Classroom diagnosis analysis method based on deep learning

Abstract

The invention discloses a classroom diagnosis analysis method based on deep learning, and relates to the technical field of education evaluation. According to the method, through a classroom diagnosis evaluation link, hand raising and speaking action behaviors of students are identified from video data, the number of times is counted, an initiative score is given, an enthusiasm evaluation index and an enthusiasm optimization index are calculated, and the enthusiasm of the students in classroom participation is considered; and meanwhile, counting the duration of abnormal behavior fragments such as sleeping, mobile phone playing and distraction, calculating an abnormal behavior index and an abnormal value, and evaluating the classroom concentration degree and the abnormal behavior influence of the students. Besides, the extracurricular combination evaluation combines a student state evaluation index for classroom performance evaluation with an assessment score of a corresponding type of classroom in a set time to obtain a comprehensive diagnosis evaluation index, and the students are comprehensively evaluated from the two aspects of classroom performance and knowledge mastering degree, so that the evaluation result is closer to the real learning condition of the students.

Description

Technical Field

[0001] The present invention relates to the technical field of educational evaluation, and specifically to a classroom diagnosis and analysis method based on deep learning. Background Technique

[0002] In today's educational field, the evaluation of classroom teaching quality is crucial for improving educational effects and promoting the development of students.

[0003] However, the following deficiencies still exist in the actual application process of the existing classroom quality diagnosis methods: Only using examination scores to judge students' learning achievements, this method only focuses on students' memory and understanding of knowledge, ignoring key factors such as students' learning processes, participation degrees, and learning attitudes in the classroom; Lacking continuous tracking and analysis of the dynamic classroom process, it usually conducts a one-time evaluation at specific time nodes, such as the end of the semester, and cannot provide real-time feedback on the changes and development trends of students in the classroom learning process; this makes it difficult for teachers to adjust teaching strategies in a timely manner and cannot carry out targeted teaching interventions according to the actual situation of students, affecting the improvement of teaching effects.

[0004] Therefore, a classroom diagnosis and analysis method based on deep learning is introduced. Summary of the Invention

[0005] The purpose of the present invention is to solve the problems pointed out in the background technique, and to propose a classroom diagnosis and analysis method based on deep learning.

[0006] The purpose of the present invention can be achieved through the following technical solutions: A classroom diagnosis and analysis method based on deep learning, including: Classroom data collection: Collect videos of various types of classrooms during the teaching process, and preprocess the collected video data. After the preprocessing is completed, store it in a pre-constructed evaluation database; Classroom diagnosis setting: Set the number of diagnostic classrooms corresponding to each type of classroom. When the cumulative number of classes of a certain type of classroom reaches the corresponding set number of diagnostic classrooms, extract the video data collected for each class of the corresponding type of classroom from the evaluation database; Classroom diagnosis evaluation: Input the video data collected for each class of the corresponding type of classroom into a pre-constructed deep learning model for comprehensive evaluation, so as to determine the student status evaluation index kgt of each student in the corresponding type of classroom during this classroom diagnosis process; where t represents the number of each student, t = 1, 2,..., r, and r is the total number of students; Extra-curricular combined assessment: Based on the student status assessment index kgt of each student in the corresponding type of class during the class diagnosis process, combined with the assessment results of each student's performance in the corresponding type of class within the set time range, the comprehensive diagnosis assessment index zpt of each student in the corresponding type of class during the class diagnosis process is output.

[0007] As a preferred embodiment of the present invention, the video data collected from each class of the corresponding type of class is input into a pre-constructed deep learning model for comprehensive evaluation. Specifically: Identify the action behaviors of students from the video data collected from each class of the corresponding type of class; count the number of times each student raises their hand and stands up to speak in each class from the action behaviors of each class of the corresponding type of class. Set an initiative score corresponding to each hand-raising behavior and standing-up speaking behavior respectively; multiply the number of times each student raises their hand and stands up to speak in each class in the video data collected from each class of the corresponding type of class by the corresponding initiative scores respectively, so as to obtain the hand-raising initiative score and speaking initiative score of each student in each class. Set the weight coefficients corresponding to the hand-raising initiative score and speaking initiative score respectively, multiply the hand-raising initiative score and speaking initiative score of each student in each class by the corresponding set weight coefficients respectively, and then sum to obtain the enthusiasm assessment index of each student in each class. For the enthusiasm assessment index of each student in each class, remove the highest enthusiasm assessment index and the lowest enthusiasm assessment index, and calculate the average value of the remaining groups of enthusiasm assessment indexes after removal, which is used as the enthusiasm optimization index of each student during the class diagnosis process.

