Skip to Main Content
Facial expression recognition has become one important research direction in HCI area, because facial expression can effectively reflect the inner emotion of people. In some areas (e.g. remote education), expression recognition system, as an auxiliary tool, is used to analyze mental states of students and plays an important role. In this paper, we developed a student-state recognition system based on facial expression. We improved local binary pattern (LBP) into an algorithm that based on specific area to extract facial features, and presented features with LBP histograms of expression image. Finally, support vector machine and nearest neighbor classification are combined to classify expressions, then classify the mental states of students based on expression. We conducted experiments with samples from a Japanese female facial expression database. Our study obtained an average facial expression recognition rate of 71.35%. High accuracy and real-time make the system applied in classroom well to recognize mental states of students.