Intelligent evaluation system and method for classroom teaching

Through classroom recording videos and deep learning technology, students' images and bone point information are extracted, and a comprehensive evaluation model is constructed, which solves the problems of lag and unreality of traditional classroom evaluation methods, and achieves more scientific and flexible classroom quality scoring.

CN119067504APending Publication Date: 2024-12-03SHANGHAI FANGLUEMENKOU EDUCATION TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411239205.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-05
Publication Date
2024-12-03

AI Technical Summary

Technical Problem

Traditional classroom teaching evaluation methods lack scientificity and flexibility, and cannot effectively reflect the dynamic changes in students' classrooms, resulting in lagging and untrue evaluation results.

Method used

By obtaining classroom recorded videos, extracting student image information, using YOLOV3 and OpenPose algorithms to detect face keyframes and bone point information, combining facial expressions and posture characteristics, a comprehensive evaluation model is constructed, and classroom quality scores are output.

Benefits of technology

It improves the scientificity, flexibility and authenticity of classroom quality evaluation and can dynamically reflect students' classroom status.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119067504A_ABST
    Figure CN119067504A_ABST
Patent Text Reader

Abstract

The invention relates to an intelligent evaluation system and method for classroom teaching, and belongs to the technical field of intelligent teaching. The method comprises the steps of obtaining a target face key frame through processing according to student image information, obtaining classroom attendance information according to the target face key frame, obtaining facial two-dimensional image information through processing according to the student image information through a face detection algorithm, obtaining facial expression features according to the facial two-dimensional image information, and obtaining classroom attendance information according to the facial expression features. According to student image information, key skeleton point information of a human body is detected through an OpenPose algorithm, a joint heat map is obtained according to the key skeleton point information, posture features are obtained through feature extraction according to the joint heat map, and a comprehensive evaluation model is constructed according to an expression feature evaluation matrix and a posture feature evaluation matrix in combination with classroom attendance features. The classroom quality score is output according to the comprehensive evaluation model, the classroom quality is scored according to the classroom state of the student, and the scientificity, flexibility and authenticity of classroom quality evaluation are improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of smart teaching, and specifically relates to an intelligent evaluation system and method for classroom teaching. Background Art

[0002] With the development of society, education has received more and more attention and importance from people. The quality of school education affects people's growth and, to a certain extent, the progress of society, and has attracted much attention from the society. Classroom teaching is the main form of school education. The quality of classroom teaching directly affects students' learning effect and the quality of talent training. Therefore, teaching evaluation of classroom teaching is an important part of teaching quality assurance, which has attracted much attention from educators and has become a hot issue in the school's teaching reform and teaching management. The traditional classroom teaching evaluation method includes teachers adjusting the teaching process and teaching methods by issuing tests, attendance and classroom performance. This is an inefficient and outdated evaluation method. At the same time, schools and colleges at all levels also monitor and manage courses. Generally, teaching supervisors are arranged to attend classes or patrol the classroom from time to time to evaluate teachers' classroom teaching. The traditional classroom quality evaluation standards are single and lack scientificity and flexibility. The traditional evaluation standards are more for teacher evaluation rather than classroom evaluation. In addition, the student status in the classroom is ignored, resulting in students' tasks only to follow and obey. Teachers can only evaluate students' knowledge acceptance based on the published test scores, and there is a problem of information lag. On the other hand, the traditional classroom teaching evaluation is more based on quantitative and qualitative evaluation, which pays too much attention to subjective data judgment and ignores the dynamic changes of students in the classroom, resulting in the inability of classroom teaching evaluation to reflect the true situation of teaching quality. Summary of the invention

