Development and testing of a markerless system for real-time 3D feedback of kinematic parameters in a canoe counter-current facility using multi-camera data
Project duration: 01/2022 – 12/2022
In elite sport, the continuous improvement of training and analysis methods is of enormous importance. The aim is, in particular, to make the best possible use of athletes’ individual potential to achieve optimal technical performance. In high-performance training, this is achieved most effectively through technique training with immediate feedback. Athletes receive visual feedback synchronised with their movements. This type of training is known as ‘measuring station training’ and is only possible in a laboratory under specific conditions, using a canoe counter-current system.
Whilst athletes are carrying out training runs, they are simultaneously filmed using a multi-camera system. The video recordings of the athletes produced in this way are subsequently analysed using specialised software. Part of this analysis involves determining the 3D kinematic parameters of the movement, knowledge of which ultimately enables the identification of areas for improvement. However, at present, this analysis can only be carried out with a high degree of manual intervention and is very time-consuming.
In this project, the technical analysis of kinematic parameters in the canoe counter-current facility is being transformed into a real-time analysis application. To this end, appropriate AI-based algorithms – primarily from the field of image and video processing – are being developed over the course of the project and implemented to meet real-time requirements. In future, a measurement system is intended to record selected kinematic parameters without the use of markers during training – or even with stroke-specific precision – and to provide appropriate feedback on display devices, thereby enabling immediate intervention with regard to sporting technique.
This project is funded by the Federal Institute for Sports Science (BISp) following a resolution by the German Bundestag.
Contact
Prof. Dr.-Ing. Mirco Fuchs
Chair of Computer Vision and Machine Learning
Telephone: +49 (0)341 3076 3104
Email: mirco(dot)fuchs(at)htwk-leipzig.de
