Researchers from the University of Waterloo utilized artificial intelligence (AI) tools to efficiently capture and analyze data from professional hockey games, revolutionizing the field of hockey analytics with significant implications for the sports industry.
The current practice of hockey analytics relies on manual analysis of game video footage, with crucial decisions about players’ careers in professional hockey, particularly in the NHL, being made based on this information.
Dr. David Clausi, a professor in Waterloo’s Department of Systems Design Engineering, explained, “The aim of our research is to interpret hockey games through video more effectively and efficiently than humans can alone. The speed and complexity of the game make it impossible for one person to track everything happening on the ice.”
Hockey players move quickly and unpredictably, making it challenging to track and analyze each player during a game due to the limitations of camera visibility and uniform identification. This manual process is prone to errors.
The AI tool developed by Clausi, Dr. John Zelek, research assistant professor Yuhao Chen, and a team of graduate students utilizes deep learning techniques to automate and enhance player tracking analysis.
In collaboration with Stathletes, an Ontario-based hockey performance data and analytics company, the research team manually annotated game footage frame-by-frame from NHL broadcasts, teaching the AI system to observe, compile data, and generate accurate analyses and predictions.
The system’s algorithms demonstrated high accuracy rates, scoring 94.5% for player tracking, 97% for team identification, and 83% for individual player identification.
While the research team continues to refine the prototype, Stathletes is already using the system to analyze hockey game footage. The potential for commercialization extends beyond hockey to other team sports like soccer or field hockey through retraining the system’s components.
“Our system can provide valuable data for various purposes,” Zelek stated. “Coaches can use it for strategic planning, scouts can identify talent, and statisticians can enhance teams’ performance. It has the potential to revolutionize the sports industry.”
