Classroom behavior feedback teaching system based on AIGC

Abstract

The invention discloses a classroom behavior feedback teaching system based on AIGC, and relates to the technical field of artificial intelligence. According to the system, classroom video acquisition units are cameras installed at multiple places of a classroom; the classroom behavior recognition module is a classroom behavior recognition model based on deep learning training; the classroom analysis module is used for performing analysis according to an identification result of the classroom behavior identification module and judging and analyzing behaviors of the user in the classroom; the virtual image production module is used for generating a three-dimensional virtual image according to the obtained facial expression information of the teacher in combination with the voice information; the behavior sample library is used for storing behavior samples of teachers or students. According to the invention, the audio and video information of the classroom is collected, model training is carried out on the audio and video information, and the three-dimensional virtual image of the teacher is generated, so that students can select favorite teachers for class on the Internet and can interact with the virtual teachers, the learning channels of the students are increased, and the cost of tutoring training of the students after class is reduced.

Description

Technical Field

[0001] The present invention belongs to the field of artificial intelligence technology, and in particular relates to an AIGC classroom behavior feedback teaching system. Background Art

[0002] With the continuous advancement of science and technology, the application of artificial intelligence in the field of education has gradually attracted people's attention. In today's society, the importance of education is increasing day by day, and many parents strive for better learning resources for their children, such as hiring tutors or one-on-one tutoring. However, finding the right famous teacher and conducting face-to-face teaching is a rather difficult task.

[0003] In this context, a teaching system based on AIGC classroom behavior feedback has ushered in development opportunities. This system is equivalent to a 24-hour private virtual teacher who can answer questions about various subjects at any time, teach multi-disciplinary courses, and provide students with personalized learning support. This new educational method is expected to reduce students' academic pressure, while also reducing parents' anxiety and creating a more free and flexible learning environment for students. Summary of the invention

[0004] The purpose of the present invention is to provide a teaching system based on AIGC classroom behavior feedback. By collecting audio and video information of the classroom and performing model training on the audio and video information, a three-dimensional virtual image of the teacher is generated. Students can choose their favorite teachers to take classes on the Internet, solving the existing problems of high tutoring fees, a single learning environment for students, and high academic pressure.

[0005] In order to solve the above technical problems, the present invention is achieved through the following technical solutions:

[0006] The present invention is a classroom behavior feedback teaching system based on AIGC, including a classroom video acquisition unit, a classroom behavior recognition module, a classroom behavior analysis module, a virtual image generation module, a behavior sample library, and a knowledge base;

[0007] The classroom video acquisition unit is a camera installed in multiple places in the classroom; the camera collects classroom images from multiple angles and divides the classroom images according to multiple user images;

[0008] The classroom behavior recognition module is based on a classroom behavior recognition model trained by deep learning;

[0009] The classroom analysis module is used to analyze the user's behavior in the classroom according to the recognition result of the classroom behavior recognition module;

[0010] The virtual image production module is used to generate a three-dimensional virtual image based on the teacher's facial expression information combined with voice information;

[0011] The behavior sample library is used to store behavior samples of teachers or students;

[0012] The knowledge base is used to store course content;

[0013] The specific training process of the classroom behavior recognition model is as follows:

[0014] Step S1: obtain the classroom image segmentation picture and pre-process it to obtain the video frame with human body in the image;

[0015] Step S2: performing image scaling and data set enhancement on the video frame;

[0016] Step S3: extract features from the data set to obtain feature vectors for each image;

[0017] Step S4: setting model parameters, including: model complexity, validation set ratio, image batch size, and training cycle;

[0018] Step S5: Model training is performed using the feature vector of each image;

[0019] Step S6: Use the validation set to verify the trained model and use the test set to evaluate the model performance.

