Optimised meeting recording and annotation using real-time video analysis

Abstract

The research detailed in this paper represents the confluence of various vision technologies to provide a powerful, real-time tool for human behavioural analysis. Gesture recognition algorithms are amalgamated with a robust multi-person tracker based on particle filtering to monitor the position and orientation of multiple people, and moreover to understand their focus of attention and gesticular activity. Additionally, an integrated virtual video director is demonstrated that can automatically control active cameras to produce an optimum record of visual events in real-time. © 2008 Springer-Verlag Berlin Heidelberg.