Digital technologies in recent years have profoundly transformed sport related activities and operations, in fact the whole ecosystem of sports. Therefore, DJNkft the government office assigned with implementing digital technologies to support nation-level strategic objectives, deals with the timely and systematic introduction of digital technologies and solutions in the fields of both elite and amateur sports.
In line with these goals, i.e. within the framework of digital technologies based sports development, Neuron Solutions has been assigned by DJNkft to prepare a feasibility study on Artificial Intelligence (AI) enabled video motion analysis for Rhythmic Gymnastics (RG).
Video analysis based tools with AI-enabled functions are already applied in most elite sports for activities ranging from athlete scouting and recruitment through training, performance analysis, maintaining athletes’ health and fitness to sport business such as media and advertising.
Some of the best known application areas are analytic support of broadcasting popular games such as soccer or basketball (e.g. counting runs and assists) or tools helping scoring and judging (e.g. hawk’s eye in tennis or video goal judge in soccer).
RG counts among the most complex sports, from the points of view of both doing and scoring the exercises. There are sequences of fast and complicated motion combinations for which there are many things to consider ranging from how perfect are the body pose and position through timing, rhythm and synchronicity to artistic and total impact. Certainly this high complexity and speed of motions of exercises results in situations where even highly experienced judges do not not have a consent in scoring.
In our digital age, it looks evident that we should render digital technologies available for both facilitating judges’ decisions and supporting coaches and athletes in training.
In the feasibility study, we gave an introduction to AI-based video analytic technologies in terms of what they are capable of today and what further development directions are possible. Next, we outlined the options of creating an AI-enabled video analysis based scoring and evaluation system for RG.
As we concluded in the study, there are already advanced AI-enabled video motion analysis techniques, but solutions specific to RG have not been created. Although, there are significant technological challenges in developing such solutions, and in addition require considerable time and resources, they could be implemented on the basis of the currently available technologies. Such systems would actually provide significant support in sports development.
This was our first assignment when we had to deal with AI in sports. We have found this topic not only important but also quite exciting. It would be even more exciting to implement such a AI-based video analysis system for sports.
Cover photo: sportfaktor.hu