Video-based system for technique improvement in aquatic sports

Purpose and goal

The aim of the project was to study the technical feasibility of video-based, automated analysis of swimmer performance and technique. Within the scope of the project, advanced computer vision algorithms were developed both for localization and target tracking, based on the latest advances in machine learning and deep neural networks. The feasibility of using a hand-held camera rig for this purpose was also studied. The main goals of the project were met and the desired performance achieved.

Expected results and effects

As the project outcome was mainly positive, Mometric will continue development of the system. Moving forward we will focus on the camera rig ergonomic and technical design, as we judge these to be the largest hurdles on the way to a commercially viable product.

Planned approach and implementation

The research and development work was conducted part-time over a longer period. This turned out to be an advantage as interactions with hardware suppliers took much longer than anticipated; while waiting for physical prototypes, software development could continue at the planned pace without being blocked. However, there was not enough time to complete the final work package comprising the web portal as the other tasks were prioritized and took longer than expected.