Sports betting company

The client is a large sports betting company, wagering on many sports events across the globe at any given time. We were engaged as part of the research department, to test the performance impact of modern deep learning approaches as well as provide new capabilities using computer vision and model explainability.

Among other deliverables, we produced an end-to-end computer vision system to track and classify boats in speedboat races. This involved using both traditional computer vision techniques and deep learning (labelling, fine tuning, deployment) to classify the races. We also piloted our proprietary software iMpLY to aid in machine learning interpretability and diagnostics.

The computer vision approach is now being used in a number of strategic projects at this company. The work done in testing production models and comparing performance allowed the business to gain a deeper insight into their betting models and finetune them accordingly.