[0008] As a preferred embodiment of the present invention, inputting the video data collected from each class of the corresponding type of class into a pre-constructed deep learning model for comprehensive evaluation further includes: Extract the enthusiasm assessment index of each student in each class, arrange the groups of enthusiasm assessment indexes according to the order of each class, and calculate the difference between adjacent groups of enthusiasm assessment indexes after the arrangement; the difference calculation order is to subtract the enthusiasm assessment index on the left from the enthusiasm assessment index on the right in the adjacent two groups of enthusiasm assessment indexes. If the calculation result of a certain group of differences shows a negative value, take the absolute value and mark it as a positive decline value. If the calculation result of a certain group of differences shows a positive value, directly mark it as a positive increase value. Sum all the positive rising values and positive falling values respectively to obtain the total positive rising value and the total positive falling value. Through ratio calculation, that is, the total positive rising value / the total positive falling value, the active trend index of each student in this classroom diagnosis process is obtained; Preset the index value ranges corresponding to each group of the active trend index, and set an additional coefficient corresponding to each group of index value ranges; Match the active trend index of each student in this classroom diagnosis process with the preset index value ranges of each group to determine the additional coefficient of each student in this classroom diagnosis process; Multiply the additional coefficient determined by each student by the enthusiasm optimization index to obtain the performance value uyt of each student in this classroom diagnosis process.

[0009] As a preferred implementation manner of the present invention, inputting the video data collected from each class corresponding to the corresponding type of classroom into a pre-constructed deep learning model for comprehensive evaluation further includes: Extract the sleeping segments, mobile phone playing segments, and distracted segments of each student during each class from the action behaviors of each class corresponding to the corresponding type of classroom; Analyze the duration of each group of sleeping segments of each student during each class and accumulate them to obtain the cumulative sleeping duration of each student during each class; Analyze the duration of each group of mobile phone playing segments of each student during each class and accumulate them to obtain the cumulative mobile phone playing duration of each student during each class; Analyze the duration of each group of distracted segments of each student during each class and accumulate them to obtain the cumulative distracted duration of each student during each class; Extract the class duration corresponding to a single class, and calculate the proportions of the cumulative sleeping duration, cumulative mobile phone playing duration, and cumulative distracted duration of each student in the class duration respectively to obtain the sleeping proportion, mobile phone playing proportion, and distracted proportion of each student during each class; Set the weight coefficients corresponding to the sleeping proportion, mobile phone playing proportion, and distracted proportion respectively, multiply the sleeping proportion, mobile phone playing proportion, and distracted proportion of each student during each class by the corresponding set weight coefficients respectively, and then sum them to obtain the difference index of each student during each class; Preset the reference threshold index range corresponding to the difference index; Compare the difference index of each student during each class with the preset reference threshold index range, count the number of difference indexes of each student higher than the reference threshold index range, denoted as the high difference number; count the number of difference indexes of each student within the reference threshold index range, denoted as the low difference number; Set the abnormal behavior impact scores corresponding to the index range higher than the reference threshold and the index range within the reference threshold respectively; multiply the number of high abnormalities and the number of low abnormalities by the corresponding set abnormal behavior impact scores respectively, and then sum to obtain the abnormality value uqt of each student during this classroom diagnosis.

[0010] As a preferred embodiment of the present invention, determine the student status evaluation index kgt of each student during this classroom diagnosis for the corresponding type of classroom, specifically: Obtain the appearance value uyt and the abnormality value uqt of each student during this classroom diagnosis, and preset the passing appearance value um and the allowable abnormality value us of the appearance value uyt and the abnormality value uqt corresponding to each type of classroom; According to the formula , calculate the appearance value uyt and the abnormality value uqt of each student during this classroom diagnosis, so as to determine the student status evaluation index kgt of each student during this classroom diagnosis for the corresponding type of classroom; where Are the influence weight factors of the appearance value uyt and the abnormality value uqt respectively; Extra-curricular combined evaluation: Based on the student status evaluation index kgt of each student during this classroom diagnosis for the corresponding type of classroom, combine the assessment results of each student's corresponding type of classroom within the set time range, so as to output the comprehensive diagnosis and evaluation index zpt of each student during this classroom diagnosis for the corresponding type of classroom.

