AI-driven classroom real-time interaction and teaching quality intelligent optimization system and method

Abstract

The invention relates to the technical field of teaching quality, in particular to an AI-driven classroom real-time interaction and teaching quality intelligent optimization system and method.The AI-driven classroom real-time interaction and teaching quality intelligent optimization system comprises a classroom test terminal used for determining student states, retrieving related knowledge points to generate classroom test questions and outputting answer results to generate a review report; the analysis and evaluation module is used for judging the current learning hidden danger type of the classroom according to the number of students meeting the classroom concentration condition; the interaction strategy module is used for determining a corresponding teaching interaction strategy according to the learning hidden danger type; the feedback optimization module is used for determining the memory difficulty coefficient of the test questions, correspondingly adjusting the test question quantity and the test question type proportion, judging the memory absorption degree and the understanding degree of the student according to the test question type, and correspondingly generating a teaching quality optimization reference; and adjusting the judgment standard for judging the class concentration condition and the attention concentration degree according to the question answering result of the test question type. The teaching strategy is adjusted in real time according to the classroom state and the learning condition.

Description

Technical Field

[0001] The present invention relates to the field of teaching quality technology, and in particular to an AI-driven real-time classroom interaction and teaching quality intelligent optimization system and method. Background Art

[0002] In the current educational environment, students' classroom dynamics, such as facial expressions and concentration, have a significant impact on teaching quality and effectiveness. At the same time, due to various factors such as individual differences among students and teachers' teaching methods, the evaluation and regulation of teaching quality face many challenges.

[0003] Chinese Patent Publication No.: CN114677249A discloses an online continuous improvement system for classroom teaching quality based on artificial intelligence technology, including a server, a lecture status management system, an academic affairs management terminal, a mobile terminal and a teacher management terminal. The system uses multiple cameras and body sensors to identify students' lecture status, count the number of students who attend classes (sign in), are late or leave early, and count the names and student numbers of students who look at their phones, doze off, chat with each other, or make loud noises. At the same time, the system can display the students' lecture status and headcount, the lecturer's lecture speed, etc. in real time on the multimedia display screen on the podium, dynamically track students' lecture status, and promptly remind teachers to adopt relevant interactive measures or change (improve) teaching methods. It can be seen that the online continuous improvement system for classroom teaching quality based on artificial intelligence technology has the following problems: It is easy to increase students' learning pressure, resulting in insufficient understanding of knowledge points, thus reducing the learning experience. At the same time, teachers cannot adjust teaching strategies in real time according to students' learning situation, which affects teaching effectiveness. Summary of the invention

[0004] To this end, the present invention provides an AI-driven real-time classroom interaction and teaching quality intelligent optimization system and method to overcome the problems in the prior art that learning pressure is easily increased, resulting in insufficient understanding of knowledge, and teachers are unable to adjust teaching strategies in real time according to students' learning situation.

[0005] To achieve the above objectives, the present invention provides an AI-driven classroom real-time interaction and teaching quality intelligent optimization system, comprising: In-class test terminal, which is used to identify the position of key points, determine the student status according to the angle between the position vectors, retrieve the relevant knowledge points according to the teacher's voice text, and generate in-class test questions according to the relevant knowledge points according to the initial number of test questions; An analysis and evaluation module, which is connected to the in-class test terminal, is divided into a detection stage and an evaluation stage, and is used to record the class duration and detect the determined student status duration in the detection stage, and evaluate whether the student meets the classroom concentration condition according to the initial detection cycle in the evaluation stage, numerically score the student, and determine the current learning hazard type of the classroom according to the proportion of the number of students who meet the classroom concentration condition; An interaction strategy module, which is connected to the analysis and evaluation module, is used to determine the corresponding teaching interaction strategy according to the learning hazard type, evaluate whether the student status is stable to execute the first adjustment instruction or the second adjustment instruction according to the fluctuation of the numerical score, or execute the third adjustment instruction for students who do not meet the attention concentration level; A feedback and optimization module, which is connected to the in-class test terminal, is used to determine the memory difficulty coefficient of the in-class quiz questions and correspondingly adjust the number of questions and the proportion of question types, generate a reference for optimizing teaching quality according to the memory absorption degree and understanding degree of the students judged by the question types, and adjust the standard proportion for determining the classroom concentration condition and the attention concentration level according to the answering results of the question types.

[0006] Further, the in-class test terminal includes a number of learning front-end modules and a teaching voice module, A learning front-end module, which is connected to the in-class test terminal, includes a detection mode and a quiz mode. The detection mode is used to collect face data by collecting the face in the face image located by the student's face, and the quiz mode is used to display the questions determined by the in-class test terminal; A teaching voice module, which is connected to the in-class test terminal, is used to collect and amplify the teacher's voice.

[0007] Further, the in-class test terminal calculates the angle between the head and the horizontal line based on the face data, and compares it with the set angle threshold to determine whether the corresponding student is in a head-up state or a head-down state; The analysis and evaluation module detects the proportion of the number of students who meet the classroom concentration condition, compares it with the set concentration ratio, and determines that the current learning hazard type of the classroom is a hidden learning hazard type or an obvious learning hazard type; The classroom concentration condition is that the ratio of the duration of the head-up state to the class duration is greater than the first standard ratio.

