Research On Classroom Quality Analysis Based On Facial Expression Recognition

Classroom is the main position of teachers’ teaching and students’ learning,so the analysis of classroom quality reflects the teaching level of a school and the applicability of teachers’ teaching to students.However,in today’s education industry,for the analysis of classroom quality,every school or every educational institution has uneven theoretical foundations and evaluation methods,and has never reached a unified and efficient standard.Therefore,in combination with facial expression recognition technology,it is a difficult problem that we urgently need to solve now to develop new low-cost or cost-controllable,high-precision and high-reliability classroom quality analysis.A new model based on facial expression recognition of video sequences is presented in this article to verify the effect of facial expression recognition in commonly used data sets,and it is used in the new classroom quality analysis system given in this article.The experimental results show that it can be present.The classroom teaching analysis at the stage provides a relatively reliable reference.The specific tasks are as follows:(1)For video sequences,a new expression recognition model DSTN-BiLSTM is proposed,which is divided into four steps: extract the expression space feature information;extract the dynamic feature information of the expression;fuse the first two features;The final expression recognition is carried out by combining the BiLSTM recurrent neural network.The feature information of expression space is extracted by using a deep CNN based on Inception,because it emphasizes "depth" and can extract more subtle and more critical expression feature information.The way to obtain the dynamic feature of expression is to use two uninterrupted expression images of the expression sequence as the input of the network,so that the network can obtain the dynamic information of short-term memory.The network chooses the shallower CNN,which contains two Conv Layer,after each Conv layer,the BN layer is used,whose function is to allow network training to converge to the local optimum faster.There are also two Max-Pooling layers,and finally an FC layer.Regarding the fusion of the two features,after experimental analysis,connection fusion is the best choice,because its recognition rate on CK+ is the highest.Finally,use it as the input of the BiLSTM network to obtain the dynamic change information of facial expressions over the entire time series,and then give the expression recognition results.(2)Compare and verify the recognition rate of the newly proposed facial expression recognition model.The new facial expression recognition model is divided into several sub-models,which are split and combined according to their structure.The experimental comparisons are carried out on the CK+,Oulu-CASIA,and MMI data sets.The experimental results show that the facial expression recognition model given in this article has a recognition rate All of the above are better,and comparing its recognition rate with other network models,it also verifies that the model in this article is reliable for classroom quality analysis.(3)Combining the facial expression recognition model given in this article,and comparing the facial expressions and facial expression characteristics of students,a new classroom quality analysis method is proposed.Expression recognition can output a confidence level for each type of learning emotion as the possibility of this expression.This possibility is used as the standard for grading students’ listening status,and then class quality classification is given according to this evaluation method.(4)Verify the reliability of the proposed classroom quality analysis design method through experiments.The author chooses a class from a high school for analysis,takes 20 minutes of time,and divides the video into 14,400 pictures at one frame interval,obtains10,000 valid pictures,and selects five students to track and score the design of this article Comparing and analyzing the scores with teachers,the results show that there is a strong correlation between the two,that is,the design scores for classroom quality based on facial expression recognition in this article are reasonable and reliable.