Student Expression Recognition in Smart Education Environment based on Convolutional Neural Network

Abstract

With the rapid construction of smart education and the continuous development of artificial intelligence technology, the advantages of assessment based on convolutional neural networks for teaching scenarios in smart learning environments are emerging. The practical applications tend to be more profound. We propose a convolutional neural network-based student expression recognition model and an assessment algorithm in the smart education environment to analyze the external learning characteristics of learners and realize the assessment and diagnosis of classroom learning effects. A student facial expression recognition dataset based on actual classroom scenarios is constructed, and teaching assessment is recognized by fusing classroom student expression states. Experiments are conducted on the publicly represented dataset and the self-built student expression recognition dataset to verify the effectiveness of the proposed student expression recognition model and the feasibility of the self-built student expression dataset. Finally, the proposed model achieved 89.31 % accuracy on the RAF -DB dataset and 80.59% on the self-built dataset. The results show that the proposed method performs well on both the public expression dataset and the self-built student expression dataset.