Piracy detection in online soccer streaming with video content inspection: an application to the Portuguese market
Diogo Pontes, Claudino Costa, Ricardo Gomes Faria, Pedro Fidalgo, José Henrique Brito · 2023
This study presents the earlier steps in the detection of pirate soccer streams using computer vision. The goal is to provide broadcasters, advertisers, and sports organizations with a powerful tool to protect their intellectual property and revenue. The focus of this study is the Portuguese market, therefore the majority of work relates to the country’s major soccer competitions and TV broadcasters. The study employs a DenseNet201 frame classifier and YOLOv5 object detector models trained with transfer learning using a dataset of manually annotated images extracted from publicly available sources. The study demonstrates that shot classification and broadcaster logo detection are highly effective in identifying visual elements in pirate streams, allowing the detection of unauthorized use of copyrighted content. The results show that the proposed method can be powerful solution for the growing problem of pirate soccer streams, achieving high accuracy in real-time applications, enabling broadcasters to take immediate action to protect their content and revenue.