Online education interactive teaching method, device, equipment and medium

Abstract

The invention relates to an online education interactive teaching method and device, equipment and a medium. The method comprises the following steps: acquiring student face image data and current teaching knowledge point data, and calculating student attention parameters according to the image data to obtain student attention parameter data; knowledge point association degree data is generated through the current teaching knowledge point data; dynamic teaching adjustment strategy data is generated based on the student attention parameter data and the knowledge point correlation degree data, teaching feedback operation is executed according to the strategy data, a teaching feedback execution result is obtained, and the teaching feedback operation comprises interface thermodynamic diagram superposition and micro-test insertion. According to the method, through the facial image data of the students and the teaching knowledge point data, not only can the attention of the students be accurately quantified and the association degree of the knowledge points be scientifically evaluated, but also a dynamic teaching strategy can be generated and feedback operation including thermodynamic diagram superposition and micro-test insertion can be executed, so that the pertinence, interactivity and effectiveness of online teaching are remarkably improved.

Description

Technical Field

[0001] The present invention belongs to the field of computer technology, and particularly relates to an online education interactive teaching method, device, equipment and medium. Background Art

[0002] With the development of technology in the field of online education, remote teaching technology based on the Internet has emerged. This technology breaks the limitations of time and space through the Internet, enabling teachers and students to carry out teaching activities without being in the same physical space, and having characteristics such as convenient sharing of teaching resources and high learning flexibility. However, in traditional online education interactive teaching strategies, fixed teaching processes and modes are mostly adopted, that is, teachers teach according to pre-set lesson plans, transmit knowledge through means such as text, voice, and video, students complete homework and tests after class, and teachers give feedback based on the results of homework and tests. For students' classroom attention, it mainly relies on subjective observation by teachers, such as observing students' speaking situations, question answering situations, etc., and it is difficult to comprehensively and accurately grasp the concentration of each student in class. Moreover, for the importance of knowledge points and students' mastery levels, most are based on teachers' experience judgments, with strong subjectivity, and it is impossible to conduct personalized teaching according to students' actual learning situations, thus affecting teaching effects and students' learning efficiency. Summary of the Invention

[0003] Based on this, in view of the above technical problems, it is necessary to provide an online education interactive teaching method, device, equipment and medium to improve the accuracy and personalization of teaching, and enhance students' learning participation and teaching effects.

[0004] In a first aspect, the present application provides an online education interactive teaching method, including:

[0005] Obtaining student facial image data at the student side and current teaching knowledge point data at the teacher side;

[0006] Calculating student attention parameters based on the student facial image data to obtain student attention parameter data, where the student attention parameter data includes eyeball focus offset angle data and corrected facial orientation angle data;

[0007] Generating knowledge point correlation data through the current teaching knowledge point data, and the knowledge point correlation data is calculated by the proportion of the number of questions related to the current knowledge point in the historical wrong question data;

[0008] Generating dynamic teaching adjustment strategy data based on the student attention parameter data and the knowledge point correlation data, and performing a teaching feedback operation according to the dynamic teaching adjustment strategy data to obtain a teaching feedback execution result, where the teaching feedback operation includes interface heat map overlay and micro-test insertion.

[0009] In one embodiment, obtaining student facial image data at the student end and current teaching knowledge point data at the teacher end, including:

[0010] Obtaining student facial image data through the camera at the student end, and calculating the original facial orientation angle data according to the physical position deviation parameters between the camera and the screen. The physical position deviation parameters include the horizontal offset between the camera and the center of the screen and the camera focal length;

[0011] Performing correction processing on the original facial orientation angle data and the physical position deviation parameters through a preset geometric correction formula to obtain the corrected facial orientation angle data;

[0012] Obtaining the screen sharing content based on the teaching software at the teacher end, extracting and verifying the current teaching knowledge point label data through a preset teaching syllabus database, and using the verified current teaching knowledge point label data as the current teaching knowledge point data.

[0013] In one embodiment, calculating the student attention parameter based on the student facial image data to obtain the student attention parameter data. The student attention parameter data includes the eyeball focus offset angle data and the corrected facial orientation angle data, including:

[0014] Performing pupil positioning processing on the student facial image data, extracting the pupil center coordinate data, and calculating the original eyeball focus offset angle according to the position deviation between the pupil center coordinate data and the preset screen calibration point;

[0015] Performing head pose analysis on the student facial image data to obtain the head pitch angle data. When it is detected that the head pitch angle data exceeds the preset angle, based on the original eyeball focus offset angle data and the head pitch angle data, performing correction processing through a compensation formula to obtain the corrected eyeball focus offset angle data;

[0016] Inputting the corrected eyeball focus offset angle data, the corrected facial orientation angle data, and the interaction event frequency data into a multimodal fusion algorithm for weighted calculation to generate the group attention score data;

[0017] Performing attenuation correction processing on the group attention score data according to the elapsed time data of the course to generate the group attention correction value data, and combining the corrected eyeball focus offset angle data, the corrected facial orientation angle data, and the group attention correction value data to generate the student attention parameter data.

[0018] In one embodiment, inputting the corrected eyeball focus offset angle data, the corrected facial orientation angle data, and the interaction event frequency data into a multimodal fusion algorithm for weighted calculation to generate the group attention score data, including:

[0019] Calculate the eye focus offset influence factor data through the sigmoid function according to the corrected eye focus offset angle data and the preset eye offset threshold;

[0020] Calculate the facial orientation cosine value data through the cosine similarity algorithm based on the corrected facial orientation angle data and the preset screen positive direction reference data;

[0021] Calculate the interaction growth rate data through the interaction event frequency data;

[0022] Perform weighted summation on the eye focus offset influence factor data, facial orientation cosine value data, and interaction growth rate data according to the preset weight coefficients to generate the group attention score data.