[0008] Further, the analysis and evaluation module outputs the determined learning hazard type to the interaction strategy module according to the initial detection cycle in the analysis stage, and the interaction strategy module generates the corresponding teaching interaction strategy according to the learning hazard type; If the current learning hazard type is a hidden learning hazard type, the interaction strategy module determines that the teaching interaction strategy is to determine whether the student status is stable according to the fluctuation of a number of numerical scores and execute the corresponding adjustment instruction; If the current learning hidden danger type is an obvious learning hidden danger type, the interaction strategy module determines that the teaching interaction strategy is to execute a third adjustment instruction for students who do not meet the attention concentration level, and generates reference information for question-and-answer interaction.

[0009] Further, the interaction strategy module calculates the variance value based on the average score of several data-based scores to determine the fluctuation of the data-based scores. If the variance value is greater than or equal to the standard variance, the interaction strategy module determines that the fluctuation of the data-based scores is large and the student's state is unstable, executes the first adjustment instruction, and generates a reference reminder for group discussion. If the variance value is less than the standard variance, the interaction strategy module determines that the fluctuation of the data-based scores is small and the student's state is stable, executes the second adjustment instruction, and generates a reference reminder for in-class quizzes.

[0010] The condition that the interaction strategy module determines does not meet the attention concentration level is that the duration ratio of the head-down state to the class duration is greater than the second standard ratio, or the duration ratio of the head-turning state to the head-up duration is greater than the third standard ratio.

[0011] Further, the in-class test terminal real-time detects the teaching speed of the teacher, converts the teacher's voice into text, and generates in-class quiz questions according to the initial number of questions. The in-class quiz questions include basic questions and thinking expansion questions. The in-class test terminal generates basic questions according to the knowledge points involved in the text content according to the number of basic questions, or extracts the keywords of the text content and matches thinking expansion questions in the question bank according to the number of expansion questions. The in-class test terminal automatically grades the answers to the in-class quiz questions and generates a grading report. The grading report includes the correct rate of each question and each option, as well as references for optimizing teaching quality.

[0012] Further, the feedback optimization module detects the number of knowledge points retrieved by the in-class test terminal, as well as the teaching duration and difficulty labels corresponding to any knowledge point, and calculates the memory difficulty coefficient in real time. The feedback optimization module evaluates the mastery degree of knowledge points according to the memory difficulty coefficient and the standard coefficient to generate references for optimizing teaching quality, or adjusts the number of questions and the proportion of question types.

[0013] Further, the feedback optimization module determines whether the memory absorption degree of students is qualified according to the basic correct rate of basic questions, and determines whether to generate a reference value for the teaching duration of knowledge points or a reference value for teaching speed according to the expansion correct rate of thinking expansion questions to determine the understanding degree of students. If the basic accuracy rate and the extended accuracy rate are less than the standard accuracy rate, the feedback optimization module increases the first standard ratio, decreases the second standard ratio and the third standard ratio according to the standard accuracy rate, the basic accuracy rate and the extended accuracy rate, and adjusts the determination criteria for the classroom focus condition and the attention concentration level.

[0014] An AI-driven intelligent optimization method for classroom real-time interaction and teaching quality, comprising: Collect student face images, locate the faces in the images, collect face data, and determine the student status according to the face data by judging the students' head-up and head-down situations. Evaluate whether the students meet the classroom focus condition according to the student status, score the students data-based, and determine the current learning hidden danger type of the classroom according to the proportion of the number of students who meet the classroom focus condition. Determine the corresponding teaching interaction strategy according to the learning hidden danger type, that is, evaluate whether the student status is stable according to the fluctuation of the data-based score, execute the corresponding adjustment instruction to generate a reference reminder for group discussion or in-class quiz, or generate a reference message for question interaction for the students who do not meet the attention concentration level. Convert the teacher's voice into text, retrieve the knowledge points involved, generate in-class quiz questions according to the initial number of questions during the in-class quiz, and output the answer results after the quiz to generate a grading report. Determine the memory difficulty coefficient of the in-class quiz questions, and correspondingly adjust the number of questions and the proportion of question types. Judge the memory absorption degree and understanding degree of the students according to the question types, and correspondingly generate a reference for optimizing the teaching quality. Adjust the determination criteria for the classroom focus condition and the attention concentration level according to the answer results of the question types.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: The present invention generates a real-time interaction strategy based on monitoring data. If it is found that the students' attention drops, it automatically pushes interaction questions and organizes group discussions. According to the students' answering situations, it intelligently adjusts the depth and progress of the subsequent teaching content, can enliven the classroom atmosphere, enhance teacher-student interaction, and improve teaching quality; by real-time analyzing the students' expressions, attention concentration levels, etc. in the classroom dynamics, it automatically adjusts the teaching rhythm and organizes interaction sessions, promotes students to actively participate in the classroom, and makes the teaching more targeted and efficient; at the same time, the system records the effects of each interaction and the students' learning achievements, diagnoses the problems that occur in the teaching process, such as weak knowledge point mastery and decreased learning interest, provides personalized diagnostic analysis and optimization suggestions and feedback to the teacher, and the teacher can further optimize the teaching method and interaction strategy based on this to achieve continuous improvement of teaching quality.