[0020] As a preferred technical solution, in step S3, the feature vector of each image is obtained, and a feature vector database is constructed to store and analyze the characteristics of the classroom behaviors of teachers and students. The specific feature vector processing flow is as follows:

[0021] Step S21: accurately extract the target area of ​​the image with people, calculate and obtain the feature vector representation of each image, the formula is:

[0022]

[0023] In the formula, A is the feature vector of the image, is a set of image vector features;

[0024] Step S22: Calculate the distance between the normal and abnormal behavior patterns in the training sample and the preset regular behavior template;

[0025] Step S23: Calculate the distance between the feature vector of the behavior to be tested and the feature vector of the imitation behavior. The specific formula is as follows:

[0026] H(A,B)=max[h(A,B),h(B,A)];

[0027] Where B is the feature vector of the image to be identified, h(A,B) is the one-way distance from vector A to vector B, and h(B,A) is the one-way distance from vector B to vector A;

[0028] Step S24: Evaluate according to the set threshold T. If H(A, B) ≥ T, the feature vector does not match, indicating that the behavior is abnormal; if H(A, B) < T, the feature vector matches, indicating that the behavior is normal.

[0029] As a preferred technical solution, the classroom analysis module uses Kalman filtering to select parameters for target tracking. The calculation formula for the target state transfer and observation update is as follows:

[0030]

[0031] In the formula, X(k) represents the state vector, Z(k) represents the observation vector, x(k) represents the lateral position of the target mass center, y(k) represents the longitudinal position of the target mass center, v(k) represents the moving speed of the target on the horizontal axis, and v y (k) represents the moving speed on the ordinate.

[0032] As a preferred technical solution, the behavior sample library has a built-in template sample library for each behavior setting. When a behavior is not found in the template sample library, the behavior sample library will compare the behavior with the identified behavior template to identify the most similar behavior.

[0033] As a preferred technical solution, the specific process of comparing the behavior with the identified behavior template is as follows:

[0034] Step B1: Standardize the original behavior data, that is, subtract the mean of each variable and divide it by the standard deviation to eliminate the impact of dimensions and magnitudes between different features, remove noise, inconsistency and duplication of data, and ensure the quality and accuracy of the data;

[0035] Step B2: Calculate the covariance matrix and perform eigenvalue decomposition to obtain eigenvalues ​​and eigenvectors;

[0036] Step B3: Select the eigenvector with the largest eigenvalue according to the size of the eigenvalue; select the eigenvector with the largest eigenvalue according to the size of the eigenvalue, and these eigenvectors represent the main change direction of the data; project the original data onto the selected eigenvector to obtain the data representation after dimensionality reduction;

[0037] Step B4: Project the original data onto the selected feature vector to obtain the data representation after dimensionality reduction;

[0038] Step B5: In the data space after dimensionality reduction, calculate the distance between each behavior feature point and the center of the category to which it belongs, or calculate the reconstruction error of the data point;

[0039] Step B7: When the distance or reconstruction error of a behavior feature point exceeds a preset threshold, the behavior feature point is considered to be an abnormal behavior.

[0040] After abnormal behavior determination is completed, the results need to be evaluated and optimized, using evaluation indicators (such as precision, recall, etc.) to evaluate the performance of the anomaly detection algorithm. Adjust the threshold or other parameters based on the evaluation results to optimize the accuracy and efficiency of anomaly detection.

[0041] As a preferred technical solution, in step B2, the calculation formula of the covariance matrix is:

[0042]

[0043] In the formula, X cent is the decentralized data set, m is the number of samples; the dimensionality reduction of the data is achieved. These steps together constitute the core content of the PCA algorithm, making the algorithm widely used in data processing and analysis.

