A three-dimensional anisotropic diffusion equation-based video recognition model for classroom concentration evaluation in English language teaching (Q2064675)

Summary: In this paper, a three-dimensional anisotropic diffusion equation is used to conduct an in-depth study and analysis of students' concentration in video recognition in English teaching classrooms. A multifeature fusion face live detection method based on diffusion model extracts Diffusion Kernel (DK) features and depth features from diffusion-processed face images, respectively. DK features provide a nonlinear description of the correlation between successive face images and express face image sequences in the temporal dimension; depth features are extracted by a pretrained depth neural network model that can express the complex nonlinear mapping relationships of images and reflect the more abstract implicit information inside face images. To improve the effectiveness of the face image features, the extracted DK features and depth features are fused using a multicore learning method to obtain the best combination and the corresponding weights. The two features complement each other, and the fused features are more discriminative, which provides a strong basis for the live determination of face images. Experiments show that the method has excellent performance and can effectively discriminate the live nature of faces in images and resist forged face attacks. Based on the above face detection and expression recognition algorithms, the classroom concentration analysis system based on expression recognition is designed to achieve real-time acquisition and processing of classroom images, complete student classroom attendance records using face detection and face recognition methods, and analyze students' concentration from the face integrity and facial expression of students facing the blackboard by combining face detection and expression recognition to visualize and display students' classroom data for teachers, students, and parents with more data support and help.