Project description

Project Description
Soccer has a large impact in society and is a multi-billion-dollar industry, capping $91 billion of annual market revenue, with $28.7 billion from the European Soccer market alone. Recent advances in computer vision helped to provide automated tools to understand and analyze broadcast games, localize the field and its lines, detect players, their motion and pose and track the ball position. In this thesis, we will go beyond these approaches that remain at the perception level and will develop novel algorithms for learning the semantics to automatically achieve a deeper understanding of the game. We will consider from shallow event discovery (e.g. goal, fault, substitution, corner kick) to high-level comprehension of player movements, team formations and tactics.
For this purpose, we will leverage on three main ingredients: 1) The specific knowledge of soccer by part of the members of the company, either professional coaches or players; 2) The ability to capture and annotate vast amounts of data of real matches, both with positional and tactical labels; 3) The use of deep learning algorithms for data-driven modeling and knowledge discovery.

The outcome of this thesis will be integrated in the company product, an intelligent tool for team and coach support, but it can also be applied to other related-applications, like in the development of automatic broadcasting tools, or video summarization and personalization.

The project will be carried in cooperation with the following:
? – Optima Sports System, in Sant Joan Despí.
? – Institute of Robotics and Industrial Informatics at the Polytechnic University of Catalonia.

The scientific advisor at IRI will be Dr. Francesc Moreno Noguer.
The responsible for part of the company will be Dr. Maurici López Felip.