[0044] As a preferred technical solution, the virtual image generation module obtains the teacher's voice information, facial information, and eye movement information in the classroom from the classroom video acquisition unit for preprocessing, and converts the voice information into text information; the facial information includes lip movement information, facial expression information, and mouth opening and closing degree information; the eye movement information includes eye trajectory information, pupil size information, blinking frequency, and eye saccade distance; wherein: eye trajectory: eye trajectory refers to the sequence of positions where the eyes stay when observing an object, which can reflect the individual's visual attention distribution and the order of information processing. Eye saccade distance: eye saccade distance refers to the distance from one fixation point to another fixation point, which reflects the efficiency of visual search and the strategy of target selection; pupil size: changes in pupil size are related to the individual's physiological and psychological state, such as fatigue or cognitive load. Blinking frequency: blinking frequency refers to the number of blinks in a certain period of time, which can be used as an indicator to assess the individual's tension. The voice information includes audio contour, pitch contour, and pitch feature vector. The lip movement information includes lip movement trajectory, lip movement speed, lip movement frequency and lip movement amplitude; wherein, lip movement trajectory: lip movement trajectory refers to the path of the lips when speaking or making a sound. By capturing these trajectories, the pronunciation habits and mouth shape changes of individuals can be analyzed, and then used in scenarios such as voice recognition and identity authentication; lip movement speed: lip movement speed refers to the distance the lips move per unit time; this feature reflects the rhythm and speed of an individual's speech, which is of great significance for understanding the naturalness and fluency of language; lip movement frequency: lip movement frequency refers to the number of times the lips complete a specific movement within a certain period of time. This feature is related to the individual's language habits and accents, and helps to distinguish the language characteristics of different regions or groups; lip movement amplitude: lip movement amplitude refers to the maximum opening degree of the lips when speaking or making a sound. This feature is closely related to the emotional expression of an individual, such as opening the mouth wide when surprised or closing the lips tightly when angry.

[0045] As a preferred technical solution, the gaze point change feature information of the eye movement information is estimated using computer vision technology, and a radar chart is drawn for the gaze point. The student's concentration is judged by analyzing the stability, duration and position of the gaze point on the screen, and the chart is summarized into a report to provide feedback to the teacher in the classroom after class.

[0046] The present invention has the following beneficial effects:

[0047] (1) The present invention generates a three-dimensional virtual image of a teacher by collecting audio and video information of the classroom and performing model training on the audio and video information. Students can choose their favorite teachers to attend classes on the Internet and can interact with the virtual teachers, thereby increasing students' learning channels and reducing the cost of after-school tutoring and training for students.

[0048] (2) The present invention collects images of students' facial expressions during class, collects students' face images through a deep learning module, and uses analysis of the stability, duration and position of the gaze point on the screen to determine the students' concentration. The results are summarized in a report format and fed back to teachers in the classroom after class, so that teachers can understand students' concentration in class and adjust the students' status.

[0049] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for describing the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0051] Figure 1 A schematic diagram of the structure of a teaching system based on AIGC classroom behavior feedback of the present invention;

[0052] Figure 2 Training flowchart for the classroom behavior recognition model. DETAILED DESCRIPTION

[0053] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0054] In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0055] In order to make the purpose, technical solution and advantages of this application clearer, the following Figure 1 The implementation methods of the present application are described in further detail.

[0056] Before introducing the embodiments of the present application, the AIGC technology is first described.

[0057] AIGC is the abbreviation of Artificial Intelligence Generated Content, which is a new content production method that uses artificial intelligence technology to automatically create text, images, videos and other content.

[0058] Specifically, it refers to methods based on artificial intelligence technologies such as generative adversarial networks (GANs) and large pre-trained models, which generate relevant content with appropriate generalization capabilities through learning and identifying existing data.

[0059] It inherits the advantages of professional-generated content (PGC) and user-generated content (UGC), and gives full play to its technological advantages to create a new form of digital content generation and interaction.

[0060] The specific implementation of AIGC technology involves multiple aspects, including machine learning, computer vision, natural language processing, optimization algorithms, etc.

[0061] Key technologies such as generative adversarial networks (GAN), diffusion models, and Transformer pre-trained large models provide strong technical support for AIGC.

[0062] In order to make the purpose, technical solutions and advantages of this application more clear, the following Figure 1 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.

[0063] See also Figure 1 As shown, the present invention is a teaching system based on AIGC classroom behavior feedback, including a classroom video acquisition unit, a classroom behavior recognition module, a classroom behavior analysis module, a virtual image generation module, a behavior sample library, and a knowledge base; the classroom video acquisition unit is a camera installed in multiple places in the classroom; the camera collects classroom images from multiple angles and segments the classroom images according to multiple user images; the classroom behavior recognition module is based on a classroom behavior recognition model trained by deep learning; the classroom analysis module is used to analyze according to the recognition results of the classroom behavior recognition module, and judge the user's behavior in the classroom for analysis; the virtual image production module is used to generate a three-dimensional virtual image based on the teacher's facial expression information combined with voice information; the behavior sample library is used to store the behavior samples of teachers or students; the knowledge base is used to store course content;