[0023] In one embodiment, generate dynamic teaching adjustment strategy data according to the student attention parameter data and the knowledge point correlation data, including:

[0024] Based on the group attention correction value data, calculate the attention distribution variance data of the students. When the group attention correction value data is lower than 0.4 and the knowledge point correlation data exceeds 0.7, generate the knowledge point repeated explanation instruction data;

[0025] When the attention distribution variance data is continuously higher than 0.2 for 5 minutes, generate the random group discussion instruction data;

[0026] Dynamically adjust the priority weights of the knowledge point repeated explanation instruction data and the random group discussion instruction data according to the historical teaching effect data. The historical teaching effect data includes the student knowledge point retention rate and the interaction participation rate;

[0027] Perform conflict detection and resource occupancy rate assessment on the knowledge point repeated explanation instruction data and the random group discussion instruction data after the priority weight adjustment to generate the dynamic teaching adjustment strategy data. Among them, the conflict detection includes judging the overlap of the instruction execution time window and the hardware device load conflict;

[0028] The resource occupancy rate assessment includes calculating the CPU occupancy rate of the teacher side and the network bandwidth occupancy rate of the student side.

[0029] In one embodiment, perform teaching feedback operations according to the dynamic teaching adjustment strategy data to obtain the teaching feedback execution result. The teaching feedback operations include interface heat map overlay and micro-test insertion, including:

[0030] Calculate the transparency data of the teacher side interface heat map according to the low attention student ratio data in the dynamic teaching adjustment strategy data. The low attention student ratio data is calculated by the proportion of the number of students whose group attention correction value data is lower than 0.4;

[0031] Generate bone conduction pulse signal intensity data based on the knowledge point correlation data and the current speaking speed data of the teacher side;

[0032] Extract the student historical correct rate data through the historical wrong question data, and calculate the attention change rate data based on the time series of the group attention correction value data;

[0033] Generate micro-test difficulty data according to the student historical correct rate data and the attention change rate data;

[0034] Send the transparency data, the bone conduction pulse signal intensity data, and the micro-test difficulty data to the student side and the teacher side, and receive the feedback data of the micro-test loading completion rate from the student side and the feedback data of the heat map rendering delay from the teacher side, and generate the teaching feedback execution result.

[0035] In one embodiment, perform attenuation correction processing on the group attention score data according to the course elapsed time data to generate group attention correction value data, including:

[0036] Calculate the course progress ratio data according to the total course duration data and the course elapsed time data;

[0037] When the course progress ratio data exceeds 0.7, calculate the time attenuation factor according to the preset formula, and the preset formula is:

[0038] λ=0.02×(t - 0.7T total )

[0039] where λ is the time attenuation factor, T total is the total course duration data, and t is the course elapsed time data;

[0040] Perform non-linear mapping on the time attenuation factor according to the sigmoid function to generate the attenuation correction coefficient;

[0041] Use the attenuation correction coefficient to correct the group attention score data to generate group attention correction value data.

[0042] In the second aspect, the present application also provides an online education interactive teaching device, including:

[0043] A data acquisition module for acquiring the student face image data of the student side and the current teaching knowledge point data of the teacher side;

[0044] An attention analysis module for calculating student attention parameters according to the student face image data to obtain student attention parameter data, and the student attention parameter data includes the eye focus offset angle data and the corrected face orientation angle data;

[0045] A knowledge association module, which is used to generate knowledge point association degree data through the current teaching knowledge point data. The knowledge point association degree data is calculated by the proportion of the number of questions related to the current knowledge point in the historical wrong question data.

[0046] A teaching optimization module, which is used to generate dynamic teaching adjustment strategy data based on the student attention parameter data and the knowledge point association degree data, and execute teaching feedback operations according to the dynamic teaching adjustment strategy data to obtain the teaching feedback execution result. The teaching feedback operations include interface heat map overlay and micro-test insertion.

[0047] In a third aspect, the present application also provides a computer device, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the steps in the first aspect are implemented.

[0048] In a fourth aspect, the present application also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps in the first aspect are implemented.

[0049] For the above online education interactive teaching method, device, equipment and medium, firstly, by obtaining the student face image data at the student end and the current teaching knowledge point data at the teacher end, it is possible to comprehensively collect the student learning status information and teaching content information, providing a data basis for subsequent teaching analysis and adjustment. Secondly, calculating the student attention parameters according to the student face image data can accurately quantify the student's attention in the classroom from multiple dimensions. Moreover, the method generates knowledge point association degree data through the current teaching knowledge point data, and this data is calculated by the proportion of the number of questions related to the current knowledge point in the historical wrong question data, which can clearly reflect the difficulty and importance of the current knowledge point for students, providing a basis for determining teaching key points and allocating teaching resources, and effectively avoiding the blindness of teaching. In addition, generating dynamic teaching adjustment strategy data based on the student attention parameter data and the knowledge point association degree data can flexibly adjust teaching strategies according to the student's real-time learning status and the actual situation of knowledge points, making the teaching process more in line with the needs of students and improving the pertinence and effectiveness of teaching.

[0050] Finally, operations such as interface heat map overlay and micro-test insertion are executed according to the dynamic teaching adjustment strategy data to obtain the teaching feedback execution result. Among them, the interface heat map overlay can visually display the attention distribution of students, helping teachers quickly understand the overall classroom state, and through the micro-test insertion, it is possible to timely check the student's mastery of knowledge points, facilitating teachers to timely adjust the teaching progress and methods. This process can not only enable teachers to timely master the teaching effect, but also timely discover problems existing in the teaching process and make improvements, further improving the teaching quality.