[0003] In order to solve the above problems existing in the prior art, the present invention provides an intelligent evaluation system and method for classroom teaching, which obtains classroom video recordings, calls a database to obtain class member information, extracts student image information according to the classroom video recordings, obtains target face key frames according to the student image information through processing, obtains classroom attendance information according to the target face key frames, obtains facial two-dimensional image information according to the student image information through face detection algorithm, obtains facial expression features according to the facial two-dimensional image information, detects key bone point information of the human body according to the student image information through OpenPose algorithm, obtains joint heat map according to the key bone point information, and obtains the key bone point information according to the key bone point information. The posture features are obtained by feature extraction from the joint heat map, and the classroom attendance features are obtained by feature extraction based on the classroom attendance information. The expression feature evaluation matrix is ​​constructed based on the facial expression features and the expression feature evaluation matrix. The posture feature evaluation matrix is ​​constructed based on the posture features and the posture feature evaluation indicators. A comprehensive evaluation model is constructed based on the expression feature evaluation matrix and the posture feature evaluation matrix combined with the classroom attendance features. The classroom quality score is output based on the comprehensive evaluation model, which realizes the scoring of classroom quality according to the students' classroom status. The classroom quality score is obtained by fusion processing based on the classroom attendance information, facial expression features and posture features, which improves the scientificity, flexibility and authenticity of the classroom quality evaluation.

[0004] The purpose of the present invention can be achieved through the following technical solutions: An intelligent evaluation method for classroom teaching, specifically comprising the following steps: S1: Get the classroom video; S2: Call the database to obtain class member information, extract student image information according to the classroom video recording, process the student image information through the YOLOV3 target detection method to obtain a target face frame set, obtain a target face key frame according to the target face frame set through the target tracking algorithm, and obtain class attendance information according to the target face key frame, the class member information includes student ID and class member face image information, and the class attendance information includes information of students who have attended and information of students who have not attended; S3: obtaining two-dimensional facial image information through a face detection algorithm according to the student image information, obtaining a denoised facial image through image processing according to the two-dimensional facial image information, obtaining three-dimensional facial image information through deep learning according to the denoised facial image, obtaining a first facial feature and a second facial feature through feature extraction according to the three-dimensional facial image information, and obtaining a facial expression feature through feature fusion model processing according to the first facial feature and the second facial feature; S4: Detect key bone point information of the human body through the OpenPose algorithm according to the student image information, obtain denoised key bone point information through normalization processing according to the key bone point information, obtain a joint heat map through Gaussian mapping according to the denoised key bone point information, and obtain posture features through feature extraction according to the joint heat map; S5: Obtain expression feature evaluation indicators and posture feature evaluation indicators, obtain classroom attendance features through feature extraction based on the classroom attendance information, construct an expression feature evaluation matrix based on the facial expression features and the expression feature evaluation matrix, construct a posture feature evaluation matrix based on the posture features and the posture feature evaluation indicators, construct a comprehensive evaluation model based on the expression feature evaluation matrix and the posture feature evaluation matrix combined with the classroom attendance features, and output a classroom quality score based on the comprehensive evaluation model.

[0005] Preferably, the step S2 specifically includes training a face recognition network model according to the face image information of the class members, processing the target face key frame through the face recognition network model to obtain a target face matching degree, when the target face matching degree is greater than or equal to a preset threshold, obtaining the attendance student information according to the corresponding student ID, and filtering out the student ID of the absent student according to the attendance student information and the class member information to obtain the absent student information; Preferably, the step S3 specifically includes the following steps: S301: acquiring an expression image sequence according to the denoised facial image information, obtaining a first standard frame and a second standard frame according to the expression image sequence, obtaining facial key point coordinates according to the first standard frame, obtaining a facial key point matrix according to the facial key point coordinates through a convolutional neural network, and obtaining the facial three-dimensional image information according to the facial key point matrix; S302: Obtain the first facial features through feature extraction according to the facial key point matrix, obtain expression image information through image segmentation according to the second standard frame, obtain a valid expression image through grayscale normalization processing according to the expression image information, obtain the second facial features by performing feature extraction on the valid expression image, and obtain the facial expression features by processing the first facial features and the second facial features through the feature fusion model.

[0006] Preferably, the step S302 specifically includes expressing the feature fusion model as: , where O represents the facial expression feature, O1 represents the first facial feature, O2 represents the second facial feature, and α and β represent weighting coefficients.

[0007] Preferably, the step S4 specifically includes obtaining a posture feature triplet by feature extraction according to the target person key frame, and obtaining a joint heat map by Gaussian mapping according to the posture feature triplet, and the formula is: , where J is the joint heat map, is the posture feature triplet, is the Gaussian map contrast parameter.