[0011] As a preferred embodiment of the present invention, output the comprehensive diagnosis and evaluation index zpt of each student during this classroom diagnosis for the corresponding type of classroom, specifically: Obtain the assessment results of each student's corresponding type of classroom within the set time range; calculate the average value of the assessment results of each student's corresponding type of classroom, so as to obtain the performance value of each student's corresponding type of classroom; Set the groups of score value ranges corresponding to the performance value of the corresponding type of classroom, and each group of score value ranges corresponds to a status adjustment coefficient; Match the performance value of each student's corresponding type of classroom with the groups of score value ranges to determine the status adjustment coefficient of each student's corresponding type of classroom; Multiply the student status evaluation index kgt of each student during this classroom diagnosis by the determined status adjustment coefficient to obtain the comprehensive diagnosis and evaluation index zpt of each student during this classroom diagnosis for the corresponding type of classroom.

[0012] As a preferred embodiment of the present invention, it further includes: Diagnostic result output: Based on the comprehensive diagnostic evaluation index zpt of each student in each type of class during the class diagnosis process, output the corresponding class quality level for each type of class; the class quality levels include abnormal class quality level, general class quality level, and excellent class quality level.

[0013] As a preferred implementation manner of the present invention, output the corresponding class quality level for each type of class, specifically: Calculate the average value of the comprehensive diagnostic evaluation index zpt of each student in each type of class during the class diagnosis process, and use the calculated average value as the class status evaluation index for each type of class; Set three groups of index value ranges corresponding to the class status evaluation index, and each of the three groups of index value ranges corresponds to a class quality level; Match the class status evaluation index of each type of class during the class diagnosis process with the set three groups of index value ranges to determine the class quality level of each type of class during the class diagnosis process.

[0014] Compared with the prior art, the beneficial effects of the present invention are: Through the class diagnosis and evaluation link, the present invention identifies the actions of students raising their hands and standing up to speak from video data, counts the number of times and assigns initiative scores, calculates the enthusiasm evaluation index and enthusiasm optimization index, and considers the enthusiasm of students' class participation; at the same time, it counts the duration of abnormal behavior segments such as sleeping, playing with mobile phones, and being absent-minded, calculates the abnormal index and abnormal value, and evaluates the students' class concentration and the impact of abnormal behaviors. In addition, the extracurricular combined evaluation combines the student status evaluation index of the class performance evaluation with the assessment results of the corresponding type of class within the set time to obtain a comprehensive diagnostic evaluation index, comprehensively evaluating students from both aspects of class performance and knowledge mastery, making the evaluation results closer to the students' real learning situation; The present invention obtains the active trend index by calculating the ratio of the total active increase value to the total active decrease value, providing a quantitative comprehensive index for the change in students' enthusiasm. This index can more comprehensively reflect the overall trend of students' class enthusiasm within a period of time, and adjusts the enthusiasm optimization index according to the active trend index, considering the change trend of students' enthusiasm, improving the accuracy and comprehensiveness of the analysis of students' class performance; The present invention combines the student status evaluation index of the class performance evaluation with the assessment results of the corresponding type of class within the set time to obtain a comprehensive diagnostic evaluation index, comprehensively evaluating students from both aspects of class performance and knowledge mastery, making the evaluation results closer to the students' real learning situation; By calculating the mean of the comprehensive diagnostic evaluation indices of students in each type of classroom, the present invention obtains the classroom status evaluation index, and sets three groups of index value ranges corresponding to different classroom quality levels, which helps schools and teachers to comprehensively understand the classroom teaching quality, discover problems existing in the teaching process, and provide a strong basis for teaching management and decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] For the convenience of those skilled in the art to understand, the present invention will be further described below in conjunction with the accompanying drawings.

[0016] Figure 1 It is a flowchart of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0017] The technical solutions of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative work shall fall within the protection scope of the present invention.