[0016] Furthermore, the classroom learning state refers to the degree to which students concentrate and are fully engaged in learning tasks in the classroom. Currently, the classroom learning state relies on the subjective observation of teachers or the self-report of students, which is highly subjective and inaccurate. This system conducts face recognition on students through the corresponding learning front-end module in front of the students, judges the situation of students looking up and down accordingly, and further evaluates whether the students meet the classroom concentration conditions based on the students' states. According to the number of students who meet the classroom concentration conditions, the current learning hazard types in the classroom are determined, including the obvious learning hazard with a relatively large number of students in the head-down state and the hidden learning hazard that seemingly meets the classroom concentration conditions but requires further analysis. This provides a basis for generating corresponding teaching interaction strategies in the follow-up. At the same time, this method of determining the learning hazard types based on the number of eligible students avoids the adverse impact of changes in the student base on the evaluation criteria for learning hazard types caused by the inability of the learning front-end module to detect face data due to equipment or student behavior reasons, thereby improving the accuracy and adaptability of evaluating the classroom state.

[0017] Furthermore, when most students seemingly meet the classroom concentration conditions on the surface, it is often difficult for teachers to detect the learning hazards in the classroom. When the current learning hazard type is the hidden learning hazard type, this invention evaluates the fluctuations in the students' states based on the fluctuations in the data-based scores. When the state fluctuations are unstable, discussions are needed to enliven the classroom atmosphere. When the state fluctuations are stable, in-class quizzes can be conducted to generate corresponding reference reminders for teachers, improving the classroom atmosphere and learning effects, and avoiding the adverse classroom atmosphere from causing students' inattention and thus affecting the learning effects. At the same time, due to the differences in learning habits and attention concentration abilities among different students, when the current learning hazard type is the obvious learning hazard type that is easy to detect, reference reminders for question-and-answer interactions are generated based on the attention concentration degree of each student, effectively improving the classroom participation and learning effects.

[0018] Furthermore, in the existing teaching model, teachers need to prepare for the in-class test by themselves according to the lesson plan, and can often only conduct the test at a fixed teaching node, and cannot evaluate the knowledge mastery degree and feedback the teaching quality according to the test results of the students in real time; the present invention determines the test time node that needs the in-class test according to the current learning hidden danger type and state fluctuation analyzed and evaluated, and conducts the in-class test at the time node when the student state is stable, that is, the attention is most concentrated, so as to ensure that the students have a more solid grasp of the knowledge and improve the learning effect, optimize the teaching rhythm of the teacher to make the teaching content distribution more reasonable and consolidate the classroom knowledge points, and automatically correct and improve the efficiency and fairness of teaching evaluation, and generate a review report including the correctness of each question and each option, so as to assist the teacher to explain the test questions of the in-class test in a targeted manner; at the same time, the number of test questions of the in-class test is adjusted according to the number of knowledge points involved in the retrieved teaching content, the test content and time are flexibly adjusted, and the test questions are divided into basic questions and thinking extension questions, so as to provide a basis for the subsequent evaluation of the students' memory absorption degree and thinking understanding degree of the knowledge points, and reflect the teaching quality of the teacher.

[0019] Furthermore, the key to teaching quality lies in evaluating the students' absorption, memory and understanding degree of knowledge points. The present invention detects the generated in-class tests through a feedback optimization module, and adjusts the number of test questions and the proportion of test question types according to the number of knowledge points and the corresponding difficulty labels as well as the memory difficulty coefficient of the test questions evaluated by the teaching time, so as to avoid the in-class tests taking up too much teaching time; judges the students' memory absorption degree according to the basic questions at different memory difficulties, judges the students' understanding degree according to the thinking extension questions, reminds the teachers to pay attention to the teaching speed according to the students' memory absorption degree, and determines whether to remind the teachers to pay attention to the teaching time allocated for the knowledge point according to the students' understanding degree, so as to avoid the knowledge points with a higher degree of mastery taking up too much teaching time; at the same time, reflects the difficulty of the teaching content according to the basic accuracy rate and the extension accuracy rate of the test questions, adjusts the judgment criteria for judging the classroom concentration conditions and the degree of concentration according to the difficulty of the teaching content, and requires students to be in a state of raising their heads and concentrating their attention when the teaching content is more difficult, so as to improve the adaptability and accuracy of judging the type of classroom hidden dangers and taking corresponding strategies and the degree of concentration. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a unit connection diagram of the classroom real-time interaction and teaching quality intelligent optimization system in an embodiment of the present invention; Figure 2 A flow chart of a method for real-time classroom interaction and intelligent optimization of teaching quality in an embodiment of the present invention; Figure 3 A schematic side view of the shape of a learning front-end module in an embodiment of the present invention; Figure 4 A front view showing the shape of a learning front end module in an embodiment of the present invention; Figure 5 It is a schematic flow chart for determining the types of classroom learning hazards in the embodiments of the present invention; Figure 6 It is a schematic flow chart for generating a reference for optimizing teaching quality according to the memory difficulty coefficient in the embodiments of the present invention.

[0021] In the figure: 1 - display, 2 - camera, 3 - learning front-end module. Detailed implementation manners

[0022] In order to make the objectives and advantages of the present invention clearer and more understandable, the present invention will be further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0023] The preferred implementation manners of the present invention will be described below with reference to the accompanying drawings. Those skilled in the art should understand that these implementation manners are only used to explain the technical principles of the present invention and do not limit the protection scope of the present invention.

[0024] It should be noted that in the description of the present invention, the terms indicating directions or positional relationships such as "upper", "lower", "left", "right", "inner", "outer", etc. are based on the directions or positional relationships shown in the drawings. This is only for the convenience of description and does not indicate or imply that the device or component must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be construed as a limitation of the present invention.