[0008] Preferably, the S5 specifically includes the expression feature evaluation matrix represented as: , where R1 represents the expression feature evaluation matrix, represents the evaluation result of the i-th expression feature evaluation index on the j-th facial expression feature, and the posture feature evaluation matrix is ​​expressed as: , where T1 represents the posture feature evaluation matrix, Represents the evaluation result of the mth posture feature evaluation index on the nth posture feature.

[0009] Preferably, the step S5 further comprises constructing the comprehensive evaluation model according to the expression feature evaluation matrix and the posture feature evaluation matrix in combination with the classroom attendance feature, and the comprehensive evaluation model is expressed as: , where w1, w2 and w3 represent weight vectors, R1 represents the expression feature evaluation matrix, T1 represents the posture feature evaluation matrix, represents the class attendance characteristics, Represents the quality rating of the class.

[0010] An intelligent evaluation system for classroom teaching, including the following modules: Video acquisition module, used to acquire classroom recording videos; A classroom attendance module is used to call a database to obtain class member information, extract student image information based on the classroom video, obtain a target face frame set based on the student image information by using a YOLOV3 target detection method, obtain a target face key frame based on the target face frame set by using a target tracking algorithm, and obtain classroom attendance information based on the target face key frame, wherein the class member information includes student ID and class member face image information, and the classroom attendance information includes information on students who have attended and information on students who have not attended; A facial analysis module, configured to obtain two-dimensional facial image information by processing the student image information through a face detection algorithm, obtain a denoised facial image by image processing based on the two-dimensional facial image information, obtain three-dimensional facial image information by deep learning based on the denoised facial image, obtain a first facial feature and a second facial feature by feature extraction based on the three-dimensional facial image information, and obtain a facial expression feature by processing the first facial feature and the second facial feature through a feature fusion model; A posture analysis module is used to detect key bone point information of the human body through the OpenPose algorithm according to the student image information, obtain denoised key bone point information through normalization processing according to the key bone point information, obtain a joint heat map through Gaussian mapping according to the denoised key bone point information, and obtain posture features through feature extraction according to the joint heat map; A classroom evaluation module is used to obtain expression feature evaluation indicators and posture feature evaluation indicators, obtain classroom attendance features through feature extraction based on the classroom attendance information, construct an expression feature evaluation matrix based on the facial expression features and the expression feature evaluation matrix, construct a posture feature evaluation matrix based on the posture features and the posture feature evaluation indicators, construct a comprehensive evaluation model based on the expression feature evaluation matrix and the posture feature evaluation matrix combined with the classroom attendance features, and output a classroom quality score based on the comprehensive evaluation model.

[0011] The beneficial effects of the present invention are: 1. Extract student image information according to the classroom video recording, obtain target face frame set by YOLOV3 target detection method according to the student image information, obtain target face key frame according to the target face frame set by target tracking algorithm, obtain classroom attendance information according to the target face key frame, realize classroom attendance, and the obtained classroom attendance information is conducive to classroom quality evaluation; 2. According to the student image information, the key bone point information of the human body is detected by the OpenPose algorithm. The denoised key bone point information is obtained after normalization processing. The joint heat map is obtained by Gaussian mapping according to the denoised key bone point information. The posture feature is obtained by feature extraction according to the joint heat map. Due to factors such as shooting distance and individual differences, the data processed by the OpenPose algorithm may contain a large amount of error data. The key point coordinate deviation caused by individual differences is reduced by normalization processing, and the redundancy of the data is reduced while ensuring the validity of the key bone point information. 3. According to the classroom attendance information, classroom attendance features are obtained through feature extraction, an expression feature evaluation matrix is ​​constructed according to the facial expression features and the expression feature evaluation matrix, a posture feature evaluation matrix is ​​constructed according to the posture features and the posture feature evaluation indicators, a comprehensive evaluation model is constructed according to the expression feature evaluation matrix and the posture feature evaluation matrix combined with the classroom attendance features, and a classroom quality score is output according to the comprehensive evaluation model, thereby achieving classroom quality scoring according to the student's classroom status, and obtaining a classroom quality score through fusion processing according to the classroom attendance information, facial expression features and posture features, thereby improving the scientificity, flexibility and authenticity of classroom quality evaluation. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to facilitate understanding by those skilled in the art, the present invention is further described below with reference to the accompanying drawings.

