Emotion recognition method using facial expressions and situation

This paper tries to propose a method to recognize emotion of a user by systems that interact with human users. The goal of the method is achieving the same recognition results with a particular person. The main characteristic of the method is taking temporal sequence of the recognition result into account in its recognition process. The characteristic is derived from knowledge from psychology. The method uses a sequence of facial expressions of persons to estimate instantaneous emotion for a certain moment and uses the recognition result to recognize the emotional states by taking the emotional context into account. Experiments with three subjects as the particular persons who recognize the emotions of ten persons in medical interviews with a virtual medical doctor system, show that the emotion recognition using the previous recognition results outperforms that of only from the sequence of facial expressions.