[0025] In addition, it should be noted that in the description of the present invention, unless otherwise clearly defined and limited, the terms "installation", "connection", and "connection" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the communication inside two components. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific situations.

[0026] Please refer to Figures 1-6 as shown in Figure 1 It is a unit connection diagram of the classroom real-time interaction and teaching quality intelligent optimization system in the embodiments of the present invention; Figure 2 It is a schematic flow chart of the classroom real-time interaction and teaching quality intelligent optimization method in the embodiments of the present invention; Figure 3 It is a schematic side view of the shape of the learning front-end module in the embodiments of the present invention; Figure 4 It is a schematic front view of the shape of the learning front-end module in the embodiments of the present invention; Figure 5 It is a schematic flow chart for determining the types of classroom learning hazards in the embodiments of the present invention; Figure 6Schematic flowchart of generating reference for optimizing teaching quality according to memory difficulty coefficient in the embodiments of the present invention.

[0027] The present invention provides an AI-driven classroom real-time interaction and intelligent teaching quality optimization system, including: An in-class test terminal, which is used to identify and locate the positions of key points, determine the student status according to the included angle between position vectors, retrieve the knowledge points involved according to the teacher's voice text, and generate in-class quiz questions according to the initial number of questions and the knowledge points involved; The in-class test terminal includes a number of learning front-end modules and a teaching voice module. A learning front-end module, which is connected to the in-class test terminal, includes a detection mode and a quiz mode. The detection mode is used to collect the face image of the student to locate the face in the image and collect face data. The quiz mode is used to display the questions determined by the in-class test terminal; A teaching voice module, which is connected to the in-class test terminal, is used to collect and amplify the teacher's voice; An analysis and evaluation module, which is connected to the in-class test terminal, is divided into a detection stage and an evaluation stage. It is used to record the class duration and the duration of the determined student status in the detection stage, evaluate whether the student meets the classroom concentration condition according to the initial detection period in the evaluation stage, score the student in a data-driven manner, and determine the current learning hazard type of the classroom according to the proportion of the number of students meeting the classroom concentration condition; An interactive strategy module, which is connected to the analysis and evaluation module, is used to determine the corresponding teaching interaction strategy according to the learning hazard type, evaluate whether the student status is stable according to the fluctuation of the data-driven score, execute the first adjustment instruction or the second adjustment instruction, or execute the third adjustment instruction for students who do not meet the attention concentration level; A feedback and optimization module, which is connected to the in-class test terminal, is used to determine the memory difficulty coefficient of the in-class quiz questions, correspondingly adjust the number of questions and the proportion of question types, generate a reference for optimizing teaching quality according to the memory absorption degree and understanding degree of the student judged by the question type, and adjust the standard proportion of judging the classroom concentration condition and the attention concentration level according to the answer results of the question type.

[0028] Specifically, the present invention generates real-time interaction strategies based on monitoring data. If it is found that the students' attention has declined, interactive questions will be automatically pushed and group discussions will be organized. According to the students' answers, the depth and progress of the subsequent teaching content will be intelligently adjusted, which can enliven the classroom atmosphere, enhance teacher-student interaction, and improve teaching quality. By analyzing students' expressions, attention levels, etc. in the classroom dynamics in real time, the teaching rhythm will be automatically adjusted to organize interactive sessions, promoting students' active participation in the classroom and making teaching more targeted and efficient. At the same time, the system records the effects of each interaction and students' learning achievements, diagnoses problems that occur in the teaching process, such as weak mastery of knowledge points and decreased learning interest, provides personalized diagnostic analysis and optimization suggestions, and feeds them back to the teacher. The teacher can further optimize the teaching methods and interaction strategies based on this to continuously improve the teaching quality.

[0029] In this embodiment, the classroom real-time interaction includes question-and-answer interaction, group discussion, and in-class quiz.

[0030] The learning front-end module 3 is provided with a display 1, and a camera 2 is provided on the top of the display.

[0031] The learning front-end module uses the camera to collect the images of students in front of the display, uses the face detection algorithm to locate the faces in the images, and performs face recognition on the students. In implementation, face positioning can be achieved through various libraries such as OpenCV, Dlib, FaceNet, etc.

[0032] When the in-class test terminal can collect face data by the camera, it calculates the angle between the head and the horizontal line according to the face data. If the angle between the head and the horizontal line is greater than the first angle threshold, the in-class test terminal determines that the student corresponding to the face is in a head-up state or a head-down state. If the angle between the head and the horizontal line is less than or equal to the first angle threshold, the in-class test terminal determines that the student corresponding to the face is in a head-down state. In this embodiment, the first angle threshold is 30°, and the second angle threshold is -30°.

[0033] Specifically, the in-class test terminal sets the eyes and nose in the face image as key points, identifies and locates the positions of the key points, calculates the position vectors of the eyes and the nose, and calculates the angle between the position vectors, and outputs the angle between the position vectors as the angle between the head and the horizontal line.

[0034] The analysis and evaluation module is divided into a detection stage and an evaluation stage. From the start of the learning front-end module to the initial detection period is the detection stage. The analysis and evaluation module records the class duration, the head-up duration of the student corresponding to the face determined by the in-class test terminal in the head-up state, and the head-down duration in the head-down state during the detection stage. In this embodiment, the initial detection period is 15 minutes and the initial detection period is adjustable.