[0013] Figure 1 It is a flow chart of the intelligent evaluation method for classroom teaching of the present invention. DETAILED DESCRIPTION

[0014] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation methods, structures, features and effects of the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments.

[0015] See also Figure 1 , an intelligent evaluation method for classroom teaching, specifically comprising the following steps: S1: Get the classroom video; S2: Call the database to obtain class member information, extract student image information according to the classroom video recording, process the student image information through the YOLOV3 target detection method to obtain a target face frame set, obtain a target face key frame according to the target face frame set through the target tracking algorithm, and obtain class attendance information according to the target face key frame, the class member information includes student ID and class member face image information, and the class attendance information includes information of students who have attended and information of students who have not attended; S3: obtaining two-dimensional facial image information through a face detection algorithm according to the student image information, obtaining a denoised facial image through image processing according to the two-dimensional facial image information, obtaining three-dimensional facial image information through deep learning according to the denoised facial image, obtaining a first facial feature and a second facial feature through feature extraction according to the three-dimensional facial image information, and obtaining a facial expression feature through feature fusion model processing according to the first facial feature and the second facial feature; S4: Detect key bone point information of the human body through the OpenPose algorithm according to the student image information, obtain denoised key bone point information through normalization processing according to the key bone point information, obtain a joint heat map through Gaussian mapping according to the denoised key bone point information, and obtain posture features through feature extraction according to the joint heat map; S5: Obtain expression feature evaluation indicators and posture feature evaluation indicators, obtain classroom attendance features through feature extraction based on the classroom attendance information, construct an expression feature evaluation matrix based on the facial expression features and the expression feature evaluation matrix, construct a posture feature evaluation matrix based on the posture features and the posture feature evaluation indicators, construct a comprehensive evaluation model based on the expression feature evaluation matrix and the posture feature evaluation matrix combined with the classroom attendance features, and output a classroom quality score based on the comprehensive evaluation model.

[0016] It should be further explained that there is a one-to-one correspondence between the student ID and the facial image information of the class members.

[0017] Specifically, the step S2 specifically includes training a face recognition network model according to the face image information of the class members, processing the target face key frame through the face recognition network model to obtain the target face matching degree, when the target face matching degree is greater than or equal to a preset threshold, obtaining the attendance student information according to the corresponding student ID, and filtering out the student ID of the absent student according to the attendance student information and the class member information to obtain the absent student information; Specifically, step S3 includes the following steps: Student image information is extracted based on the classroom recording video, and a target face frame set is obtained by processing the student image information through the YOLOV3 target detection method. A target face key frame is obtained based on the target face frame set through a target tracking algorithm, and classroom attendance information is obtained based on the target face key frame, thereby realizing classroom attendance. The obtained classroom attendance information is conducive to classroom quality evaluation.

[0018] S301: acquiring an expression image sequence according to the denoised facial image information, obtaining a first standard frame and a second standard frame according to the expression image sequence, obtaining facial key point coordinates according to the first standard frame, obtaining a facial key point matrix according to the facial key point coordinates through a convolutional neural network, and obtaining the facial three-dimensional image information according to the facial key point matrix; S302: Obtain the first facial features through feature extraction according to the facial key point matrix, obtain expression image information through image segmentation according to the second standard frame, obtain a valid expression image through grayscale normalization processing according to the expression image information, obtain the second facial features by performing feature extraction on the valid expression image, and obtain the facial expression features by processing the first facial features and the second facial features through the feature fusion model.

[0019] Specifically, the step S302 specifically includes the feature fusion model being expressed as: , where O represents the facial expression feature, O1 represents the first facial feature, O2 represents the second facial feature, and α and β represent weighting coefficients.

[0020] Specifically, the step S4 specifically includes obtaining a posture feature triplet by feature extraction according to the target person key frame, and obtaining a joint heat map by Gaussian mapping according to the posture feature triplet, and the formula is: , where J is the joint heat map, is the posture feature triplet, is the Gaussian map contrast parameter.

