Practice-orientated research in women's high-performance basketball

The topic of "women in sport" is becoming increasingly relevant and is receiving more and more attention in the public debate. Numerous sporting events with female participants are inspiring spectators. Various political initiatives are currently aimed at promoting gender equality and equal opportunities. Nevertheless, there is still an imbalance in competitive sport, particularly in terms of public perception, structural framework conditions, funding opportunities and also in research, as the example of the gender data gap makes clear.

Professor Daniel Memmert and Dr. Marc Garnica Caparrós from the Institute of Exercise Training and Sport Informatics (ITS) at the German Sport University Cologne have been working for years on Artificial Intelligence and Machine Learning on the basis of various sports data as part of DFG and BMBF projects. For example, a new platform called "floodlight" was developed for sports data analysis, which was designed as a freely accessible package to simplify and standardise data processing.

The Institute of Exercise Training and Sport Informatics (ITS) has now acquired new funding from the Federal Institute of Sports Science (BISp) together with the German Basketball Federation (DBB). As part of the research focus "Women and Girls in Competitive Sports" (FeMaLe). The title: Using computer vision and machine learning to improve training and competitiveness in women's 3x3 basketball.

The aim of this two-year project is to support the DBB with state-of-the-art data collection techniques for 3x3 women's basketball competitions using computer vision and information retrieval technology. The resulting data sets will be owned by the DBB and made available for knowledge extraction and transfer. In addition, an information extraction pipeline will be used in co-operation between DBB and ITS to collect retrospective data from all broadcast matches. In addition, the resulting event and tracking data sets will be analysed to gain evidence-based insights into the women's 3x3 basketball discipline. Another important aspect of this project is the transferability of the knowledge gained to the field of conventional women's 5x5 basketball. Firstly, the results of this project will be directly applicable to the training and performance of 5x5 basketball players. Secondly, and more importantly, 3x3 basketball will be used as a research field for women's basketball. Therefore, the data-based findings of this project will be translated into concrete recommendations for action.

The Managing Director of the Institute of Exercise Training and Sport Informatics, Prof. Daniel Memmert, sums up: "We are delighted that we are now able to apply our expertise in 3:3 women's basketball at national squad level on the basis of our fundamental research in AI and ML."

Peter Radegast, Science Coordinator of the DBB, adds: "With the effective use of the developed computer vision models, the DBB and the organisers of 3x3 competitions will be able to provide real-time information and statistics to the coaching staff and journalism. This will improve the quality of our sport and the training of players and teams."