Chameleo-AD: Real-time targeted interactive advertisement for digital signage

In this paper we present an integrated approach to interactive, real-time targeted multimedia advertisement for digital signage, based on visual audience analysis. Digital Signage and Digital-out-of-Home (DooH) systems nowadays still employ static content scheduling, and only few of them offer basic interaction capabilities, usually based on touch screens. In our approach we use visual sensors attached to a display to measure anonymous properties of the present audience in real-time and to enable touch-less interaction. A combined visual classification scheme analyses facial and full-body image data and integrates results in a long-term observation of individuals. Based on a measured audience profile, considering gender and age, an adaptive media player queries targeted rich media content from a database and presents it to the audience. Content selection is based on an audience state model which considers the spatio-temporal characteristics of the audience (distance, dwell-time) and a content model representing the relations of content and audience. The system automatically offers free-hand gesture interaction to the user, once a certain level of attention is reached. To our knowledge the presented system is the first approach to integrating long-term observation based on facial/full-body visual data, content adaptation and gesture interaction.