Comparative performance evaluation of classifiers for Facial Expression Recognition

Towards building new, friendlier human-computer interaction and multimedia interactive services systems, we developed a image processing system which consists of the face detection module, which first determines automatically whether or not there are any faces in given images and, if so, returns the location and extent of each face and a facial expression classification module, which allow the classification of several facial expressions. In order to increase the accuracy of the facial expression classification module, we developed four different classifiers, namely:(1) Multilayer perceptrons, (2) Radial basis networks, (3) K-nearest neighbor classifiers and, (4) Support vector machines. In this paper we make an evaluation of performance of these classifiers versus the human's expression recognition performance for five expression: 'neutral', 'happy', 'surprised', 'angry' and 'disgusted'.