Lightweight Solution for English Course Quality Evaluation Supported by Edge Intelligence

ABSTRACT With the development of artificial intelligence and computer vision technologies, smart education supported by artificial intelligence and computer vision has gradually replaced the traditional classroom teaching mode, such as English courses. However, the existing studies about smart education mainly focus on student classroom behavior recognition and neglect quality evaluation of teaching and learning in the classroom. On the other hand, it still lacks a lightweight solution for smart education in the network environment with limited resources. This paper establishes a lightweight English course quality evaluation scheme to tackle the above issues. First, a camera is used to collect the information of class; then, a lightweight network, which consists of a MobileNet v4 Small and a YOLO network, is deployed in the edge computing device to automatically process the information from the camera for attendance rate estimation and students' behavior recognition; lastly, the server in the remote control room collects all the information from classrooms to implement course quality evaluation. The experimental and simulated results show that student detection and student behavior recognition can meet the requirements in the real application.