[0064] See also Figure 2 As shown in the figure, the specific training process of the classroom behavior recognition model is as follows:

[0065] Step S1: obtain the classroom image segmentation picture and pre-process it to obtain the video frame with human body in the image;

[0066] Step S2: performing image scaling and data set enhancement on the video frame;

[0067] Step S3: extract features from the data set to obtain feature vectors for each image;

[0068] Step S4: setting model parameters, including: model complexity, validation set ratio, image batch size, and training cycle;

[0069] Step S5: Model training is performed using the feature vector of each image;

[0070] Step S6: Use the validation set to verify the trained model and use the test set to evaluate the model performance.

[0071] In step S3, the feature vector of each image is obtained, and a feature vector database is constructed to store and analyze the characteristics of the classroom behaviors of teachers and students. The specific feature vector processing flow is as follows:

[0072] Step S21: accurately extract the target area of ​​the image with people, calculate and obtain the feature vector representation of each image, the formula is:

[0073]

[0074] In the formula, A is the feature vector of the image, is a set of image vector features;

[0075] Step S22: Calculate the distance between the normal and abnormal behavior patterns in the training sample and the preset regular behavior template;

[0076] Step S23: Calculate the distance between the feature vector of the behavior to be tested and the feature vector of the imitation behavior. The specific formula is as follows:

[0077] H(A,B)=max[h(A,B),h(B,A)];

[0078] Where B is the feature vector of the image to be identified, h(A,B) is the one-way distance from vector A to vector B, and h(B,A) is the one-way distance from vector B to vector A;

[0079] Step S24: Evaluate according to the set threshold T. If H(A, B) ≥ T, the feature vector does not match, indicating that the behavior is abnormal; if H(A, B) < T, the feature vector matches, indicating that the behavior is normal.

[0080] The classroom analysis module uses Kalman filtering to select parameters for target tracking. The calculation formula for the target state transfer and observation update is as follows:

[0081]

[0082] In the formula, X(k) represents the state vector, Z(k) represents the observation vector, x(k) represents the lateral position of the target mass center, y(k) represents the longitudinal position of the target mass center, v(k) represents the moving speed of the target on the horizontal axis, and v y (k) represents the moving speed on the ordinate.

[0083] The behavior sample library has a built-in template sample library for each behavior setting. When a behavior is not found in the template sample library, the behavior sample library will compare the behavior with the identified behavior template to identify the most similar behavior.

[0084] The specific process of comparing behaviors with identified behavior templates is as follows:

[0085] Step B1: Standardize the original behavior data, that is, subtract the mean of each variable and divide it by the standard deviation to eliminate the impact of dimensions and magnitudes between different features, remove noise, inconsistency and duplication of data, and ensure the quality and accuracy of the data;

[0086] Step B2: Calculate the covariance matrix and perform eigenvalue decomposition to obtain eigenvalues ​​and eigenvectors;

[0087] Step B3: Select the eigenvector with the largest eigenvalue according to the size of the eigenvalue; select the eigenvector with the largest eigenvalue according to the size of the eigenvalue, and these eigenvectors represent the main change direction of the data; project the original data onto the selected eigenvector to obtain the data representation after dimensionality reduction;

[0088] Step B4: Project the original data onto the selected feature vector to obtain the data representation after dimensionality reduction;

[0089] Step B5: In the data space after dimensionality reduction, calculate the distance between each behavior feature point and the center of the category to which it belongs, or calculate the reconstruction error of the data point;

[0090] Step B7: When the distance or reconstruction error of a behavior feature point exceeds a preset threshold, the behavior feature point is considered to be an abnormal behavior.

[0091] After abnormal behavior determination is completed, the results need to be evaluated and optimized, using evaluation indicators (such as precision, recall, etc.) to evaluate the performance of the anomaly detection algorithm. Adjust the threshold or other parameters based on the evaluation results to optimize the accuracy and efficiency of anomaly detection.