[0021] According to the student image information, the key bone point information of the human body is detected by the OpenPose algorithm, and the denoised key bone point information is obtained after normalization processing. According to the denoised key bone point information, a joint heat map is obtained by Gaussian mapping, and posture features are obtained by feature extraction based on the joint heat map. Due to factors such as shooting distance and individual differences, the data processed by the OpenPose algorithm may contain a large amount of error data. This embodiment adopts the means of normalization processing to reduce the key point coordinate deviation caused by individual differences, and reduces the redundancy of data while ensuring the validity of the key bone point information.

[0022] Specifically, the S5 specifically includes the expression feature evaluation matrix represented as: , where R1 represents the expression feature evaluation matrix, represents the evaluation result of the i-th expression feature evaluation index on the j-th facial expression feature, and the posture feature evaluation matrix is ​​expressed as: , where T1 represents the posture feature evaluation matrix, Represents the evaluation result of the mth posture feature evaluation index on the nth posture feature.

[0023] Specifically, the step S5 further includes constructing the comprehensive evaluation model according to the expression feature evaluation matrix and the posture feature evaluation matrix combined with the classroom attendance feature, and the comprehensive evaluation model is expressed as: , where w1, w2 and w3 represent weight vectors, R1 represents the expression feature evaluation matrix, T1 represents the posture feature evaluation matrix, represents the class attendance characteristics, Represents the quality rating of the class.

[0024] Class attendance features are obtained through feature extraction based on the class attendance information, an expression feature evaluation matrix is ​​constructed based on the facial expression features and the expression feature evaluation matrix, a posture feature evaluation matrix is ​​constructed based on the posture features and the posture feature evaluation indicators, a comprehensive evaluation model is constructed based on the expression feature evaluation matrix and the posture feature evaluation matrix combined with the class attendance features, and a class quality score is output based on the comprehensive evaluation model, thereby achieving a score for class quality based on the class status of students, and obtaining a class quality score through fusion processing based on the class attendance information, facial expression features and posture features, thereby improving the scientificity, flexibility and authenticity of class quality evaluation.

[0025] Furthermore, the present application also provides an intelligent evaluation system for classroom teaching, comprising the following modules: Video acquisition module, used to acquire classroom recording videos; A classroom attendance module is used to call a database to obtain class member information, extract student image information based on the classroom video, obtain a target face frame set based on the student image information by using a YOLOV3 target detection method, obtain a target face key frame based on the target face frame set by using a target tracking algorithm, and obtain classroom attendance information based on the target face key frame, wherein the class member information includes student ID and class member face image information, and the classroom attendance information includes information on students who have attended and information on students who have not attended; A facial analysis module, configured to obtain two-dimensional facial image information by processing the student image information through a face detection algorithm, obtain a denoised facial image by image processing based on the two-dimensional facial image information, obtain three-dimensional facial image information by deep learning based on the denoised facial image, obtain a first facial feature and a second facial feature by feature extraction based on the three-dimensional facial image information, and obtain a facial expression feature by processing the first facial feature and the second facial feature through a feature fusion model; A posture analysis module is used to detect key bone point information of the human body through the OpenPose algorithm according to the student image information, obtain denoised key bone point information through normalization processing according to the key bone point information, obtain a joint heat map through Gaussian mapping according to the denoised key bone point information, and obtain posture features through feature extraction according to the joint heat map; A classroom evaluation module is used to obtain expression feature evaluation indicators and posture feature evaluation indicators, obtain classroom attendance features through feature extraction based on the classroom attendance information, construct an expression feature evaluation matrix based on the facial expression features and the expression feature evaluation matrix, construct a posture feature evaluation matrix based on the posture features and the posture feature evaluation indicators, construct a comprehensive evaluation model based on the expression feature evaluation matrix and the posture feature evaluation matrix combined with the classroom attendance features, and output a classroom quality score based on the comprehensive evaluation model.