[0051] Compared with the traditional online education teaching method, this method effectively solves the problems of being unable to accurately understand students' attention, difficult to determine teaching key points, and lack of flexibility in teaching strategies, realizes the precision and personalization of teaching, and enhances the teaching effect of online education and students' learning experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following will briefly introduce the drawings required for use in the description of the embodiments or related technologies. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0053] Figure 1 A method flowchart of an online education interactive teaching method provided for an exemplary embodiment of the present invention;

[0054] Figure 2 A schematic structural diagram of an online education interactive teaching device provided for an exemplary embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0055] In order to make the purpose, technical solutions and advantages of the present application more clear, the following further details the present application in conjunction with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0056] In one embodiment, as Figure 1 shown, an online education interactive teaching method is provided. In this embodiment, it is exemplified that the method is applied to a terminal. It can be understood that the method can also be applied to a server, and can also be applied to a device including a terminal and a server, and is realized through the interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0057] S101: Obtain the student facial image data at the student end and the current teaching knowledge point data at the teacher end.

[0058] Specifically, the camera at the student end can capture the student's facial image data at a preset frequency, such as 30 frames per second. And the teaching content at the teacher end can be obtained through the teacher end device, such as teaching software, so as to extract the current teaching knowledge point data. The current teaching knowledge point data can be the knowledge point text, picture or video content that the teacher is explaining. By obtaining the student facial image data and the current teaching knowledge point data, the relevance between the student's learning state and the teaching content can be perceived in real time, providing basic data support for subsequent attention analysis and teaching optimization.

[0059] S102: Calculate the student attention parameters based on the student facial image data to obtain the student attention parameter data, where the student attention parameter data includes the eye focus offset angle data and the corrected facial orientation angle data.

[0060] Specifically, based on the student facial image data, facial features such as the student's eye position and facial orientation can be analyzed through an image recognition algorithm, such as a facial pose estimation model based on deep learning, and then the student attention parameter data can be output. Among them, the offset angle between the eye focus and the center of the screen can be calculated by identifying the focus position of the student's eyes to obtain the eye focus offset angle data. The smaller the offset angle, the more concentrated the student's attention; the larger the offset angle, the more likely the student's attention is scattered. In addition, the angle between the facial orientation and the screen can be calculated by analyzing the student's facial orientation, and the student's facial orientation can be corrected in combination with historical data to obtain the corrected facial orientation angle data. This corrected angle data can more accurately reflect the student's attention state.

[0061] S103: Generate knowledge point correlation data through the current teaching knowledge point data, and the knowledge point correlation data is calculated by the proportion of the number of questions related to the current knowledge point in the historical wrong question data.

[0062] Specifically, the wrong question data related to the current knowledge point can be extracted from the student's historical learning records. These wrong question data can be the questions that the student got wrong in historical tests or exercises. And the number of wrong questions related to the current teaching knowledge point can be counted, and the proportion of it in the total number of the student's historical wrong questions can be calculated. For example, if the current knowledge point is "trigonometric functions", the number of wrong questions in the trigonometric function related questions of the student can be counted and the proportion can be calculated to obtain the knowledge point correlation data. The higher the correlation, the more obvious the weak link of the student in the current knowledge point.

[0063] S104: Generate dynamic teaching adjustment strategy data based on the student attention parameter data and the knowledge point correlation data, and perform teaching feedback operations according to the dynamic teaching adjustment strategy data to obtain the teaching feedback execution result, where the teaching feedback operations include interface heat map overlay and micro-test insertion.

[0064] Specifically, the comprehensive score can be calculated by combining the student attention parameters and the knowledge point correlation data through a preset weight distribution algorithm. Based on this comprehensive score, the corresponding dynamic teaching adjustment strategy data can be generated. For example, if the comprehensive score is low, it indicates that the students' attention is scattered and their mastery of knowledge points is poor, and a more frequent micro-test insertion strategy can be generated; if the comprehensive score is high, the micro-test insertion frequency can be reduced, and the content overlay of the interface heat map can be optimized. Schematically, this interface heat map is generated based on the student attention parameter data and can intuitively display the areas where the students' attention is concentrated. Therefore, teachers can adjust the display method of teaching content according to this heat map, for example, placing key content in the areas where the students' attention is concentrated. The micro-test insertion is based on the knowledge point correlation data and inserts micro-test questions related to the current knowledge point. These micro-test questions can be multiple-choice questions, fill-in-the-blank questions, or short-answer questions, so that students can consolidate the knowledge points, and teachers can understand the students' learning status in real time through the students' answering situations.

[0065] In the above method, by obtaining the student facial image data on the student side and the current teaching knowledge point data on the teacher side, it effectively makes up for the deficiency that traditional online education is difficult to comprehensively grasp the students' learning status and teaching key points, and can comprehensively integrate the external performance information of students in the classroom and the key information of teaching content, thus providing a comprehensive data basis for subsequent teaching analysis and adjustment. Secondly, calculating the student attention parameter data including the eye focus offset angle data and the corrected facial orientation angle data based on the student facial image data can not only accurately quantify the students' attention in the classroom, but also comprehensively consider the students' concentration from multiple dimensions, avoiding the one-sidedness and limitations of relying solely on the teacher's subjective observation.

[0066] Moreover, by calculating the knowledge point correlation data through the proportion of the number of questions related to the current teaching knowledge point data in the historical wrong question data, it can accurately reflect the difficulty and importance of the current knowledge point for students, effectively improving the ability to grasp the teaching key points and difficulties. Finally, generating the dynamic teaching adjustment strategy data based on the student attention parameter data and the knowledge point correlation data, and performing the teaching feedback operation according to the dynamic teaching adjustment strategy data to obtain the teaching feedback execution result. Through this process, it can provide teachers with intuitive feedback on the students' attention distribution and knowledge mastery, timely discover the problems existing in the teaching process and make adjustments, and thus improve the teaching quality.

[0067] Compared with the traditional online education method, this method further improves the teaching quality and interaction effect of online education through multi-source data acquisition, attention analysis, knowledge point correlation calculation, and dynamic teaching adjustment and feedback.

