Group Activity Recognition System in Football Videos

This research will investigate a group activity recognition system that detects and recognises high-level group activity in football matches.

Objectives

• Identifying suitable datasets for the video understanding model by hand collection or using an existing dataset.
• A video understanding model should be researched and proposed to achieve the goal of group activity recognition. The methods used by the two streams (Figure and Ground) will be investigated.
• The proposed model should be evaluated using appropriate metrics and compared with state-of-the-art methods.
• Expansion of approach to use a self-supervised method to move beyond football understanding.

Supervisors

• Dr Andrew Gilbert
• Dr Femi Adeyemi-Ejeye
• Professor Shamika Almeida

Open to UK and international candidates.

Applicants are expected to hold a first or upper-second class degree in a relevant discipline (or equivalent overseas qualification), or a lower second plus a good Masters degree (distinction normally required).

IELTS Academic: 6.5 or above (or equivalent) with 6.0 in each individual category.