A Unified System for Processing News and Detecting Advertisements in TV Streams

Abstract

Media monitoring plays an important role in the success of companies by protecting their corporate, staff, and brand reputation. With the emergence of ICT, monitoring has emerged as an inevitable means of covering various fields of media: newspapers, online news, broadcast news (TV, radio), and social networks. Companies need a 360-degree view of media sources in near real-time that reports what's going on. In this paper, we proposed a unified system for processing news and detecting advertisements in TV streams. The system architecture is completely described, and the automatic processing has been detailed. Our contributions include a TV stream segmentation method, a news topics segmentation method, and an advertisements classification model. These methods have been evaluated and given promising results. However, more improvements can be made in the future.

Keywords

• Multimedia processing
• parallel processing
• deep learning
• TV stream analysis
• news identification
• advertisement extraction
• media monitoring

References

1. Kompatsiaris, Y., Merialdo, B. and Lian, S.: TV Content Analysis: Techniques and Applications. In: Boca Raton, FL: CRC Press. (2012).
2. Minaee, S., Bouazizi, I., Kolan, P. and Najafzadeh, H.: Ad-Net: Audio-Visual Convolutional Neural Network for Advertisement Detection In Videos. In: arXiv preprint arXiv:1806.08612. (2018).