Research On Evaluation Model Of Classroom Attention Of Students Based On Face Recognition Technology

Education informatization has gradually become a key topic of national concern.Improving teaching efficiency through education informatization is a major goal of curriculum reform in education.Classroom teaching is a space for teachers and students to interact and interact,and it is the main way for teachers to guide students to learn and students to explore knowledge.In classroom teaching,students'classroom performance is an important part.The student's classroom performance includes not only the student's lectures and group discussions,but also the teacher's concentration during the class.Whether the student's performance is concentrated is often reflected in the student's facial expression and classroom head-up rate,and these aspects also reflect Students' efficiency in class and their ability to accept new knowledge.However,after literature investigation and research,it is found that the existing methods of assessment of students' classroom concentration still have the following problems:first,in a one-to-many teaching classroom environment,it is easy to cause incomplete evaluation and not timely;second,the evaluation method The single factor cannot fully consider the factors that affect students' classroom concentration;third,a reasonable joint evaluation model has not been established between different evaluation indicators,and the influence of a certain indicator is relatively large,which cannot effectively reduce the error;fourth,expression recognition The recognition effect of the algorithm is not ideal,resulting in a certain error in the evaluation of concentration.In response to the above problems,this study improved two neural network algorithms for facial expression recognition,respectively,an improved model based on VGG network and an improved model based on separable convolutional network.And using the improved VGG network model to establish a joint concentration evaluation model based on expression and head-up rate,the purpose is to make a more comprehensive,objective and scientific analysis of the students' classroom concentration through the joint evaluation model.Compared with other algorithms,the improved algorithm based on the VGG network achieved the best results on the FER2013 data set,and greatly saved the time of model training,which greatly improved the efficiency.The improved model based on separable convolution has the second highest accuracy on the test set,but its convergence is better than the former.In the evaluation of students' classroom concentration,this paper first uses the face detection algorithm in face recognition technology to detect faces and intercept expression data,and calculates the head-up rate;then uses the improved VGG model for facial expression recognition and gives corresponding The weight of the expression calculates the expression score;finally,the head-up rate calculated at the same moment is multiplied by the expression score as the final concentration score.By conducting experiments in actual classrooms and analyzing the results of the experiments,draw corresponding conclusions and provide teaching suggestions for teachers.For the reliability of the model,this paper verifies the model through practice tests,teacher questions,and interviews with students and teachers.The results show that the joint evaluation method based on expression and head-up rate proposed in this paper has high accuracy and reliability.