"Chinese scientists" created an AI referee for football matches, its accuracy is already 84%
Football matches are now analysed not only by VAR but also by artificial intelligence. Researchers from Shanghai's National Chiao Tung University (NCTU), in collaboration with Alibaba, have created an innovative MatchVision system that can recognise events on the football field, analyse them and even generate text comments. This is a new stage in the interaction between technology and sports.
Here's What We Know
MatchVision is based on the largest football video database to date - SoccerReplay-1988. This database includes 1988 full matches from the biggest football leagues in Europe and the Champions League for the period from 2014 to 2024. This is almost 3,300 hours of video, which are systematised and annotated to help artificial intelligence learn.
A special feature of MatchVision is its ability to recognise up to 24 different types of events, such as goals, shots, fouls, offsides, and even tactical changes in possession. But the real highlight is the ability to analyse rule violations from different angles, determine their severity and create commentary to better understand the key moments of the game.
For those who want to know more: VAR (Video Assistant Referee)
VAR (Video Assistant Referee ) is a video replay system used in football to assist the referee in making difficult decisions. It allows you to review controversial moments to ensure accuracy and fairness in the game. The main functions of VAR are to check goals for violations (offside, foul, etc.), the correctness of penalties, flagrant fouls for red cards, and to correct referee mistakes, for example, when the offender is incorrectly identified. The system works like this: in a special room, VAR assistants watch the match in real time using multi-camera recording, and in the event of a controversial situation, they signal the head referee. The referee can review the moment on a monitor near the pitch and make the final decision. The benefits of VAR include improved decision accuracy, reduced errors and minimised human error, but the system is not without criticism due to the slowdown in play, long pauses and subjectivity of individual decisions. VAR was officially introduced at the 2018 FIFA World Cup and has since become an integral part of professional football.
Results and accuracy
The system achieved an accuracy rate of 84% in event recognition tests and outperformed all existing models in generating text comments. This means that artificial intelligence not only understands what is happening on the pitch, but is also able to explain it in a way that fans can understand. For example, the system can describe the context and significance of a particular foul, including an assessment of its seriousness.
Open future
The developers plan to make the code and model available to the community via GitHub. This opens up new prospects for automating sports video analysis, creating detailed match reviews, and even improving the VAR system.
MatchVision demonstrates how artificial intelligence can not only make the work of commentators and coaches easier, but also become an important tool for football fans who want to understand their favourite game more deeply.
Now the question remains: will AI be able to become a full-fledged referee one day? The future will tell.
Source: Github.io
