Research On Mining Intelligent Teaching State In Classroom Teaching Video

With the application of information network technology in the field of intelligent education,a large number of big data resources in the field of education have been accumulated,including text,image,video and other data in the field of education.These huge data contain rich and valuable information.The advent of the era of educational big data has brought new ideas and academic research hotspots to intelligent teaching.As people pay more and more attention to education,the traditional classroom teaching has been unable to meet people’s needs.Traditional classroom teaching has some limitations in information transmission and feedback between teachers and students due to the limited energy of teachers and too many students.Therefore,using classroom teaching video data,combined with computer vision and video intelligent processing technology,to detect and identify the state of students in the classroom,and build an intelligent teaching state mining system,so that teachers can timely understand the changes of students’ classroom state and students’ attention,which has important research value and application prospect.The work completed in this thesis mainly includes the following four aspects:(1)Aiming at the problem that it is impossible to obtain effective student expression features in classroom teaching scene,a student expression feature learning algorithm based on deep attention network is proposed.Through data enhancement and student face detection technology to process the classroom teaching video data,we can extract the student face position efficiently.Based on the deep learning method,this thesis constructs the student expression feature learning model of classroom teaching video based on the deep attention network.The feature fusion method is used to enhance the nonlinear expression of students’expression feature learning model through the fusion of high-level features and low-level features;By fusing multi-branch network and adaptive weight allocation mechanism,a deep attention convolution neural network is constructed to learn multi-perspective features such as local features,occluded features and global features of students’expression.Different weights are allocated to different areas of the face through the attention mechanism to solve the impact of students’ face occlusion and the network paying too much attention to non-important areas on expression recognition,and filter redundant information.Realize the effective learning of students’ facial expression characteristics in classroom scenes.The experimental results show that this algorithm has the best feature extraction effect compared with the frontier algorithm in recent years.(2)Aiming at the problems of single data and general methods not suitable for classroom scenes in student expression recognition,a student expression recognition method based on deep learning is proposed.Using the deep learning method,an expression recognition model based on generative adversarial network is constructed,and the expression database is edited with different features.The expression data of different emotions are generated by the generator to enhance the data set;The distortion of identity data is reduced by controlling the expression in the editing process;The problem of artifact in expression editing is solved by cascaded network.On this basis,combined with the deep attention convolution neural network,the student expression recognition model is constructed to reduce the impact of head posture,identity deviation and single data on the student expression recognition,and realize the accurate and efficient acquisition of the student expression information in the classroom scene.Experimental results show that this algorithm is better than the frontier algorithm in recent years.(3)Aiming at the problems that classroom evaluation methods consider less features and are not objective enough,an intelligent teaching state mining algorithm based on multi-feature fusion is proposed.Through the combination of classroom nature,students’ head posture and students’expression,judge the state of each student and evaluate the overall classroom state.The intelligent teaching state evaluation model is constructed by integrating the characteristics of classroom nature,students’ expression and students’ head posture,and the students’ head posture is obtained by behavior recognition algorithm;The student expression is obtained through the student expression recognition algorithm;Judge the student’s state in combination with the nature of the classroom,the student’s expression and head posture,score the student’s classroom performance through the scoring algorithm,build a scientific and intelligent scoring method,deeply mine the student’s state information in the classroom video,realize the mapping mechanism from the student’s expression information to the listening state,and mine the overall teaching state.(4)Realize the intelligent teaching state mining system in classroom teaching video,and build a system with excellent performance such as feasibility,robustness and accuracy.The system is divided into three functional modules:Student expression feature learning module,student expression recognition module based on deep learning,and intelligent teaching state mining module based on multi feature fusion.Realize the intelligent teaching state mining function integrating the nature of the classroom,students’ head posture and students’ expression,and complete the mining of each student and the overall classroom state.The experimental and test results show that the system realizes the accurate and objective evaluation of the classroom state,designs an easy to expand algorithm interface and user-friendly interactive interface,and comprehensively displays the operation results of each module function.