canoe rAIce

Project duration: 01/2023 – 12/2024

Due to the very high and ever-increasing level of performance in canoe racing, particularly at the top international level, the optimisation of race strategy and competition tactics is becoming increasingly important. Relevant parameters for assessing the performance of boats competing in races can currently only be determined manually, which is a time-consuming process. Ultimately, this results in only limited opportunities to adjust one’s own race tactics in light of competing boats.

In this project, algorithms are being developed and integrated into software to determine, fully automatically and immediately after races, the parameters relevant to coaches for competition analysis. This will provide access to unique insights and open up new scope for action. These analyses are based on camera footage of the entire race over a distance of up to 1,000 metres. From these recordings, the individual boats, their routes and corresponding times are to be automatically recorded by combining classic computer vision algorithms, modern deep learning methods and predictive models.

The new analytical techniques developed as part of the project are intended to be used in future international competitions and to help improve the chances of German athletes achieving top placings in the disciplines of kayaking and canoeing, in both single and team boats.