Use of machine learning (ML) for image and video analyses would often include face recognition, personalization and recommendations. An emerging trend is the application of AI technology for TV advertising. In this paper, we present the unique challenges in applying machine learning to carrier-class video advertising. We focus the discussion on a specific use case that is common to all ad supported TV services.
The selected use case is Ad Ingest Quality Control (QC). In the United States, TV commercials are subjected to various rules and regulations. Ads containing specific content (e.g. Alcohol, firearms) are barred from airing during certain TV programs. Identifying these categories may pose a challenge, as off the-shelf machine learning products are more oriented towards facial recognition. That is to be expected perhaps, as the video ML products were primarily intended for surveillance and sports applications.
However, our research indicates that by judiciously combing metadata from multiple data streams, machine learning analysis results can be improved.
The intent of the paper is to outline the results and recommendations of a proof-of-concept study that will be helpful to the carrier-class video services community.