[0035] During the evaluation phase, the analysis and evaluation module determines whether a student meets the classroom concentration condition according to the initial detection period based on the head-up state, numerically scores the students corresponding to each learning front-end module, and determines the current learning hazard type in the classroom according to the number of students who meet the classroom concentration condition. If the ratio of the number of students who meet the classroom concentration condition to the number of students whose faces are captured by the camera is greater than the concentration ratio, the analysis and evaluation module determines that the current learning hazard type in the classroom is a hidden learning hazard type and outputs the classroom learning state. Specifically, the classroom learning state can be displayed in real time as green or good, or the number of students who meet the classroom concentration condition can be output in real time.

[0036] If the ratio of the number of students who meet the classroom concentration condition to the number of students whose faces are captured by the camera is less than or equal to the concentration ratio, the analysis and evaluation module determines that the current learning hazard type in the classroom is an overt learning hazard type. Among them, the concentration ratio is 85% and the value of the concentration ratio is adjustable.

[0037] The classroom concentration condition is that the ratio of the head-up duration in the head-up state to the class duration is greater than the first standard ratio. Among them, the first standard ratio is a preset value set according to the average ratio of the head-up duration to the class duration in the head-up state of students in the publicly available student classroom video dataset, and the first standard ratio can be adjusted according to different age groups and different disciplines.

[0038] Specifically, the classroom learning state refers to the degree to which students concentrate and focus on learning tasks in the classroom. Currently, the classroom learning state depends on the subjective observation of teachers or the self-report of students, which is highly subjective and inaccurate. This system performs face recognition on students through the corresponding learning front-end module in front of the students, determines the head-up and head-down situations of the students accordingly, further evaluates whether the students meet the classroom concentration condition based on the student status, and determines the current learning hazard type in the classroom according to the number of students who meet the classroom concentration condition, including the overt learning hazard with a large number of students in the head-down state and the hidden learning hazard that seemingly has a large number of students meeting the classroom concentration condition but requires further analysis, providing a basis for generating corresponding teaching interaction strategies in the future. At the same time, the method of determining the learning hazard type through the number of students who meet the conditions avoids the adverse impact on the evaluation standard of the learning hazard type caused by the change in the student base due to the failure of the learning front-end module to detect face data due to equipment or student behavior reasons, thereby improving the accuracy and adaptability of evaluating the classroom state.

[0039] It is understandable that the student behaviors refer to behaviors such as absences caused by factors such as illness and short-term departures caused by behaviors such as going to the toilet, which result in the camera not being able to detect the face.

[0040] The analysis and evaluation module digitally scores any student according to the ratio of the head-up duration in the head-up state to the head-down duration in the head-down state; In the analysis stage, the analysis and evaluation module outputs the determined learning hazard types to the interaction strategy module at the initial detection period, and the interaction strategy module generates corresponding teaching interaction strategies according to the learning hazard types. If the current learning hazard type is a hidden learning hazard type, the interaction strategy module determines that the teaching interaction strategy is to determine whether the student state is stable according to the fluctuations of several digital scores, and execute the corresponding adjustment instructions to take corresponding interaction measures; The interaction strategy module calculates the variance value according to the average score of several digital scores to determine the fluctuation of the digital scores. If the variance value is greater than or equal to the standard variance, the interaction strategy module determines that the fluctuation of the digital scores is large and the student state is unstable, executes the first adjustment instruction, and generates a reference reminder for group discussion; If the variance value is less than the standard variance, the interaction strategy module determines that the fluctuation of the digital scores is small and the student state is stable, executes the second adjustment instruction, and generates a reference reminder for in-class quizzes; Specifically, the generated reference reminder can be displayed on the podium or the in-class test terminal, or on the display screen at any position, or input into the teacher's voice module in the form of voice for broadcast, as long as the teacher can receive the reminder information; Among them, the initial detection period is 15 minutes, and the standard ratio is 75%.

[0041] If the current learning hazard type is an obvious learning hazard type, the interaction strategy module determines that the teaching interaction strategy is to execute the third adjustment instruction for students who do not meet the attention concentration level, and generate reference information for question interaction; The interaction strategy module determines that the attention concentration level is not met. If the duration ratio of the head-down duration in the head-down state to the class duration is greater than the second standard ratio, or the duration ratio of the head-turning duration in the head-turning state to the head-up duration is greater than the third standard ratio, the interaction strategy module determines that the attention concentration level is not met, executes the third adjustment instruction, and generates a reference reminder for question interaction; The reference reminder for question interaction is a list of student names arranged according to the second standard ratio or the third standard ratio.

[0042] Specifically, when most students on the surface meet the conditions of classroom concentration, it is often difficult for teachers to detect the learning hazards in the classroom. When the current type of learning hazard is the hidden learning hazard type, this invention evaluates the fluctuations in the state of students based on the fluctuations in the data-based scores. When the state fluctuations are unstable, discussions are needed to enliven the classroom atmosphere. When the state fluctuations are stable, in-class quizzes can be conducted to generate corresponding reference reminders for teachers, improving the classroom atmosphere and learning effects, and avoiding the adverse classroom atmosphere from causing students' inattention and thus affecting learning effects. At the same time, due to the differences in the learning habits and attention concentration abilities of different students, when the current type of learning hazard is the obvious learning hazard type that is easy to detect, reference reminders for question-and-answer interactions are generated based on the attention concentration level of each student, effectively improving classroom participation and learning effects.

