Automatic facial expression analysis of students in teaching environments

Abstract

Based on students’ facial expressions, the teacher in class can know the students’ comprehension of the lecture, which has been a standard of teaching effect evaluation. In order to solve the problem of high cost and low efficiency caused by employing human analysts to observe classroom teaching effect, in this paper we present a novel and high-efficiency prototype system, that automatically analyzes students’ expressions. The fusion feature called Uniform Local Gabor Binary Pattern Histogram Sequence (ULGBPHS) is employed in the system. Using K-nearest neighbor (KNN) classifier, we obtain an average recognition rate of 79% on students’ expressions database with five types of expressions. The experiment shows that the proposed system is feasible, and is able to improve the efficiency of teaching evaluation.