[0018] Please refer to Figure 1 as shown, a classroom diagnosis and analysis method based on deep learning includes: Classroom data collection: Using cameras pre-deployed in the target classroom area, collect videos of various types of classrooms during the teaching process; various types of classrooms are classrooms of various subjects, such as Chinese, mathematics, and English, etc.; and perform preprocessing on the collected video data; preprocessing is to perform image enhancement and denoising on the collected video data; after preprocessing is completed, store it in the pre-constructed evaluation database. It should be noted that the cameras are specifically installed with high-definition cameras at the four corners of the classroom ceiling and above the blackboard. These cameras have ultra-wide angles and high-definition picture quality, and can capture every move of teachers and students in the classroom in all directions.

[0019] Classroom diagnosis setting: Set the number of diagnostic classrooms corresponding to each type of classroom. When the cumulative number of classes of a certain type of classroom reaches the corresponding set number of diagnostic classrooms, extract the video data collected in each class corresponding to the corresponding type of classroom from the evaluation database. It should be noted that, for example, for the Chinese classroom, a comprehensive diagnostic analysis is carried out every time the cumulative number of classes reaches 15; for the mathematics classroom, a diagnostic analysis is carried out every time the cumulative number of classes reaches 18; for the English classroom, a diagnostic analysis is carried out every time the cumulative number of classes reaches 20. Monitor the number of classes for each type of class. When the set number of diagnostic classes is reached, data retrieval is performed through an SQL query statement in the evaluation database. For example, if the cumulative number of Chinese classes reaches the set number of diagnostic classes (15 classes), then search for Chinese classes in the evaluation database and perform data retrieval and extraction.

[0020] Classroom diagnosis and evaluation: Input the video data collected for each class of the corresponding type of class into a pre-constructed deep learning model for comprehensive evaluation, so as to determine the student status evaluation index kgt of each student in the corresponding type of class during this classroom diagnosis; where t represents the number of each student, t = 1, 2,..., r, and r is the total number of students. Specifically: Identify the action behaviors of students from the video data collected for each class of the corresponding type of class; count the number of times each student raises their hand and stands up to speak in each class from the action behaviors of the corresponding type of class in each class. It should be noted that the identification method for identifying students' action behaviors specifically refers to the student classroom behavior identification technology based on SSD-MobileNetV2 in CN114067432A. The behavior types include: "reading", "sleeping", "playing with mobile phones", "standing up", "raising hands", "listening to class", "being distracted", "writing". The behavior types corresponding to each abnormal behavior include: "sleeping", "playing with mobile phones", "being distracted".

[0021] Set an initiative score corresponding to each hand-raising behavior and standing-up speaking behavior respectively; the initiative score corresponding to the standing-up speaking behavior is greater than the initiative score corresponding to the hand-raising behavior. Multiply the number of times each student raises their hand and stands up to speak in each class in the video data collected for each class of the corresponding type of class by the corresponding initiative scores respectively, so as to obtain the hand-raising initiative score and speaking initiative score of each student in each class. Set the weight coefficients corresponding to the hand-raising initiative score and speaking initiative score respectively, multiply the hand-raising initiative score and speaking initiative score of each student in each class by the corresponding set weight coefficients respectively, and then sum to obtain the enthusiasm evaluation index of each student in each class. For the enthusiasm evaluation index of each student in each class, remove the highest enthusiasm evaluation index and the lowest enthusiasm evaluation index. After removal, calculate the average value of the remaining groups of enthusiasm evaluation indexes as the enthusiasm optimization index of each student during this classroom diagnosis process. It should be noted that by setting the weight coefficients of the active scoring for raising hands and the active scoring for speaking, and obtaining the enthusiasm evaluation index of students in each class through weighted summation, the different importance of different active behaviors in the evaluation is considered, making the evaluation results more scientific and reasonable; When calculating the enthusiasm optimization index of students in the process of this class diagnosis, one highest enthusiasm evaluation index and one lowest enthusiasm evaluation index are removed, and then the average value of the remaining indexes is calculated. This operation effectively avoids the influence of individual extreme data on the overall evaluation results, improves the reliability and stability of the evaluation results, and can more truly reflect the enthusiasm level of students in the class; Teachers can understand the learning characteristics and participation of each student based on the enthusiasm optimization index of the students and the behavior recognition results, and thus adjust the teaching methods and strategies accordingly.

