Face Tracking Based Advertisement Effect Evaluation

Advertisements in public area are important for propagandizing commodities, but we have not got effective solutions to automatically analyze its effect. This paper presented a digital video processing-based evaluation system, which captured the scene in front of the advertisement, then performed statistics on passenger´s number, especially the ones looking to the advertisements, and the watching time intervals and patterns. With the above parameters, one could estimate the effect of the advertisements. Face detection technique is based on AdaBoost classifiers, face tracking with particle filters. The detection and the tracking are supplementary and could improve the tracking reliability. The experiments proved the effectiveness.