In this work, we present AdsISee, a real-life application for the detection and tracking of advertisements in broadcasts of sport events for supporting business analysts in the task of sponsorship evaluation and reporting. Our approach is based on different combinations of several techniques for object detection and tracking in images. In contrast to other works which use the technology of neural networks, we use alternative solutions to detect advertisements based on provided templates and without any training period. Hereby we can built an application which can be executed on standard hardware by still providing a feasible performance. Furthermore, our evaluations show that we can achieve comparable results against existing approaches, which are based on neural networks, for sponsorship evaluation.
Dr. Andreas Weiler (Zurich University of Applied Sciences School of Engineering)