[0022] Extract the enthusiasm evaluation index of each student in each class, arrange the enthusiasm evaluation indexes of each group in the order of each class, and calculate the difference between adjacent groups of enthusiasm evaluation indexes after the arrangement; the difference calculation order is to subtract the enthusiasm evaluation index on the left from the enthusiasm evaluation index on the right in adjacent groups of enthusiasm evaluation indexes; If the calculation result of the difference of a certain group shows a negative value, take the absolute value and mark it as the positive decline value. If the calculation result of the difference of a certain group shows a positive value, directly mark it as the positive rise value; Sum up all the positive rise values and positive decline values respectively, so as to obtain the total positive rise value and the total positive decline value. Through ratio calculation, that is, the total positive rise value / the total positive decline value, the active trend index of each student in the process of this class diagnosis is obtained; It should be noted that the active trend index is obtained by calculating the ratio of the total positive rise value to the total positive decline value, which provides a quantitative comprehensive index for the change of students' enthusiasm. This index can more comprehensively reflect the overall trend of students' classroom enthusiasm in a period of time. When the active trend index is close to 1, it means that the enthusiasm of students tends to be stable. When it is greater than 1, it means that the overall trend of enthusiasm is gradually increasing. When it is less than 1, it means that the overall trend of enthusiasm is gradually decreasing.

[0023] Preset the value range of each group of indexes corresponding to the active trend index, and set an additional coefficient corresponding to each group of index value ranges; the value range of the additional coefficient is set between 0.875 - 1.236, and the higher the active trend index, the higher the corresponding additional coefficient; Match the active trend index of each student in the process of this class diagnosis with the preset value range of each group of indexes, so as to determine the additional coefficient of each student in the process of this class diagnosis; Multiply the additional coefficients determined by each student by the enthusiasm optimization index to obtain the performance value uyt of each student during this classroom diagnosis; Extract the sleeping segments, mobile phone playing segments, and distracted segments of each student during each class from the action behaviors corresponding to each class of the corresponding type of class; Analyze the duration of each group of sleeping segments of each student during each class and accumulate them to obtain the cumulative sleeping duration of each student during each class; Analyze the duration of each group of mobile phone playing segments of each student during each class and accumulate them to obtain the cumulative mobile phone playing duration of each student during each class; Analyze the duration of each group of distracted segments of each student during each class and accumulate them to obtain the cumulative distracted duration of each student during each class; Extract the class duration corresponding to a single class, and calculate the proportions of the cumulative sleeping duration, cumulative mobile phone playing duration, and cumulative distracted duration of each student in the class duration respectively, to obtain the sleeping proportion, mobile phone playing proportion, and distracted proportion of each student during each class; Set the weight coefficients corresponding to the sleeping proportion, mobile phone playing proportion, and distracted proportion respectively, multiply the sleeping proportion, mobile phone playing proportion, and distracted proportion of each student during each class by the corresponding set weight coefficients respectively, and then sum them to obtain the deviation index of each student during each class; Preset the reference threshold index range corresponding to the deviation index; being higher than the reference threshold index range indicates that the student spends most of the time on abnormal behaviors during this class, and being within the reference threshold index range indicates that the student spends most of the time on abnormal behaviors during this class; Compare the deviation index of each student during each class with the preset reference threshold index range, count the number of deviation indexes of each student higher than the reference threshold index range, denoted as the high deviation number; count the number of deviation indexes of each student within the reference threshold index range, denoted as the low deviation number; Set the abnormal behavior impact scores corresponding to being higher than the reference threshold index range and being within the reference threshold index range respectively; the abnormal behavior impact score corresponding to being higher than the reference threshold index range is higher than the abnormal behavior impact score corresponding to being within the reference threshold index range; multiply the high deviation number and the low deviation number by the corresponding set abnormal behavior impact scores respectively, and then sum them to obtain the abnormal condition value uqt of each student during this classroom diagnosis; Obtain the performance value uyt and the abnormal condition value uqt of each student during this classroom diagnosis, and preset the passing performance value um and the allowable abnormal condition value us of the performance value uyt and the abnormal condition value uqt corresponding to each type of class; According to the formula , calculate the performance value uyt and abnormal value uqt of each student during the classroom diagnosis process, so as to determine the student status evaluation index kgt of each student in the corresponding type of classroom during the classroom diagnosis process; among them are the influence weight factors of the performance value uyt and the abnormal value uqt respectively; The value of is set specifically according to each type of classroom; Extra-curricular combined evaluation: Based on the student status evaluation index kgt of each student in the corresponding type of classroom during the classroom diagnosis process, combined with the assessment results of each student's corresponding type of classroom within the set time range, the comprehensive diagnosis evaluation index zpt of each student in the corresponding type of classroom during the classroom diagnosis process is output; Specifically: Obtain the assessment scores of each student's corresponding type of classroom within the set time range; calculate the average value of the assessment scores of each student's corresponding type of classroom, so as to obtain the performance value of each student's corresponding type of classroom; Set the score value ranges corresponding to the performance values of the corresponding type of classroom, and each score value range corresponds to a status adjustment coefficient; the status adjustment coefficient range is set between 0.893-1.176, and the higher the performance value, the higher the corresponding status adjustment coefficient, and vice versa; Match the performance value of each student's corresponding type of classroom with each group of score value ranges to determine the status adjustment coefficient of each student's corresponding type of classroom; Multiply the student status evaluation index kgt of each student during the classroom diagnosis process by the determined status adjustment coefficient to obtain the comprehensive diagnosis evaluation index zpt of each student in the corresponding type of classroom during the classroom diagnosis process; It should be noted that combining the student status evaluation index obtained from behavior analysis in the classroom with the assessment results of the corresponding type of classroom within the set time range can more comprehensively evaluate students from two aspects of classroom performance and knowledge mastery, changing the limitation of relying solely on grades or classroom performance for evaluation, making the evaluation results closer to the real learning situation of students. The higher the grade, the corresponding increase in the student status evaluation index, and the lower the grade, the corresponding decrease in the student status evaluation index.

