Target Audience Response Analysis in Out-of-home Advertising Using Computer Vision

In this paper we aim to analyze the response of the target audience to out-of-home advertising panels mounted in display windows. We track people passing in front of the panels and, if they approach the panel, looking directly at it, we analyze their facial movements, to see whether the advert appeals to them or not. The approach is based on video data gathering using a static IP camera, storage and analysis, aiming to deliver real-time results. We use neural networks and support vector machines to determine facial microexpression. Results are correlated with information about the adverts displayed to be of use to advertising providers.