[0092] Step B2, the calculation formula of the covariance matrix is:

[0093]

[0094] In the formula, X centis the decentralized data set, m is the number of samples; the dimensionality reduction of the data is achieved. These steps together constitute the core content of the PCA algorithm, making the algorithm widely used in data processing and analysis.

[0095] The virtual image generation module obtains the teacher's voice information, facial information, and eye movement information from the classroom video acquisition unit for preprocessing, and converts the voice information into text information; facial information includes lip movement information, facial expression information, and mouth opening and closing degree information; eye movement information includes eye trajectory information, pupil size information, blinking frequency, and eye saccade distance; among them: eye trajectory: eye trajectory refers to the sequence of positions where the eyes stay when observing an object, which can reflect the individual's visual attention distribution and the order of information processing. Eye saccade distance: eye saccade distance refers to the distance from one fixation point to another, which reflects the efficiency of visual search and the strategy of target selection; pupil size: changes in pupil size are related to the individual's physiological and psychological state, such as fatigue or cognitive load. Blinking frequency: blinking frequency refers to the number of blinks in a certain period of time, which can be used as an indicator to assess the individual's tension. Voice information includes audio contour, pitch contour, and pitch feature vector. Lip movement information includes lip movement trajectory, lip movement speed, lip movement frequency, and lip movement amplitude; among them, lip movement trajectory: lip movement trajectory refers to the path that the lips move when speaking or making sounds. By capturing these trajectories, we can analyze the pronunciation habits and mouth shape changes of individuals, which can then be used in scenarios such as voice recognition and identity authentication; Lip movement speed: Lip movement speed refers to the distance the lips move per unit time; this feature reflects the rhythm and speed of an individual's speech, which is important for understanding the naturalness and fluency of language; Lip movement frequency: Lip movement frequency refers to the number of times the lips complete a specific movement within a certain period of time. This feature is related to the individual's language habits and accents, and helps to distinguish the language characteristics of different regions or groups; Lip movement amplitude: Lip movement amplitude refers to the maximum degree of opening of the lips when speaking or making sounds. This feature is closely related to the emotional expression of an individual, such as opening the mouth wide when surprised or closing the lips when angry.

[0096] The gaze point change feature information of eye movement information is estimated using computer vision technology, and a radar chart is drawn for the gaze point. The student's concentration is judged by analyzing the stability, duration and position of the gaze point on the screen, and the information is summarized into a report to provide feedback to the teacher in the classroom after class.

[0097] It is worth noting that in the above system embodiment, the various units included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0098] In addition, those skilled in the art can understand that all or part of the steps in the above-mentioned embodiments can be completed by instructing related hardware through a program, and the corresponding program can be stored in a computer-readable storage medium.

[0099] The preferred embodiments of the present invention disclosed above are only used to help illustrate the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A classroom behavior feedback teaching system based on AIGC, including a classroom video acquisition unit, a classroom behavior recognition module, a classroom behavior analysis module, a virtual image generation module, a behavior sample library, and a knowledge base, characterized by: The classroom video acquisition unit is a camera installed in multiple places in the classroom; the camera collects classroom images from multiple angles and divides the classroom images according to multiple user images; The classroom behavior recognition module is based on a classroom behavior recognition model trained by deep learning; The classroom analysis module is used to analyze the user's behavior in the classroom according to the recognition result of the classroom behavior recognition module; The virtual image production module is used to generate a three-dimensional virtual image based on the teacher's facial expression information combined with voice information; The behavior sample library is used to store behavior samples of teachers or students; The knowledge base is used to store course content; The specific training process of the classroom behavior recognition model is as follows: Step S1: obtain the classroom image segmentation picture and pre-process it to obtain the video frame with human body in the image; Step S2: performing image scaling and data set enhancement on the video frame; Step S3: extract features from the data set to obtain feature vectors for each image; Step S4: setting model parameters, including: model complexity, validation set ratio, image batch size, and training cycle; Step S5: Model training is performed using the feature vector of each image; Step S6: Use the validation set to verify the trained model and use the test set to evaluate the model performance.

