| With the increasing progress of science and technology,how to prefer the digital and intelligent advance of modern education classroom management has become an urgent problem to be solved.Classroom attention of students reflects the learner’s mastery of the knowledge taught by the teacher,so that both learners and teachers can change the learning and teaching project on the grounds of the learning condition,and it also influences the quality of schoolroom teaching.In recent years,many information technologies have been used to analyze students’ attention in class.The way of collecting students’ information through video technology has the advantages of good real-time performance and large coverage.To analyze students’ attention in class through video,this method has problems such as illumination and occlusion,which reduce the accuracy of face detection.At present,most researchers focus on the single characteristics of students’ attention in class,such as expression or bowing rate in class.It is difficult to accurately reflect students’ attention state in class.Based on the said questions,this essay in-depth analysis and investigation on students’ face detection in classroom scenes,classroom expressions,head posture,fatigue status and students’ sitting posture inclination,as the main criteria for judging students’ attention in class.The research content is as follows:1.This paper introduces the research progress and related technology of students’ attention concentration determination at home and abroad,expounds the technical research situation and current situation,and made a detailed research plan for the detection of student concentration.2.The improved MTCNN（Multi-task Cascaded Convolutional Networks）face detection algorithm combined with SSH（Single Stage Headless Face Detector）face detection algorithm is applied to face detection of multiple students in class and the algorithm is used for experiments to test the excellence of the algorithm.3.According to the classroom background,we redefine 6 kinds of classroom expressions,and use the improved Alexnet network to inspect students’ classroom expressions.Obtain the key points of the eyes and mouth according to the landmark technology,and judge the fatigue state of the students based on the principle of PERCLOS（Percentage of Eye Iid CIosure Over The Pupi I Over Time）.The head pose of students is detected based on the Head Pose Estimation algorithm,and the key points of human bones were detected by Openpose algorithm,and the key points were converted into coordinate data.The body inclination is described by the angle formed by the connection between the joint points on the left and right shoulders of the human body and the horizontal direction.Machine recognition and manual detection are performed respectively.Contrast experiments to test the accuracy of the algorithm.4.Facial expression,body tilt,mouth opening,eye closure,and head orientation are used as the evaluation features of attention concentration.The students’ attention concentration was scored by fuzzy judgment method,and the accuracy of experimental results was verified and analyzed according to the actual situation. |