[0026] The working principle and use process of the present invention: The video acquisition module is used to obtain the classroom video recording, and the student image information is extracted according to the classroom video recording. The target face key frame is obtained by processing the student image information, and the classroom attendance information is obtained according to the target face key frame. The student image information is processed by the face detection algorithm to obtain the facial two-dimensional image information, and the facial three-dimensional image information is obtained according to the facial two-dimensional image information. The first facial feature and the second facial feature are obtained by feature extraction according to the facial three-dimensional image information. The facial expression feature is obtained by processing the first facial feature and the second facial feature through the feature fusion model. The key bone point information of the human body is detected by the OpenPose algorithm according to the student image information. The denoised key bone point information is obtained after normalization processing according to the key bone point information. The joint heat map is obtained by Gaussian mapping according to the denoised key bone point information. The posture feature is obtained by feature extraction according to the joint heat map. The classroom attendance feature is obtained by feature extraction according to the classroom attendance information. The expression feature evaluation matrix is ​​constructed according to the facial expression feature and the expression feature evaluation matrix. The posture feature evaluation matrix is ​​constructed according to the posture feature and the posture feature evaluation index. A comprehensive evaluation model is constructed according to the expression feature evaluation matrix and the posture feature evaluation matrix combined with the classroom attendance feature, and the classroom quality score is output according to the comprehensive evaluation model.

[0027] The program code included in the method in the embodiment of the present invention can be transmitted with any appropriate medium, including but not limited to wireless, electric wire, optical cable, etc., or any suitable combination mentioned above. The computer program code for performing the operation of the present invention can be written with one or more programming languages ​​or their combination, and the programming language includes object-oriented programming languages--such as Java, Smalltalk, C++, and also includes conventional procedural programming languages--such as "C" language or similar programming languages. The program code can be executed completely on the user's computer, partially on the user's computer, executed as an independent software package, partially on the user's computer and partially on the remote computer, or executed completely on the remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (for example, using an Internet service provider to connect through the Internet).

[0028] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technical personnel in this field can make some changes or modify the technical contents disclosed above into equivalent embodiments without departing from the scope of the technical solution of the present invention. However, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present invention without departing from the content of the technical solution of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. An intelligent evaluation method for classroom teaching, characterized in that: The specific steps include: S1: Get the classroom video; S2: Call the database to obtain class member information, extract student image information according to the classroom video recording, process the student image information through the YOLOV3 target detection method to obtain a target face frame set, obtain a target face key frame according to the target face frame set through the target tracking algorithm, and obtain class attendance information according to the target face key frame, the class member information includes student ID and class member face image information, and the class attendance information includes information of students who have attended and information of students who have not attended; S3: obtaining two-dimensional facial image information through a face detection algorithm according to the student image information, obtaining a denoised facial image through image processing according to the two-dimensional facial image information, obtaining three-dimensional facial image information through deep learning according to the denoised facial image, obtaining a first facial feature and a second facial feature through feature extraction according to the three-dimensional facial image information, and obtaining a facial expression feature through feature fusion model processing according to the first facial feature and the second facial feature; S4: Detect key bone point information of the human body through the OpenPose algorithm according to the student image information, obtain denoised key bone point information through normalization processing according to the key bone point information, obtain a joint heat map through Gaussian mapping according to the denoised key bone point information, and obtain posture features through feature extraction according to the joint heat map; S5: Obtain expression feature evaluation indicators and posture feature evaluation indicators, obtain classroom attendance features through feature extraction based on the classroom attendance information, construct an expression feature evaluation matrix based on the facial expression features and the expression feature evaluation matrix, construct a posture feature evaluation matrix based on the posture features and the posture feature evaluation indicators, construct a comprehensive evaluation model based on the expression feature evaluation matrix and the posture feature evaluation matrix combined with the classroom attendance features, and output a classroom quality score based on the comprehensive evaluation model.

2. The intelligent evaluation method for classroom teaching according to claim 1, characterized in that: The step S2 specifically includes training a face recognition network model according to the face image information of the class members, processing the target face key frame through the face recognition network model to obtain the target face matching degree, when the target face matching degree is greater than or equal to a preset threshold, obtaining the attendance student information according to the corresponding student ID, and filtering out the student ID of the absent student according to the attendance student information and the class member information to obtain the absent student information.

