Towards real-time multiresolution face/head detection

Abstract

Reliable and real-time face detection is a basic ability for any Vision Based Interface. This paper combines and exploits the benefits of two different face detectors specialized each one in a specific context. The resulting system improves their respective individual performances by means of their cooperation, the integration of temporal coherence, persistence and explicit knowledge about the human face, achieving a robust and close to real-time multiresolution face detector.