[0043] After generating the reference reminder for the in-class quiz, if the learning front-end module switches to the quiz mode, the learning front-end module will display the generated in-class quiz on the monitor. The teaching voice module is used to collect and amplify the teacher's teaching voice in real time. The in-class test terminal detects the teacher's teaching speed in real time according to the teacher's teaching voice, converts the teacher's voice into text, and generates in-class quiz questions according to the initial number of questions based on the text content. The in-class quiz questions include basic questions and thinking expansion questions. The in-class test terminal retrieves the knowledge points involved according to the text content and generates basic questions according to the number of basic questions, or extracts the keywords of the text content and matches thinking expansion questions in the question bank according to the number of expansion questions. Wherein the initial number of questions is the sum of the number of basic questions and the number of expansion questions. Any one of the knowledge points is marked with a difficulty label, and the difficulty label of the thinking expansion question is greater than that of the basic question.

[0044] The in-class test terminal automatically grades the answers to the in-class quiz questions and generates a grading report. The grading report includes the correct rate of each question and each option, as well as references for optimizing teaching quality. Specifically, in the existing teaching model, teachers need to prepare for in-class tests according to the lesson plan, and can often only conduct tests at fixed teaching nodes, and cannot evaluate the knowledge mastery degree of students in real time according to the test results to feedback the teaching quality; the present invention determines the test time node that needs the in-class test according to the current learning risk type and state fluctuation analyzed and evaluated, and conducts the in-class test at the time node when the student state is stable, that is, the attention is most concentrated, to ensure that students have a more solid grasp of knowledge and improve learning effects, optimize the teacher's teaching rhythm to make the teaching content distribution more reasonable and consolidate classroom knowledge points, and automatically correct to improve the efficiency and fairness of teaching evaluation, and generate a review report including the correctness of each question and each option, to assist teachers in explaining the test questions of the in-class test in a targeted manner; at the same time, the number of test questions of the in-class test is adjusted according to the number of knowledge points involved in the retrieved teaching content, the test content and time are flexibly adjusted, and by dividing the in-class test questions into basic questions and thinking extension questions, a basis is provided for subsequent evaluation of students' memory absorption degree and thinking understanding degree of knowledge points, reflecting the teacher's teaching quality.

[0045] The teaching quality optimization reference includes the teaching time and teaching time reference value corresponding to each knowledge point, as well as the teaching speed and teaching speed reference value, and the students' memory absorption degree and thinking understanding degree; The feedback optimization module detects the number of knowledge points retrieved by the in-class test terminal and the teaching time and difficulty label corresponding to any knowledge point, and calculates the ratio of the number of knowledge points and the corresponding difficulty label to the teaching time in real time as a memory difficulty coefficient; If the memory difficulty coefficient is less than or equal to the standard coefficient, the feedback optimization module evaluates the mastery of the knowledge points and generates a teaching quality optimization reference; If the memory difficulty coefficient is greater than the standard coefficient, the feedback optimization module adjusts the number of test questions and the proportion of test question types; Specifically, the feedback optimization module reduces the number of test questions in the class test and the ratio of the number of extended test questions to the number of basic test questions in the class test according to the ratio of the standard coefficient to the memory difficulty coefficient; Among them, the standard coefficient is a preset value set according to the historical memory difficulty coefficient data of the in-class test under the standard accuracy rate, and the standard accuracy rate is 80%.

[0046] The feedback optimization module determines the students' memory absorption degree according to the basic correctness rate of basic questions. If the basic accuracy rate is greater than the standard accuracy rate, the feedback optimization module determines that the student's memory absorption degree is qualified and determines whether to generate a reference value for the teaching time of the knowledge point; If the basic accuracy rate is less than or equal to the standard accuracy rate, the feedback optimization module determines that the student's memory absorption level is unqualified and generates a teaching speed reference value; Specifically, the reference value of the teaching speed is equal to the product of the ratio of the basic correct rate and the standard correct rate and the teaching speed.

[0047] Preferably, the reference teaching speech rate provided in this embodiment is 156 words per minute.

[0048] The feedback optimization module determines the understanding degree of the student according to the extended correct rate of the thinking expansion questions, and determines whether to generate a reference value for the teaching duration of the knowledge points. If the extended correct rate is less than or equal to the standard correct rate, the feedback optimization module determines not to generate a reference value for the teaching duration of the knowledge points; If the extended correct rate is greater than the standard correct rate, the feedback optimization module determines to generate a reference value for the teaching duration of the knowledge points and adjusts the standard coefficient; Specifically, the feedback optimization module increases the standard coefficient according to the ratio of the extended correct rate to the standard correct rate, and the reference value of the teaching duration is equal to the product of the ratio of the standard correct rate to the extended correct rate, the ratio of the standard correct rate to the basic correct rate, and the teaching duration of the knowledge points.

[0049] Among them, if the basic correct rate and the extended correct rate are less than the standard correct rate, the feedback optimization module increases the first standard proportion according to the ratio of the standard correct rate to the sum of the basic correct rate and the extended correct rate. And decreases the second standard proportion and the third standard proportion according to the ratio of the sum of the basic correct rate and the extended correct rate to the standard correct rate.