[0024] Diagnosis result output: According to the comprehensive diagnosis evaluation index zpt of each student in each type of classroom during the classroom diagnosis process, output the classroom quality level corresponding to each type of classroom; the classroom quality level includes abnormal classroom quality level, general classroom quality level, and excellent classroom quality level; Calculate the average value of the comprehensive diagnosis evaluation index zpt of each student in each type of classroom during the classroom diagnosis process, and use the calculated average value as the classroom status evaluation index of each type of classroom; Set three groups of index value ranges corresponding to the classroom status evaluation index, and each group of index value ranges corresponds to a classroom quality level; the higher the classroom status evaluation index, the higher the possibility of corresponding to the excellent classroom quality level; Match the classroom status evaluation index of each type of classroom in the current classroom diagnosis process with the set three groups of index value ranges, so as to determine the classroom quality level of each type of classroom in the current classroom diagnosis process; It should be noted that according to the classroom quality level of each type of classroom in the current classroom diagnosis process, an abnormal classroom quality level and a general classroom quality level are set to correspond to a reduction ratio of the number of diagnosed classrooms respectively. If the classroom quality level of a certain type of classroom in the current classroom diagnosis process is the abnormal classroom quality level or the general classroom quality level, then on the basis of the current number of diagnosed classrooms of the corresponding type of classroom, the corresponding reduction ratio is reduced, and it is updated as the number of diagnosed classrooms of the corresponding type of classroom; if the calculation result has a decimal point, round it off.

[0025] The preferred embodiments of the present invention disclosed above are only used to help illustrate the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the present invention to the specific embodiments. Obviously, many modifications and changes can be made according to the content of this specification. These embodiments are selected and specifically described in this specification to better explain the principle and practical application of the present invention, so that those skilled in the art in the relevant technical field can understand and utilize the present invention well. The present invention is only limited by the claims and their full scope and equivalents.

