Real-Time Face Tracking for Audience Engagement

Abstract

– Audience measurement is an application area where face tracking and analysis could automatically provide reliable answers about how consumers and users behave spontaneously in real environments. The solutions can detect age, gender, size of the audience, distance from a display, dwell times and more. By knowing audience emotional reactions and engagement towards the products, content and campaigns, one can offer audience an improved experience and customized content addressing their interests and behaviour. In this paper, we present a pose-invariant face tracking method to automatically monitor audience's attention time in real-time. The current system can handle 3D pose variation up to ±55 in yaw and ±30 in pitch angles.