With potentially 150 candidates vying for spots in the games, we needed to develop a model to identify the most promising qualifiers. Analysing boxing with AI presents unique challenges due to varying camera setups, lighting conditions, and lack of fixed viewpoints. Existing boxing analysis methods fell short of optimal statistical and predictive approaches, so we had to innovate our own solution.
We developed a unique system that we can’t talk about the details of in full detail but we built a model of performance that was used to quantify and strategically analyse blind spots. We utilised multiple AI solutions for this project to get the outcome we were seeking.
