Detect and classify events in football video

In this project, the student will try to automatically classify common events in football video such as goals, corners, cards.
Master

Major football events are manually tagged today in the Norwegian and Swedish top leagues. In this project, the student will develop a system for automatically tagging events based on football video and player positional data, when available. The output can be used directly to extract highlight clips, or as input to human operators. A large event library of tagged football data is available for training.

Goal

Develop and evaluate a machine learning system for detecting football events. Academic publication of results.

Learning outcome

Practical experience in using machine learning techniques for real video events.

Qualifications

Understanding of neural nets and other machine learning techniques. INF5860 recommended.

Supervisors

Collaboration partners

ForzaSys AS