[0068] In one embodiment, obtaining student facial image data at the student side and current teaching knowledge point data at the teacher side, including:

[0069] Obtaining student facial image data through the camera at the student side, and calculating the original facial orientation angle data according to the physical position deviation parameters between the camera and the screen. The physical position deviation parameters include the horizontal offset between the camera and the center of the screen and the camera focal length;

[0070] Performing correction processing on the original facial orientation angle data according to the physical position deviation parameters through a preset geometric correction formula to obtain the corrected facial orientation angle data;

[0071] Obtaining the screen sharing content based on the teaching software at the teacher side, extracting and verifying the current teaching knowledge point label data through a preset teaching syllabus database, and using the verified current teaching knowledge point label data as the current teaching knowledge point data.

[0072] Specifically, the horizontal offset between the camera and the center of the screen refers to the position difference of the camera relative to the center of the screen in the horizontal direction. The camera focal length is an important parameter of the camera optical system, which determines the distance range in which the camera can form a clear image and the image scaling ratio. Based on this physical position deviation parameter, the position and optical characteristics of the camera, as well as the feature information in the student facial image data obtained from the camera, can be comprehensively considered to initially determine the orientation angle of the student's face relative to the screen. However, since the original facial orientation angle data is affected by the physical position deviation, it can be corrected through a preset geometric correction formula. This formula can be derived based on optical principles and geometric relationships, and can then convert the original facial orientation angle data into a more accurate corrected facial orientation angle data according to the relative position and optical characteristics of the camera and the screen. At the teacher side, the screen sharing content can be obtained through the teaching software. This content contains various information displayed by the teacher during the teaching process, such as text, pictures, charts, etc. The knowledge point label data related to the current teaching can be extracted from these contents. For example, when explaining a math course, knowledge point labels such as "domain of a function" and "monotonicity of a function" may be extracted. And the extracted current teaching knowledge point label data is verified through a preset teaching syllabus database to check whether it meets the requirements and logical relevance of the teaching syllabus. The teaching syllabus database is a pre-established knowledge base that contains all the knowledge points of the course and the logical relationships between the knowledge points.

[0073] In one embodiment, calculating the student attention parameter according to the student facial image data to obtain the student attention parameter data. The student attention parameter data includes the eye focus offset angle data and the corrected facial orientation angle data, including:

[0074] Perform pupil localization processing based on the student's facial image data, extract the pupil center coordinate data, and calculate the original eyeball focus offset angle based on the position deviation between the pupil center coordinate data and the preset screen calibration points;

[0075] Perform head pose analysis based on the student's facial image data to obtain the head pitch angle data. When it is detected that the head pitch angle data exceeds the preset angle, based on the original eyeball focus offset angle data and the head pitch angle data, perform correction processing through a compensation formula to obtain the corrected eyeball focus offset angle data;

[0076] Input the corrected eyeball focus offset angle data, the corrected facial orientation angle data, and the interaction event frequency data into a multi-modal fusion algorithm for weighted calculation to generate group attention score data;

[0077] Perform attenuation correction processing on the group attention score data according to the course elapsed time data to generate group attention correction value data, and combine the corrected eyeball focus offset angle data, the corrected facial orientation angle data, and the group attention correction value data to generate student attention parameter data.

[0078] Specifically, after obtaining the student's facial image data, operations such as image enhancement, denoising, and grayscaling can be performed on it to further improve the image quality. Subsequently, a deep learning-based object detection model can be used to localize the pupils in the image and extract the coordinate data of the pupil center. This pupil center coordinate data is usually represented in two-dimensional coordinates, and it can be compared with the preset screen calibration points to calculate the position deviation between the pupil center coordinate and the calibration points, obtaining the original eyeball focus offset angle data. Schematically, the preset screen calibration points are reference points determined by the student by gazing at a specific point on the screen, such as the center of the screen, during the initial calibration stage. And by inputting the facial image data into a deep learning-based head pose estimation model, key points of the face such as eyes, nose, mouth, etc. can be detected to estimate the head pitch angle, yaw angle, and roll angle, obtaining the head pitch angle data. This head pitch angle data reflects the tilt angle of the student's head in the vertical direction. When it is detected that the head pitch angle exceeds the preset angle, for example, exceeds 30 degrees, that is, the change in the head pose may affect the accuracy of the eyeball focus offset angle, it can be corrected through a compensation formula to correct the eyeball focus offset angle data.

[0079] In addition, the corrected eye focus deviation angle data, the corrected facial orientation angle data, and the interactive event frequency data can be weighted and fused to calculate the group attention score data. And as the course time progresses, the attention of students usually gradually decreases. Therefore, the attention score can be dynamically adjusted. Specifically, the group attention score data can be attenuated and corrected according to the course elapsed time data to calculate the group attention correction value data. Finally, combining the corrected eye focus deviation angle data, the corrected facial orientation angle data, and the group attention correction value data, the student attention parameter data is generated. This data can comprehensively reflect the attention state of students and provide a scientific basis for subsequent teaching optimization.

[0080] In one embodiment, the corrected eye focus deviation angle data, the corrected facial orientation angle data, and the interactive event frequency data are input into a multi-modal fusion algorithm for weighted calculation to generate the group attention score data, including:

[0081] According to the corrected eye focus deviation angle data and a preset eye deviation threshold, the eye focus deviation influence factor data is calculated through the sigmoid function;

[0082] Based on the corrected facial orientation angle data and the preset screen positive direction reference data, the facial orientation cosine value data is calculated through the cosine similarity algorithm;

[0083] The interactive growth rate data is calculated through the interactive event frequency data;

[0084] The eye focus deviation influence factor data, the facial orientation cosine value data, and the interactive growth rate data are weighted and summed according to a preset weight coefficient to generate the group attention score data.