2. According to claim 1, the AIGC classroom behavior feedback teaching system is characterized in that: In step S3, the feature vector of each image is obtained, and a feature vector database is constructed to store and analyze the characteristics of the classroom behaviors of teachers and students. The specific feature vector processing flow is as follows: Step S21: accurately extract the target area of ​​the image with people, calculate and obtain the feature vector representation of each image, the formula is: In the formula, A is the feature vector of the image, is a set of image vector features; Step S22: Calculate the distance between the normal and abnormal behavior patterns in the training sample and the preset regular behavior template; Step S23: Calculate the distance between the feature vector of the behavior to be tested and the feature vector of the imitation behavior. The specific formula is as follows: H(A,B)=max[h(A,B),h(B,A)]; Where B is the feature vector of the image to be identified, h(A,B) is the one-way distance from vector A to vector B, and h(B,A) is the one-way distance from vector B to vector A; Step S24: Evaluate according to the set threshold T. If H(A, B) ≥ T, the feature vector does not match, indicating that the behavior is abnormal; if H(A, B) < T, the feature vector matches, indicating that the behavior is normal.

3. According to claim 1, the AIGC classroom behavior feedback teaching system is characterized in that: The classroom analysis module uses Kalman filtering to select parameters for target tracking. The calculation formula for the state transfer and observation update of the target is as follows: In the formula, X(k) represents the state vector, Z(k) represents the observation vector, x(k) represents the lateral position of the target mass center, y(k) represents the longitudinal position of the target mass center, v(k) represents the moving speed of the target on the horizontal axis, and v y (k) represents the moving speed on the ordinate.

4. According to claim 1, the AIGC classroom behavior feedback teaching system is characterized in that: The behavior sample library has a built-in template sample library for each behavior setting. When a behavior is not found in the template sample library, the behavior sample library will compare the behavior with the identified behavior template to identify the most similar behavior.

5. According to claim 4, the AIGC classroom behavior feedback teaching system is characterized in that: The specific process of comparing the behavior with the identified behavior template is as follows: Step B1, standardize the original behavioral data, that is, subtract the mean of each variable and divide it by the standard deviation; Step B2: Calculate the covariance matrix and perform eigenvalue decomposition to obtain eigenvalues ​​and eigenvectors; Step B3: Select the eigenvector with the largest eigenvalue according to the size of the eigenvalue; Step B4: Project the original data onto the selected feature vector to obtain the data representation after dimensionality reduction; Step B5: In the data space after dimensionality reduction, calculate the distance between each behavior feature point and the center of the category to which it belongs, or calculate the reconstruction error of the data point; Step B7: When the distance or reconstruction error of a behavior feature point exceeds a preset threshold, the behavior feature point is considered to be an abnormal behavior.

6. The AIGC classroom behavior feedback teaching system according to claim 5 is characterized in that: In step B2, the calculation formula of the covariance matrix is: Where, X cent is the decentralized data set, and m is the number of samples.

7. The AIGC classroom behavior feedback teaching system according to claim 1 is characterized in that: The virtual image generation module obtains the teacher's voice information, facial information, and eye movement information in the classroom from the classroom video acquisition unit for pre-processing, and converts the voice information into text information; the facial information includes lip movement information, facial expression information, and mouth opening and closing degree information; the eye movement information includes eyeball trajectory information, pupil size information, blinking frequency, and eye saccade distance; the voice information includes audio contour, pitch contour, and pitch feature vector; the lip movement information includes lip movement trajectory, lip movement speed, lip movement frequency, and lip movement amplitude.

8. The AIGC classroom behavior feedback teaching system according to claim 7 is characterized in that: The voice information is converted into text information and transmitted to the large language model, and the corresponding answer is returned; the question and the answer are put into machine learning for learning to form a model for judging the end of the question; if it is judged that the conversation is ended, the corresponding ending words are returned, if the conversation is not ended, it continues to receive.

9. The AIGC classroom behavior feedback teaching system according to claim 7 is characterized in that: The gaze point change feature information of the eye movement information is estimated using computer vision technology, and a radar chart is drawn for the gaze point. The student's concentration is judged by analyzing the stability, duration and position of the gaze point on the screen, and the information is summarized into a report to provide feedback to the teacher in the classroom after class.

Images