This chair aims to develop cutting-edge approaches in artificial intelligence (AI) for the automatic understanding of sports images and videos. Created with the support of EVS, it marks a significant milestone in the alliance between academic research and industry in the field of AI applied to sports media. Professor Cioppa, now a lecturer at the Montefiore Institute within ULiège’s Faculty of Applied Sciences, will lead the scientific direction of the chair.

A strategic partnership at the crossroads of sports, AI, and innovation

EVS, recognized globally for the quality of its broadcast images during the world’s most iconic events, has been integrating AI techniques into its solutions for nearly a decade. Among its flagship innovations:

•    Xeebra®, a multi-camera review system that uses AI for offside line placement
•    XtraMotion®, a replay effects solution that uses Generative AI to create super slow-motion replays without the need for high-frame-rate cameras, eliminate motion blur, and add a cinematic shallow depth-of-field effect

Building on this experience, EVS is strengthening its collaboration with ULiège to anticipate future uses of AI in sports.

High-impact research

In a context where the volume of sports video content is growing exponentially, automatic analysis tools are becoming essential. The chair will develop innovative solutions based on the following research areas:

1.    High-dimensional data perception: Extracting and interpreting semantic information from rich and complex data such as video, which combines spatial, temporal, and color dimensions.
2.    Long-form video understanding: Designing AI models capable of processing the full temporal scope of a video to extract relevant semantic information, even over long durations (e.g., a 90-minute football match).
3.    Memory in neural networks for video: Developing AI models that can model, organize, and efficiently use temporal memory to retain relevant context over long sequences.
4.    Multimodal combinations: Leveraging multiple data modalities—video, audio, and sensor signals—to enrich the understanding of sports.
5.    Generative AI: Plausibly generating missing information, such as new viewpoints, to offer immersive spectator experiences (e.g., watching the game from the field via a VR headset).

SoccerNet: a scientific and community catalyst

Anthony Cioppa is also co-founder of SoccerNet, the largest open-source video analysis dataset for sports. Launched in 2018, SoccerNet has become a global benchmark, bringing together top research teams from academia and industry through its annual SoccerNet Challenges.

Major disruptions ahead thanks to Artificial Intelligence

The rise of AI promises to revolutionize how sports are captured, analyzed, refereed, and experienced. Computer vision is becoming capable of recognizing player movements without markers or GPS, estimating ball position and depth from simple video, generating statistics or summaries in natural language, and even answering questions about controversial referee decisions.