[0085] Specifically, the corrected eye focus deviation angle data reflects the actual focus deviation of the students' eyes in the classroom. The preset eye deviation threshold is a reference standard to judge whether the degree of eye deviation exceeds the normal range. For example, under normal circumstances, if the eyes of a student move within a certain area, it is considered focused, and exceeding this area may mean distraction, and the angle corresponding to the boundary of this area is the eye deviation threshold. The sigmoid function has the characteristic of mapping data between 0 and 1, and is suitable for converting the eye focus deviation situation into an influencing factor. Through the conversion of this function, the influence degree of eye focus deviation on students' attention can be intuitively quantified. For example, when the corrected eye focus deviation angle is close to or less than the eye deviation threshold, it indicates that the degree of eye focus deviation of the student is small, the attention is relatively concentrated, and the data of the influencing factor of eye focus deviation calculated by the sigmoid function is close to 1. The corrected facial orientation angle data accurately reflects the orientation angle of the student's face relative to the screen. The preset screen positive direction reference data is a set standard direction, and the central axis direction of the screen can be used as the positive direction reference. Subsequently, the cosine similarity algorithm can be used to calculate the cosine value of the included angle between the two vectors in the corrected facial orientation angle data and the screen positive direction reference data to reflect the influence of the student's facial orientation on attention.

[0086] In addition, the interactive event frequency data records the frequency of students' participation in interactions during the class, such as the statistics of the number of times of interactive behaviors such as students answering questions, raising hands to speak, and participating in group discussions. The interactive growth rate data is obtained by further processing the interactive event frequency data within a preset interval. A higher interactive growth rate means that the enthusiasm of students to participate in interactions is increasing, reflecting that students are more concentrated in attention and have a higher participation rate in the class. By weighted summing the data of the influencing factor of eye focus deviation, the cosine value data of facial orientation, and the interactive growth rate data according to the preset weight coefficients, the influence of multiple factors on students' attention is comprehensively considered, and the finally generated group attention score data can more comprehensively and accurately reflect the attention status of the student group in the class, providing a strong basis for the adjustment of subsequent teaching strategies.

[0087] In one embodiment, dynamic teaching adjustment strategy data is generated based on the student attention parameter data and the knowledge point relevance data, including:

[0088] Based on the group attention correction value data, calculate the attention distribution variance data of the students. When the group attention correction value data is lower than 0.4 and the knowledge point relevance data exceeds 0.7, generate the knowledge point repeated explanation instruction data;

[0089] When the attention distribution variance data is continuously higher than 0.2 for 5 minutes, generate the random grouping discussion instruction data;

[0090] Dynamically adjust the priority weights of the knowledge point repeated explanation instruction data and the random group discussion instruction data according to the historical teaching effect data, where the historical teaching effect data includes the student knowledge point retention rate and the interaction participation rate;

[0091] Conduct conflict detection and resource occupancy rate assessment on the knowledge point repeated explanation instruction data and the random group discussion instruction data after the priority weight adjustment to generate dynamic teaching adjustment strategy data. Among them, the conflict detection includes judging the overlap of the instruction execution time window and the hardware device load conflict;

[0092] The resource occupancy rate assessment includes calculating the CPU occupancy rate of the teacher side and the network bandwidth occupancy rate of the student side.

[0093] Specifically, the group attention correction value data reflects the degree of attention of students on the current teaching content. The attention distribution variance data can be calculated through this data. The attention distribution variance data is used to evaluate the degree of attention of the student group. The smaller the variance, the more concentrated the attention of the student group; the larger the variance, the more dispersed the attention distribution of the student group. And during the teaching process, the group attention correction value data and the knowledge point correlation data can be monitored to adjust the teaching strategy. Schematically, when the group attention correction value data is lower than 0.4, it indicates that the attention of the student group is generally low, and when the knowledge point correlation data exceeds 0.7, it indicates that the current knowledge point has a high correlation with the historical wrong question data of the students, and the students may have weak links in this knowledge point. When both of the above conditions are met at the same time, it can be considered that it is necessary to improve the students' attention and knowledge point mastery by repeating the explanation of the current knowledge point, that is, generate the knowledge point repeated explanation instruction data. The instruction data includes the number of repeated explanations, the explanation methods such as animation demonstration, example explanation, etc., and the explanation time window.

[0094] In addition, in this embodiment, the attention distribution variance data can be continuously monitored. When the attention distribution variance data is higher than 0.2 for 5 consecutive minutes, it can be considered that the attention distribution of the student group is too scattered, that is, the participation of students can be improved through interactive means. Schematically, random grouping discussion instruction data can be generated to improve students' attention and knowledge mastery ability through interaction. The instruction data can include grouping methods such as random grouping, grouping according to knowledge mastery, discussion topics, and discussion time windows. Moreover, the priority weights of the knowledge point repeated explanation instruction data and the random grouping discussion instruction data can be dynamically adjusted according to the historical teaching effect data to ensure the scientificity and effectiveness of the teaching adjustment strategy. Among them, the historical teaching effect data includes the knowledge point retention rate and interaction participation rate of students. Subsequently, conflict detection and resource occupancy rate evaluation can also be performed on the knowledge point repeated explanation instruction data and the random grouping discussion instruction data after the priority weight adjustment to ensure the feasibility and effectiveness of the teaching strategy. For example, in conflict detection, it can be checked whether the execution time windows of the knowledge point repeated explanation instruction and the random grouping discussion instruction overlap, or the hardware device load conditions of the teacher side and the student side can be checked to ensure that the instruction execution will not cause device overload. In the resource occupancy rate evaluation, the CPU occupancy rate of the teacher side device during the execution of the instruction can be calculated to ensure that the CPU occupancy rate does not exceed the usage threshold, or the network bandwidth occupancy rate of the student side device during the execution of the instruction can be calculated to ensure that the network bandwidth occupancy rate does not exceed the occupancy threshold. Through conflict detection and resource occupancy rate evaluation, dynamic teaching adjustment strategy data can be obtained. The strategy data can include detailed information such as the execution order, time window, and resource allocation of the instruction to ensure the smooth execution of the teaching adjustment strategy.

