In computer vision, person re-identification is the task of recognizing and assigning an identity to a person previously observed, typically in a different camera view. Person re-ID techniques find use in various types of applications, e.g. surveillance systems, robotics, security and forensic applications. This thesis addresses the possibility of using person re-identification in a sports setting and compares it with the usual task of person re-identification in the public spaces. Emphasis is placed on deep learning and different approaches to re-ID in this field (classification and verification methods). This work describes several state-of-the-art person re-ID techniques and evaluates their effectiveness and efficiency on two types of data: 1) Publicly available person re-ID datasets, such as Market-1501 or CUHK03 dataset. 2) Private sports dataset consisting of football player images generated using footage from the multi-camera TRACAB player tracking system. This work also discusses the differences between both types of data and specific ways to employ re-ID in football.