[0050] Specifically, the key to teaching quality lies in evaluating the degree of absorption, memory, and understanding of knowledge points by students. The present invention detects the generated in-class quizzes through a feedback optimization module, adjusts the number of questions and the proportion of question types according to the number of knowledge points, corresponding difficulty labels, and teaching duration to evaluate the memory difficulty coefficient of the questions, so as to avoid in-class quizzes occupying too much teaching duration; judges the memory absorption degree of students according to basic questions under different memory difficulties, judges the understanding degree of students according to thinking expansion questions, reminds teachers to pay attention to the teaching speed according to the memory absorption degree of students, and determines whether to remind teachers to pay attention to the teaching duration allocated for this knowledge point according to the understanding degree of students, so as to avoid knowledge points with a higher mastery degree occupying too much teaching duration; at the same time, reflects the teaching content difficulty according to the basic correct rate and the extended correct rate of the questions, adjusts the criteria for judging classroom focus conditions and attention concentration degree, and requires students to be in a head-up state and concentrated attention more when the teaching content is more difficult, so as to improve the adaptability and accuracy of judging classroom type hidden dangers and taking corresponding strategies and attention concentration degree.

[0051] The present invention provides an AI-driven method for real-time classroom interaction and intelligent optimization of teaching quality, including: Step S1: Collect students' face images, locate the faces in the images, collect face data, and determine the students' states based on the face data to find out whether the students are looking up or down. Step S2: Evaluate whether the students meet the classroom concentration conditions according to the students' states, give the students a digital score, and determine the current learning risk types in the classroom based on the proportion of students who meet the classroom concentration conditions. Step S3: Determine the corresponding teaching interaction strategies according to the learning risk types, that is, evaluate whether the students' states are stable based on the fluctuations of the digital scores, execute the corresponding adjustment instructions, generate reference reminders for group discussions or in-class quizzes, or generate reference information for question-and-answer interactions for students who do not meet the attention concentration level. Step S4: Convert the teacher's voice into text, retrieve the knowledge points involved, generate in-class quiz questions according to the initial number of questions during the in-class quiz, and output the answer results and generate a grading report after the quiz. Step S5: Determine the memory difficulty coefficient of the in-class quiz questions and correspondingly adjust the number of questions and the proportion of question types. Step S6: Judge the students' memory absorption degree and understanding degree according to the question types and correspondingly generate references for optimizing teaching quality. Step S7: Adjust the judgment criteria for classroom concentration conditions and attention concentration level according to the answer results of the question types.

[0052] Specifically, the present invention aims to enliven the classroom atmosphere, enhance teacher-student interaction, and improve teaching quality; by analyzing students' expressions, attention concentration level, etc. in the classroom dynamics in real time, automatically adjust the teaching rhythm, organize interaction sessions, promote students' active participation in the classroom, and make teaching more targeted and efficient; at the same time, diagnose problems that occur in the teaching process, such as weak mastery of knowledge points and decreased learning interest, provide personalized diagnostic analysis and optimization suggestions, and adjust the teacher's teaching plan.

[0053] So far, the technical solution of the present invention has been described in combination with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the protection scope of the present invention.

[0054] The above are only the preferred embodiments of the present invention and are not used to limit the present invention; for those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent substitution, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. An AI-driven intelligent optimization system for real-time classroom interaction and teaching quality, characterized in that, Including: A classroom test terminal, which is used to identify and locate the key point positions, determine the student status according to the included angle between the position vectors, retrieve the knowledge points involved according to the teacher's speech text, and generate classroom quiz questions according to the initial number of questions based on the knowledge points involved; An analysis and evaluation module, which is connected to the classroom test terminal and is divided into a detection stage and an evaluation stage. It is used to record the class duration and detect the determined student status duration in the detection stage, evaluate whether the student meets the classroom concentration condition according to the initial detection cycle in the evaluation stage, score the student in a data-based manner, and determine the current learning hidden danger type of the classroom according to the proportion of the number of students who meet the classroom concentration condition; An interactive strategy module, which is connected to the analysis and evaluation module, is used to determine the corresponding teaching interactive strategy according to the learning hidden danger type, evaluate whether the student status is stable by executing the first adjustment instruction or the second adjustment instruction according to the fluctuation of the data-based score, or execute the third adjustment instruction for students who do not meet the attention concentration level; A feedback and optimization module, which is connected to the classroom test terminal, is used to determine the memory difficulty coefficient of the classroom quiz questions and correspondingly adjust the number of questions and the proportion of question types, generate a reference for optimizing the teaching quality according to the memory absorption degree and understanding degree of the students judged by the question types, and adjust the standard proportion for judging the classroom concentration condition and the attention concentration level according to the answering results of the question types.

2. The AI-driven real-time classroom interaction and intelligent teaching quality optimization system according to claim 1, wherein The classroom test terminal includes a number of learning front-end modules and a teaching voice module. The learning front-end module, which is connected to the classroom test terminal, includes a detection mode and a quiz mode. The detection mode is used to collect the face data of the students by collecting the face in the face image located in the image, and the quiz mode is used to display the questions determined by the classroom test terminal. The teaching voice module, which is connected to the classroom test terminal, is used to collect and amplify the teacher's voice.