[0095] In one embodiment, a teaching feedback operation is performed according to the dynamic teaching adjustment strategy data to obtain a teaching feedback execution result. The teaching feedback operation includes interface heat map overlay and micro-test insertion, including:

[0096] According to the low-attention student proportion data in the dynamic teaching adjustment strategy data, calculate the transparency data of the teacher-side interface heat map. The low-attention student proportion data is calculated by the proportion of the number of students whose group attention correction value data is lower than 0.4;

[0097] Generate bone conduction pulse signal intensity data based on the knowledge point correlation data and the current speaking speed data of the teacher side;

[0098] Extract the student historical correct rate data through the historical wrong question data, and calculate the attention change rate data based on the time series of the group attention correction value data;

[0099] Generate micro-test difficulty data based on the student historical correct rate data and the attention change rate data;

[0100] Send transparency data, bone conduction pulse signal intensity data, and micro-test difficulty data to the student side and the teacher side, and receive the feedback data of the micro-test loading completion rate from the student side and the feedback data of the heat map rendering delay from the teacher side to generate the teaching feedback execution result.

[0101] Specifically, the low-attention student proportion data in the dynamic teaching adjustment strategy data is a key indicator for judging the overall attention status of students in the classroom. This low-attention student proportion data can be calculated by the proportion of the number of students whose group attention correction value data is lower than 0.4. If the low-attention student proportion is relatively high, it means that a relatively large number of students in the classroom are inattentive. To enable teachers to more clearly and intuitively perceive this situation, the transparency of the heat map on the teacher side can be correspondingly reduced, making the colors representing the low-attention areas more prominent; conversely, if the low-attention student proportion is relatively low, indicating that most students are relatively concentrated, the transparency of the heat map can be relatively increased, and the display effect of the low-attention areas is relatively weakened. Through this process, it helps teachers quickly understand the attention distribution of students from the interface heat map so as to adjust teaching strategies in a timely manner.

[0102] The knowledge point correlation data reflects the degree of closeness between the currently taught knowledge point and the relevant questions in the student's historical wrong question data, and can reflect the importance and difficulty level of this knowledge point for students. The current speaking speed data on the teacher side represents the language expression speed of the teacher during the teaching process. The bone conduction pulse signal intensity data can be generated by combining the knowledge point correlation data and the current speaking speed data on the teacher side. For example, when the knowledge point correlation is high and the teacher's speaking speed is relatively fast, it indicates that the content being explained is important and the progress is a bit fast. To remind students to concentrate, the bone conduction pulse signal intensity can be correspondingly increased. Through devices such as bone conduction headphones, students can feel a stronger pulse reminder, prompting them to pull their attention back to the classroom content; conversely, if the knowledge point correlation is low or the teacher's speaking speed is normal and gentle, the bone conduction pulse signal intensity will be weakened to avoid unnecessary interference to students. Through the dynamic adjustment of the bone conduction pulse signal intensity, it can effectively assist teachers in guiding students' attention without affecting the classroom order.

[0103] Specifically, from the historical wrong-question data, the historical correct rate data of students can be extracted. This data details the correct answering ratios of students for questions related to various knowledge points during the past learning process, and can intuitively reflect the students' mastery of different knowledge points. Moreover, since the group attention correction value data changes over time during the class process, by analyzing its change trend within a certain time period, the difference in the group attention correction value data between adjacent time points can be calculated, and combined with the time interval, the attention change rate data can be obtained. If the attention change rate is positive and the value is large, it indicates that the attention of the student group is gradually increasing; if it is negative and the value is large, it means that the attention of the student group is showing a downward trend. In addition, the historical correct rate data of students and the attention change rate data are important bases for generating the micro-test difficulty data. Schematically, if the historical correct rate of students is high and the attention change rate is stable or on the rise, the difficulty of the micro-test can be appropriately increased to further challenge the students, promote the deepening and consolidation of their knowledge, and play a better role in teaching feedback.

[0104] Specifically, on the student side, after receiving the micro-test difficulty data, the student starts to load the micro-test and can monitor the loading completion rate of the micro-test in real time. The loading completion rate can reflect the impact of the student device performance and network conditions on the micro-test loading. If the loading completion rate is low, there may be a situation where some student devices are stuck or the network is poor, and corresponding measures need to be taken, such as reminding the student to check the network and trying to reload after waiting for a while. On the teacher side, the teacher receives the transparency data for rendering the interface heat map, and at this time, the feedback data on the rendering delay of the heat map on the teacher side can be obtained. If the rendering delay of the heat map is high, it may affect the teacher's timely acquisition of the student attention distribution information, and thus affect the teaching decision-making, then the system performance or data transmission needs to be optimized. By synthesizing the feedback data on the loading completion rate of the micro-test on the student side and the feedback data on the rendering delay of the heat map on the teacher side, the teaching feedback execution result is generated. This result comprehensively reflects the actual situation of the teaching feedback operation during the implementation process and provides a key reference basis for further optimizing the teaching strategy and improving the system performance in the future.

[0105] In one embodiment, performing an attenuation correction process on the group attention score data according to the course elapsed time data to generate the group attention correction value data, including:

[0106] Calculating the course progress ratio data according to the total course duration data and the course elapsed time data;

[0107] When the course progress ratio data exceeds 0.7, calculating the time attenuation factor according to a preset formula, and the preset formula is:

[0108] λ = 0.02×(t - 0.7T total )

[0109] where λ is the time decay factor, and T total is the total course duration data, and t is the elapsed course time data;

[0110] Perform a non - linear mapping on the time decay factor according to the sigmoid function to generate a decay correction coefficient;

[0111] Use the decay correction coefficient to correct the group attention score data to generate group attention corrected value data.

