In this paper, a real-time facial expression recognition system is proposed to classify images of human facial expressions into a two dimensional model of emotion. The system is comprised of a facial feature detection system that uses constrained local models to locate the features, and a facial expression recognition system that utilizes multi-class support vector machines to classify the facial expressions. The outputs of the system are values of pleasure and arousal associated with an input image. Classifications rates achieved were 76% for the pleasure dimension and 62% for the arousal dimension. An analysis on facial action parameters also revealed that some parameters were more effective at determining pleasure and arousal values than others.