3. The AI-driven classroom real-time interaction and teaching quality intelligent optimization system according to claim 1, wherein The classroom test terminal calculates the included angle between the head and the horizontal line according to the face data, and compares it with the set angle threshold to determine whether the corresponding student is in a head-up state or a head-down state; The analysis and evaluation module detects the proportion of the number of students who meet the classroom concentration condition, compares it with the set concentration ratio, and determines that the current learning hidden danger type of the classroom is a hidden learning hidden danger type or an obvious learning hidden danger type; The classroom concentration condition is that the proportion of the duration of the head-up state to the class duration is greater than the first standard proportion.

4. The AI-driven classroom real-time interaction and teaching quality intelligent optimization system according to claim 3, wherein The analysis and evaluation module outputs the determined learning hidden danger type to the interactive strategy module according to the initial detection cycle in the analysis stage, and the interactive strategy module generates the corresponding teaching interactive strategy according to the learning hidden danger type; If the current learning hidden danger type is a hidden learning hidden danger type, the interactive strategy module determines that the teaching interactive strategy is to determine whether the student status is stable according to the fluctuation of a number of data-based scores and execute the corresponding adjustment instruction; If the current learning hazard type is an overt learning hazard type, the interaction strategy module determines that the teaching interaction strategy is to execute a third adjustment instruction for students who do not meet the attention concentration level, and generates reference information for question-and-answer interaction.

5. The AI-driven classroom real-time interaction and teaching quality intelligent optimization system according to claim 4, wherein the interaction strategy module calculates the variance value based on the average score of several data-based scores to determine the fluctuation of the data-based scores, if the variance value is greater than or equal to the standard variance, the interaction strategy module determines that the fluctuation of the data-based scores is large and the student state is unstable, executes a first adjustment instruction, and generates a reference reminder for group discussion; if the variance value is less than the standard variance, the interaction strategy module determines that the fluctuation of the data-based scores is small and the student state is stable, executes a second adjustment instruction, and generates a reference reminder for in-class quiz.

6. The AI-driven classroom real-time interaction and teaching quality intelligent optimization system according to claim 4, wherein the condition that the interaction strategy module determines does not meet the attention concentration level is that the duration ratio of the head-down state to the class duration is greater than the second standard ratio, or the duration ratio of the head-turning state to the head-up duration is greater than the third standard ratio.

7. The AI-driven classroom real-time interaction and teaching quality intelligent optimization system according to claim 1, wherein the in-class test terminal real-time detects the teaching speech rate of the teacher, converts the teacher's speech into text, and generates in-class quiz questions according to the initial number of questions. The in-class quiz questions include basic questions and thinking expansion questions; the in-class test terminal retrieves the knowledge points involved according to the text content and generates basic questions according to the number of basic questions, or extracts the keywords of the text content and matches thinking expansion questions in the question bank according to the number of expansion questions; the in-class test terminal automatically grades the answers of the in-class quiz questions and generates a grading report. The grading report includes the correct rate of each question and each option, as well as a reference for teaching quality optimization.

8. The AI-driven classroom real-time interaction and teaching quality intelligent optimization system according to claim 7, wherein the feedback optimization module detects the number of knowledge points retrieved by the in-class test terminal, as well as the teaching duration and difficulty label corresponding to any knowledge point, and calculates the memory difficulty coefficient in real time; the feedback optimization module evaluates the mastery degree of knowledge points according to the memory difficulty coefficient and the standard coefficient to generate a reference for teaching quality optimization, or adjusts the number of questions and the proportion of question types.

9. The AI-driven classroom real-time interaction and teaching quality intelligent optimization system according to claim 8, wherein the feedback optimization module determines whether the memory absorption degree of the students is qualified according to the basic correct rate of the basic questions, and determines whether to generate a reference value for the teaching duration of the knowledge points or a reference value for the teaching speed according to the expansion correct rate of the thinking expansion questions to determine the understanding degree of the students; If the basic accuracy rate and the extended accuracy rate are less than the standard accuracy rate, the feedback optimization module increases the first standard proportion, decreases the second standard proportion and the third standard proportion according to the standard accuracy rate, the basic accuracy rate and the extended accuracy rate, and adjusts the judgment criteria for classroom concentration conditions and attention concentration levels.

10. An AI-driven intelligent optimization method for real-time classroom interaction and teaching quality, based on the AI-driven intelligent optimization system for real-time classroom interaction and teaching quality according to any one of claims 1-9, characterized in that Including: Collect the face images of students, locate the faces in the images, collect face data, and determine the student status according to the face data to determine the students' head-up and head-down situations; Evaluate whether the students meet the classroom concentration conditions according to the student status, score the students in a data-based manner, and determine the current learning hidden danger type of the classroom according to the proportion of the number of students who meet the classroom concentration conditions; Determine the corresponding teaching interaction strategy according to the learning hidden danger type, that is, evaluate whether the student status is stable according to the fluctuation of the data-based score, execute the corresponding adjustment instruction to generate a reference reminder for group discussion or in-class quiz, or generate a reference message for question interaction for students who do not meet the attention concentration level; Convert the teacher's voice into text, retrieve the knowledge points involved, generate in-class quiz questions according to the initial number of questions during the in-class quiz, and output the answer results after the end to generate a grading report; Determine the memory difficulty coefficient of the in-class quiz questions and correspondingly adjust the number of questions and the proportion of question types; Judge the memory absorption degree and understanding degree of the students according to the question types and correspondingly generate a reference for optimizing teaching quality; Adjust the judgment criteria for classroom concentration conditions and attention concentration levels according to the answer results of the question types.

Images