[0112] Specifically, the total course duration data is the total duration planned for the course, and the elapsed course time data is the cumulative duration of the course from the start to the current moment. The course progress ratio data reflects the completion degree of the course. In this embodiment, the course progress ratio data can be monitored. When the course progress ratio data exceeds 0.7, it can be considered that the course has entered the later stage, and the students' attention may start to decline. That is, the time decay factor can be calculated according to a preset formula. This time decay factor reflects the impact of the remaining course time on attention. As the remaining course time decreases, the time decay factor gradually decreases to represent the natural decline of attention. And a non - linear mapping can be performed on the time decay factor through the sigmoid function, mapping the input value to an interval between 0 and 1, making the change of the correction coefficient smoother. Correcting the group attention score data with the decay correction coefficient can generate group attention corrected value data, making the attention evaluation more in line with the actual situation.

[0113] Based on the same inventive concept, as Figure 2 shown, the embodiment of the present application also provides an online education interactive teaching device 200. The implementation solution provided by this device to solve problems is similar to the implementation solution described in the above method. Therefore, the specific limitations in one or more of the following online education interactive teaching device embodiments can refer to the limitations on the online education interactive teaching device method in the above text and will not be elaborated here. The device includes:

[0114] A data acquisition module 201, configured to acquire student face image data at the student side and current teaching knowledge point data at the teacher side;

[0115] An attention analysis module 202, configured to calculate student attention parameters based on the student face image data to obtain student attention parameter data, where the student attention parameter data includes eyeball focus offset angle data and corrected face orientation angle data;

[0116] A knowledge association module 203, configured to generate knowledge point association degree data through the current teaching knowledge point data, and the knowledge point association degree data is calculated by the proportion of the number of questions associated with the current knowledge point in the historical wrong - question data;

[0117] The teaching optimization module 204 is used to generate dynamic teaching adjustment strategy data based on the student attention parameter data and the knowledge point correlation data, and perform teaching feedback operations according to the dynamic teaching adjustment strategy data to obtain the teaching feedback execution result. The teaching feedback operations include interface heat map overlay and micro-test insertion.

[0118] In the above device, the data acquisition module 201 can obtain the student facial image data at the student end and the current teaching knowledge point data at the teacher end in real time, providing real-time and accurate data support for subsequent attention analysis and teaching optimization. And by capturing the student's facial image and knowledge point data, the system can dynamically perceive the relevance between the student's learning state and the teaching content, laying a foundation for subsequent teaching optimization. The attention analysis module 202 quantifies the student's attention state by analyzing the student facial image data and calculates the student attention parameters, which helps the teacher understand the student's concentration in real time and provides a scientific basis for subsequent teaching adjustment. The knowledge association module 203 can calculate the knowledge point correlation data based on the proportion of the number of questions related to the current knowledge point in the historical wrong question data through the current teaching knowledge point data. This module can identify the weak links of the student in the current knowledge point, helps the teacher understand the relevance between knowledge points, and provides data support for the optimization of teaching content. The teaching optimization module 204 can generate dynamic teaching adjustment strategy data based on the student attention parameter data and the knowledge point correlation data, and optimize teaching through teaching feedback operations. This module can dynamically adjust the teaching content and methods according to the student's attention state and knowledge point mastery, improving the interactivity and immediacy of teaching.

[0119] In an exemplary embodiment, the present invention also provides a computer device, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements the steps of the online education interactive teaching method of the present application. A multi-core processor is preferably used to improve the parallel processing ability of the system. Memory: Provide sufficient temporary storage space to support the operation of the program and the processing of data. The memory capacity should be large enough to accommodate a large amount of supply information and computing tasks.

[0120] In an exemplary embodiment, the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of an online education interactive teaching method of the present application are implemented. The computer-readable storage medium may include: read-only memory (ROM, Read Only Memory), random access memory (RAM, Random Access Memory), solid state drives (SSD, Solid State Drives), or optical discs, etc. Among them, the random access memory may include resistive random access memory (ReRAM, Resistance Random Access Memory) and dynamic random access memory (DRAM, Dynamic Random Access Memory). The above-described embodiments only express several implementation manners of the embodiments of the present application, and their descriptions are relatively specific and detailed, but should not be construed as a limitation on the scope of the patent of the embodiments of the application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the embodiments of the present application, several modifications and improvements can still be made, and these all belong to the protection scope of the embodiments of the present application.

Claims

1. An online education interactive teaching method, characterized in that, The method includes: Obtaining student facial image data at the student end and current teaching knowledge point data at the teacher end; Calculating student attention parameters based on the student facial image data to obtain student attention parameter data, where the student attention parameter data includes eyeball focus offset angle data and corrected facial orientation angle data; Generating knowledge point correlation data through the current teaching knowledge point data, and calculating the knowledge point correlation data based on the proportion of the number of questions related to the current knowledge point in the historical wrong question data; Generating dynamic teaching adjustment strategy data based on the student attention parameter data and the knowledge point correlation data, and performing a teaching feedback operation according to the dynamic teaching adjustment strategy data to obtain a teaching feedback execution result, where the teaching feedback operation includes interface heat map overlay and micro-test insertion.

2. The method according to claim 1, characterized in that The obtaining of the student facial image data at the student end and the current teaching knowledge point data at the teacher end includes: Obtaining the student facial image data through the camera at the student end, and calculating the original facial orientation angle data according to the physical position deviation parameters between the camera and the screen, where the physical position deviation parameters include the horizontal offset between the camera and the center of the screen and the camera focal length; Performing correction processing through a preset geometric correction formula according to the original facial orientation angle data and the physical position deviation parameters to obtain the corrected facial orientation angle data; Obtaining screen sharing content based on the teaching software at the teacher end, extracting and verifying the current teaching knowledge point label data through a preset teaching syllabus database, and using the verified current teaching knowledge point label data as the current teaching knowledge point data.