3. The intelligent evaluation method for classroom teaching according to claim 1 is characterized in that: The step S3 specifically comprises the following steps: S301: acquiring an expression image sequence according to the denoised facial image information, obtaining a first standard frame and a second standard frame according to the expression image sequence, obtaining facial key point coordinates according to the first standard frame, obtaining a facial key point matrix according to the facial key point coordinates through a convolutional neural network, and obtaining the facial three-dimensional image information according to the facial key point matrix; S302: Obtain the first facial features through feature extraction according to the facial key point matrix, obtain expression image information through image segmentation according to the second standard frame, obtain a valid expression image through grayscale normalization processing according to the expression image information, obtain the second facial features by performing feature extraction on the valid expression image, and obtain the facial expression features by processing the first facial features and the second facial features through the feature fusion model.

4. The intelligent evaluation method for classroom teaching according to claim 3 is characterized in that: The step S302 specifically includes the feature fusion model being represented as: , where O represents the facial expression feature, O1 represents the first facial feature, O2 represents the second facial feature, and α and β represent weighting coefficients.

5. The intelligent evaluation method for classroom teaching according to claim 1 is characterized in that: The step S4 specifically includes obtaining a posture feature triplet by feature extraction according to the target person key frame, and obtaining a joint heat map by Gaussian mapping according to the posture feature triplet, and the formula is: , where J is the joint heat map, is the posture feature triplet, is the Gaussian map contrast parameter.

6. The intelligent evaluation method for classroom teaching according to claim 1 is characterized in that: The step S5 specifically includes expressing the expression feature evaluation matrix as follows: , where R1 represents the expression feature evaluation matrix, represents the evaluation result of the i-th expression feature evaluation index on the j-th facial expression feature, and the posture feature evaluation matrix is ​​expressed as: , where T1 represents the posture feature evaluation matrix, Represents the evaluation result of the mth posture feature evaluation index on the nth posture feature.

7. The intelligent evaluation method for classroom teaching according to claim 1 is characterized in that: The step S5 specifically further includes constructing the comprehensive evaluation model according to the expression feature evaluation matrix and the posture feature evaluation matrix in combination with the classroom attendance feature, and the comprehensive evaluation model is expressed as: , where w1, w2 and w3 represent weight vectors, R1 represents the expression feature evaluation matrix, T1 represents the posture feature evaluation matrix, represents the class attendance characteristics, Represents the quality rating of the class.

8. An intelligent evaluation system for classroom teaching, the intelligent evaluation system for classroom teaching adopts the intelligent evaluation method for classroom teaching as claimed in claim 1, characterized in that: Includes the following modules: Video acquisition module, used to acquire classroom recording videos; A classroom attendance module is used to call a database to obtain class member information, extract student image information based on the classroom video, obtain a target face frame set based on the student image information by using a YOLOV3 target detection method, obtain a target face key frame based on the target face frame set by using a target tracking algorithm, and obtain classroom attendance information based on the target face key frame, wherein the class member information includes student ID and class member face image information, and the classroom attendance information includes information on students who have attended and information on students who have not attended; A facial analysis module, configured to obtain two-dimensional facial image information by processing the student image information through a face detection algorithm, obtain a denoised facial image by image processing based on the two-dimensional facial image information, obtain three-dimensional facial image information by deep learning based on the denoised facial image, obtain a first facial feature and a second facial feature by feature extraction based on the three-dimensional facial image information, and obtain a facial expression feature by processing the first facial feature and the second facial feature through a feature fusion model; A posture analysis module is used to detect key bone point information of the human body through the OpenPose algorithm according to the student image information, obtain denoised key bone point information through normalization processing according to the key bone point information, obtain a joint heat map through Gaussian mapping according to the denoised key bone point information, and obtain posture features through feature extraction according to the joint heat map; A classroom evaluation module is used to obtain expression feature evaluation indicators and posture feature evaluation indicators, obtain classroom attendance features through feature extraction based on the classroom attendance information, construct an expression feature evaluation matrix based on the facial expression features and the expression feature evaluation matrix, construct a posture feature evaluation matrix based on the posture features and the posture feature evaluation indicators, construct a comprehensive evaluation model based on the expression feature evaluation matrix and the posture feature evaluation matrix combined with the classroom attendance features, and output a classroom quality score based on the comprehensive evaluation model.