Research on classroom expression recognition based on deep circular convolution self-encoding network

The detection and recognition of facial expressions has been widely used in various fields of social life in recent years. When students are in class, it is very useful to strengthen the monitoring and management of students during class by using methods such as expression recognition. For student expression recognition, we propose a cycle convolution self-coding deep learning network. This network combines the ideas of convolutional self-encoding networks and cyclic consistency. Our experiments were performed on the classroom student expression recognition dataset, and the recognition accuracy reached 98.12%.