Research And Implementation Of Student Attention Judgment Model Based On Online Learning Video Content

Under the background of developing "Internet+" education,online education has gradually entered the public's perspective.This education mode has changed the traditional classroom teaching mode,breaking the barriers of space and time of teaching,which can liberate teachers and students.Online education has self-learning ability and rich educational resources.However,it also brings certain disadvantages.The temporal and spatial separation of teachers and students in online education leads to the lack of classroom emotional analysis.Teachers cannot interact with students in real time,and they cannot obtain students' cognitive status,classroom concentration,and classroom effects in a timely manner.Therefore,the acquisition of emotional state of students in online classes are currently urgently needed to solve.The concentration of students directly represents the degree of student's investment in the course,and it's a important significance of course evaluation.The main purpose of emotion recognition in online education is to monitor and feedback the state of learners in real time.However,the expression of facial movements is complex and changeable,so it is very difficult to detect the concentration.The purpose of this thesis is to evaluate the performance of students in online classroom through deep learning network,including face detection,expression recognition,research on the division of students' classroom state and focus judgment,and design and implement a video content-based student classroom concentration judgment system.During the process,students' status is tracked and identified,feedback,and classroom evaluations are conducted.The main research contents are as follows:First,based on the research on the existing database recording process,i formulated a detailed expression database construction plan,and recorded a set of student fatigue expression data sets in the classroom context.Taking the student group as the test object,it is different from the existing adult expression data set,and the data is focused on the collection of fatigue expressions,which can be better transferred to the classroom emotion analysis model.Secondly,research the theories of face detection and expression recognition.Through the analysis of expression data,use MTCNN network for face detection during the expression recognition process,and extract the feature points of the eyes and mouth.Using CNN+RNN network to achieve a complete fatigue expression detection model.Thirdly,this thesis divides the classroom status of students,designs a multi-dimensional classroom status evaluation system,realizes the division of classroom concentration(normal,fatigue,leaving),then designs and implements a system for identifying students' concentration based on video content.The system integrates face detection and fatigue detection models to judge student behaviors.By predicting the positions of facial feature points,the image search time is reduced and the real-time performance of the algorithm is improved.Finally,Using the facial expression data during online courses to verify the feasibility of the system.The test results show that this method can recognize the expression categories well and meet the requirements of online education.