Claims

1. A classroom diagnosis and analysis method based on deep learning, characterized in that Including: Classroom data collection: Collect videos during the teaching process of various types of classrooms, and preprocess the collected video data. After the preprocessing is completed, store it in a pre-constructed evaluation database; Classroom diagnosis setting: Set the number of diagnostic classrooms corresponding to each type of classroom. When the cumulative number of classes of a certain type of classroom reaches the corresponding set number of diagnostic classrooms, extract the video data collected for each class of the corresponding type of classroom from the evaluation database; Classroom diagnosis evaluation: Input the video data collected for each class of the corresponding type of classroom into a pre-constructed deep learning model for comprehensive evaluation, so as to determine the student status evaluation index kgt of each student in the corresponding type of classroom during this classroom diagnosis process; where t represents the number of each student, t = 1, 2,..., r, and r is the total number of students; Extra-curricular combined evaluation: Based on the student status evaluation index kgt of each student in the corresponding type of classroom during this classroom diagnosis process, combined with the assessment results of each student's performance in the corresponding type of classroom within the set time range, so as to output the comprehensive diagnosis evaluation index zpt of each student in the corresponding type of classroom during this classroom diagnosis process.

2. The classroom diagnosis and analysis method based on deep learning according to claim 1, wherein Input the video data collected for each class of the corresponding type of classroom into a pre-constructed deep learning model for comprehensive evaluation, specifically: Identify the action behaviors of students from the video data collected for each class of the corresponding type of classroom; count the number of times each student raises their hand and stands up to speak in each class from the action behaviors of each class of the corresponding type of classroom; Set a proactive score corresponding to each hand-raising behavior and standing-up-and-speaking behavior respectively; multiply the number of times each student raises their hand and stands up to speak in each class in the video data collected for each class of the corresponding type of classroom by the corresponding proactive scores respectively, so as to obtain the hand-raising proactive score and speaking proactive score of each student in each class; Set the weight coefficients corresponding to the hand-raising proactive score and speaking proactive score respectively, multiply the hand-raising proactive score and speaking proactive score of each student in each class by the corresponding set weight coefficients respectively, and then sum them up to obtain the enthusiasm evaluation index of each student in each class; For the enthusiasm evaluation index of each student in each class, remove one highest enthusiasm evaluation index and one lowest enthusiasm evaluation index, and after removal, calculate the average value of the remaining groups of enthusiasm evaluation indexes as the enthusiasm optimization index of each student during this classroom diagnosis process.

