment and labour. AttentiveVision’s technology provides them with an affordable alternative. Once deployed, the technology, based on a proprietary 8K camera and their AI model, is able to track the play across the ice, giving viewers a curated, zoomed-in viewing experience available in real-time online. Teams paying AttentiveVision’s monthly fee also gain access to the company’s suite of tools to help manage their roster, annotate games, and track statistics. “They really want to engage their fans and their families,” says Elder. “The more presence they have online, the more they’re going to be able to do that.”

A startup was never the end goal for Elder, but he was encouraged by York’s commercialization team to spin out the company from his lab. Founded in 2019, AttentiveVision currently has installations in BC and Ontario.  

While the company’s current focus is hockey, Elder believes there is a lot of potential to apply the technology to other sports like soccer or basketball. He also sees separate revenue streams in micro-targeted advertising from local companies that could be added onto the boards or ice surface. “Amateur sports are really part of the social fabric of a community,” says Elder. “We would like to be able to create an environment that is really rich and supportive of that.” 



12 AI Trends to watch for in 2024

Trustworthy AI

Key takeaways from the All In 2023 conference

South Korea trip reaffirms Canada’s global leadership in responsible AI