Research On Classroom Teaching Quality Evaluation Based On Expression Recognition And Bayesian Probability Model Statistical Analysis

The application of facial expression recognition in the classroom can allow teachers to grasp the students’ listening status and learning emotions in time.Because the current classroom teaching evaluation based on facial expression recognition is directly based on the classification results of the algorithm,and the results obtained directly are easily affected by sample characteristics,sampling methods and other factors.Therefore,the prior knowledge of facial expression classification can be a useful and necessary supplement,so that an evaluation model that is more in line with the real situation can be established.Therefore,this study takes the facial expressions of middle school classroom learners as the research object,integrates deep learning and Bayesian probability statistical analysis methods,establishes a deep neural network algorithm for real-time face detection and facial expression classification,and a standard library of classroom facial expressions Through the sequence description of classroom expression emotion index,an evaluation model based on normal distribution inference learning is established to improve the evaluation of classroom teaching quality.The main work done is as follows:(1)Analyze the distinctive facial expression characteristics of students in the classroom,and establish classification standards and standard libraries.Collect facial expression videos of classroom students,use multi-task convolutional neural network(MTCNN)for face detection and image segmentation,and select a better feature form to build a standard database.(2)According to the classification criteria,build a deep neural network with accurate and high classification efficiency,and generate a sentiment index based on the classification results.Design an algorithm that combines real-time multi-face detection and expression classification.First,test the classification efficiency on the existing open facial expression data set.The algorithm that improves the classification efficiency will be improved according to the classification standard of the self-built expression data set.Through training to form a stable expression recognition network;and according to Russell’s emotion ring theory,the sorted expression pictures are used to generate expression emotion index sequence.(3)Establish an evaluation model for classroom teaching quality and integrate teaching video data for empirical analysis.Using the learner’s expression and emotion index as the main indicator,the Bayesian probability model is used to perform a priori hypothesis and iterative analysis of the video data stream,and an approximate posterior distribution is obtained to achieve the description of the overall classroom learning state;Evaluate the model,learn the model parameters and compare and analyze the parameters according to the video recording of classroom teaching,and then complete the classroom teaching quality evaluation method for the model parameters;find the teaching problems in time according to the evaluation results,improve the teaching methods and strategies,and carry out accurate supervision to improve the education and teaching management Level.The experimental results show that the student’s emotional index conforms to the normal distribution,and the evaluation model established by the Bayesian statistical analysis method can understand the emotional state of the learner and the state of the teacher’s teaching level,thereby giving the teacher timely information feedback so that the teacher can carry out the teaching method The choice of teaching and the control of teaching progress promote the improvement of teaching quality.