3. The classroom diagnosis and analysis method based on deep learning according to claim 2, wherein Input the video data collected for each class of the corresponding type of classroom into a pre-constructed deep learning model for comprehensive evaluation, and also include: Extract the enthusiasm evaluation index of each student in each class, arrange the groups of enthusiasm evaluation indexes in the order of each class, and calculate the difference between adjacent groups of enthusiasm evaluation indexes after the arrangement; the order of the difference calculation is to subtract the enthusiasm evaluation index on the left from the enthusiasm evaluation index on the right in the adjacent two groups of enthusiasm evaluation indexes; If the calculation result of a certain group of differences shows a negative value, take the absolute value and mark it as a positive decline value. If the calculation result of a certain group of differences shows a positive value, directly mark it as a positive increase value; Sum up all the positive increase values and positive decline values respectively to obtain the total positive increase value and the total positive decline value. Through ratio calculation, that is, the total positive increase value / the total positive decline value, obtain the active trend index of each student in this classroom diagnosis process; Preset the index value ranges corresponding to the active trend index, and set an additional coefficient corresponding to each group of index value ranges; Match the active trend index of each student in this classroom diagnosis process with the preset index value ranges of each group to determine the additional coefficient of each student in this classroom diagnosis process; Multiply the additional coefficient determined for each student by the enthusiasm optimization index to obtain the performance value uyt of each student in this classroom diagnosis process; 4. The classroom diagnosis and analysis method based on deep learning according to claim 3, wherein, Input the video data collected from each class corresponding to the corresponding type of classroom into a pre-constructed deep learning model for comprehensive evaluation, which also includes: Extract the sleeping segments, mobile phone playing segments, and distracted segments of each student during each class from the action behaviors of each class corresponding to the corresponding type of classroom; Analyze the duration of each group of sleeping segments of each student during each class and accumulate them to obtain the cumulative sleeping duration of each student during each class; Analyze the duration of each group of mobile phone playing segments of each student during each class and accumulate them to obtain the cumulative mobile phone playing duration of each student during each class; Analyze the duration of each group of distracted segments of each student during each class and accumulate them to obtain the cumulative distracted duration of each student during each class; Extract the class duration corresponding to a single class, and calculate the proportions of the cumulative sleeping duration, cumulative mobile phone playing duration, and cumulative distracted duration of each student in the class duration respectively to obtain the sleeping proportion, mobile phone playing proportion, and distracted proportion of each student during each class; Set the weight coefficients corresponding to the sleeping proportion, mobile phone playing proportion, and distracted proportion respectively. Multiply the sleeping proportion, mobile phone playing proportion, and distracted proportion of each student during each class by the corresponding set weight coefficients respectively, and then sum them up to obtain the difference index of each student during each class; Preset the reference threshold index range corresponding to the difference index; Compare the difference index of each student during each class with the preset reference threshold index range, count the number of difference indices of each student higher than the reference threshold index range, denoted as the high difference number; count the number of difference indices of each student within the reference threshold index range, denoted as the low difference number; Set the abnormal behavior impact scores corresponding to higher than the reference threshold index range and within the reference threshold index range respectively; multiply the high difference number and the low difference number by the corresponding set abnormal behavior impact scores respectively, and then sum them up to obtain the abnormal condition value uqt of each student in this classroom diagnosis process; 5. The classroom diagnosis and analysis method based on deep learning according to claim 4, characterized in that, Determine the student status evaluation index kgt of each student in this classroom diagnosis process for the corresponding type of classroom, specifically: Obtain the performance values uyt and abnormal values uqt of each student during this classroom diagnosis, and preset the passing performance value um and allowable abnormal value us of the performance value uyt and abnormal value uqt corresponding to each type of classroom. According to the formula , calculate the performance value uyt and the abnormal value uqt of each student during the classroom diagnosis process, so as to determine the student status evaluation index kgt of each student in the corresponding type of classroom during the classroom diagnosis process; where are the influence weight factors of the performance value uyt and the abnormal value uqt respectively; Extra-curricular combined assessment: Based on the student status assessment index kgt of each student in the corresponding type of classroom during this classroom diagnosis, combined with the assessment results of the corresponding type of classroom for each student within the set time range, so as to output the comprehensive diagnosis assessment index zpt of each student in the corresponding type of classroom during this classroom diagnosis.

6. The classroom diagnosis and analysis method based on deep learning according to claim 5, wherein Output the comprehensive diagnosis assessment index zpt of each student in the corresponding type of classroom during this classroom diagnosis, specifically: Obtain the assessment scores of each student in the corresponding type of classroom within the set time range; calculate the average of the assessment scores of each student in the corresponding type of classroom to obtain the performance value of each student in the corresponding type of classroom. Set the score value ranges corresponding to the performance values of the corresponding type of classroom, and each score value range corresponds to a state adjustment coefficient. Match the performance values of each student in the corresponding type of classroom with the score value ranges to determine the state adjustment coefficient of each student in the corresponding type of classroom. Multiply the student status assessment index kgt of each student during this classroom diagnosis by the determined state adjustment coefficient to obtain the comprehensive diagnosis assessment index zpt of each student in the corresponding type of classroom during this classroom diagnosis.

7. The classroom diagnosis and analysis method based on deep learning according to claim 6, wherein It also includes: Diagnosis result output: According to the comprehensive diagnosis assessment index zpt of each student in each type of classroom during this classroom diagnosis, output the classroom quality level corresponding to each type of classroom; the classroom quality level includes abnormal classroom quality level, general classroom quality level, and excellent classroom quality level.

8. A classroom diagnosis and analysis method based on deep learning according to claim 7, characterized in that, Output the classroom quality level corresponding to each type of classroom, specifically: Calculate the average of the comprehensive diagnosis assessment index zpt of each student in each type of classroom during this classroom diagnosis, and use the calculated average as the classroom status assessment index of each type of classroom. Set three groups of index value ranges corresponding to the classroom status assessment index, and each group of index value ranges corresponds to a classroom quality level. Match the classroom status assessment index of each type of classroom during this classroom diagnosis with the set three groups of index value ranges to determine the classroom quality level of each type of classroom during this classroom diagnosis.

Images