3. The method according to claim 2, wherein The calculating of the student attention parameters based on the student facial image data to obtain the student attention parameter data, where the student attention parameter data includes eyeball focus offset angle data and corrected facial orientation angle data, includes: Performing pupil positioning processing based on the student facial image data, extracting pupil center coordinate data, and calculating the original eyeball focus offset angle according to the position deviation between the pupil center coordinate data and a preset screen calibration point; Performing head pose analysis based on the student facial image data to obtain head pitch angle data. When it is detected that the head pitch angle data exceeds a preset angle, performing correction processing through a compensation formula based on the original eyeball focus offset angle data and the head pitch angle data to obtain corrected eyeball focus offset angle data; Inputting the corrected eyeball focus offset angle data, the corrected facial orientation angle data, and interaction event frequency data into a multi-modal fusion algorithm for weighted calculation to generate group attention score data; Performing attenuation correction processing on the group attention score data according to the elapsed time data of the course to generate group attention correction value data, and combining the corrected eyeball focus offset angle data, the corrected facial orientation angle data, and the group attention correction value data to generate the student attention parameter data.

4. The method according to claim 3, wherein Inputting the corrected eye focus deviation angle data, the corrected facial orientation angle data, and the interaction event frequency data into a multi-modal fusion algorithm for weighted calculation to generate group attention score data includes: Calculating eye focus deviation influence factor data through a sigmoid function based on the corrected eye focus deviation angle data and a preset eye deviation threshold; Calculating facial orientation cosine value data through a cosine similarity algorithm based on the corrected facial orientation angle data and preset screen positive direction reference data; Calculating interaction growth rate data through the interaction event frequency data; Performing weighted summation on the eye focus deviation influence factor data, the facial orientation cosine value data, and the interaction growth rate data according to preset weight coefficients to generate the group attention score data.

5. The method according to claim 3, characterized in that, Generating dynamic teaching adjustment strategy data based on the student attention parameter data and the knowledge point correlation data includes: Calculating the attention distribution variance data of students based on the group attention correction value data. When the group attention correction value data is lower than 0.4 and the knowledge point correlation data exceeds 0.7, generating knowledge point repeated explanation instruction data; When the attention distribution variance data is continuously higher than 0.2 for 5 minutes, generating random group discussion instruction data; Dynamically adjusting the priority weights of the knowledge point repeated explanation instruction data and the random group discussion instruction data according to historical teaching effect data, where the historical teaching effect data includes student knowledge point retention rate and interaction participation rate; Performing conflict detection and resource occupancy rate assessment on the knowledge point repeated explanation instruction data and the random group discussion instruction data after priority weight adjustment to generate the dynamic teaching adjustment strategy data, where the conflict detection includes judging the overlap of instruction execution time windows and hardware device load conflicts; The resource occupancy rate assessment includes calculating the CPU occupancy rate of the teacher side and the network bandwidth occupancy rate of the student side.

6. The method according to claim 3, characterized in that Performing a teaching feedback operation according to the dynamic teaching adjustment strategy data to obtain a teaching feedback execution result, where the teaching feedback operation includes interface heat map overlay and micro-test insertion, including: Calculating the transparency data of the teacher side interface heat map according to the low attention student ratio data in the dynamic teaching adjustment strategy data, and the low attention student ratio data is calculated by the proportion of the number of students with the group attention correction value data lower than 0.4; Generating bone conduction pulse signal intensity data based on the knowledge point correlation data and the current speaking speed data of the teacher side; Extracting student historical correct rate data from the historical wrong question data and calculating attention change rate data based on the time series of the group attention correction value data; Generating micro-test difficulty data according to the student historical correct rate data and the attention change rate data; Send the transparency data, the bone conduction pulse signal intensity data, and the micro-test difficulty data to the student terminal and the teacher terminal, and receive the feedback data of the micro-test loading completion rate from the student terminal and the feedback data of the heat map rendering delay from the teacher terminal to generate the teaching feedback execution result.

7. The method according to claim 3, wherein The attenuation correction process of the group attention score data according to the course elapsed time data to generate the group attention correction value data includes: Calculate the course progress ratio data according to the total course duration data and the course elapsed time data; When the course progress ratio data exceeds 0.7, calculate the time decay factor according to a preset formula, and the preset formula is: λ = 0.02×(t - 0.7T total ) where λ is the time decay factor, T total is the total course duration data, and t is the elapsed course time data; Perform a non-linear mapping on the time decay factor according to the sigmoid function to generate the attenuation correction coefficient; Use the attenuation correction coefficient to correct the group attention score data to generate the group attention correction value data.

8. An online education interactive teaching device, characterized in that, The device includes: A data acquisition module for acquiring the student face image data of the student terminal and the current teaching knowledge point data of the teacher terminal; An attention analysis module for calculating the student attention parameters according to the student face image data to obtain the student attention parameter data, and the student attention parameter data includes the eye focus offset angle data and the corrected face orientation angle data; A knowledge association module for generating the knowledge point association degree data through the current teaching knowledge point data, and the knowledge point association degree data is calculated by the proportion of the number of questions associated with the current knowledge point in the historical wrong question data; A teaching optimization module for generating dynamic teaching adjustment strategy data based on the student attention parameter data and the knowledge point association degree data, and performing a teaching feedback operation according to the dynamic teaching adjustment strategy data to obtain a teaching feedback execution result, and the teaching feedback operation includes interface heat map overlay and micro-test insertion.

9. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the method according to any one of claims 1 to 7.

Images