Automatic analysis of facial expressions: The state of the art

Humans detect and interpret faces and facial expressions in a scene with li ttle or no effort. Still, development of an automated system that accomplis hes this task is rather difficult. There are several related problems: dete ction of an image segment as a face, extraction of the facial expression in formation. and classification of the expression (e.g., in emotion categorie s). A system that performs these operations accurately and in real time wou ld form a big step in achieving a human-like interaction between man and ma chine. This paper surveys the past work in solving these problems. The capa bility of the human Visual system with respect to these problems is discuss ed. too, it is meant to serve as an ultimate goal and a guide for determini ng recommendations for development of an automatic facial expression analyz er.