Screen interactive learning method and system based on artificial intelligence

Abstract

The invention discloses a screen interactive learning method and system based on artificial intelligence, and belongs to the technical field of interactive learning, and the method comprises the steps: firstly, transmitting a first interaction instruction based on artificial intelligence, randomly selecting an interaction problem from a preset question bank, and carrying out the visual display of the interaction problem through a screen; secondly, extracting a human body action image of the learner through an image recognition technology, and constructing a human body action feature model of the learner; then, collecting sudden change information by utilizing a human body action feature model, and obtaining a judgment condition for eliminating interference actions; eliminating the interference action based on the judgment condition of interference action elimination; and finally, based on artificial intelligence, according to a preset answer of the interaction question and the human body action characteristics after the interference action is eliminated, a second interaction instruction is sent out, and interactive learning is completed. According to the method, interference actions caused by randomness of human body actions are eliminated, the accuracy and logicality of screen interaction are greatly improved, and the learning effect is effectively improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of artificial intelligence and interactive learning, and specifically relates to a screen interactive learning method and system based on artificial intelligence. Background Art

[0002] At present, in various exhibition halls such as science and technology museums and history museums, the innovation of educational methods is proceeding at an unprecedented speed. Among them, the integration of artificial intelligence technology provides an unprecedented learning guidance experience for beginners and visitors who are willing to learn. In particular, screen interactive learning, as a new learning guidance method that integrates modern technology and educational concepts, has gradually become an indispensable part of the exhibition hall with its intuitive visualization effects and interesting interactive forms, greatly enriching learners' visiting experience and knowledge acquisition methods.

[0003] However, despite the increasing application of screen interactive learning models in exhibition halls, its traditional forms of expression, such as simple button answering or voice response, have limited its interactivity to a certain extent. These seemingly convenient and direct answering methods often lack deep interaction and personalized design, and are difficult to truly stimulate learners' interest and enthusiasm, resulting in a significant gap between the actual effect of screen interactive learning devices and the expected goals of the exhibition hall and the actual needs of learners.

[0004] To address this problem, a new type of screen interactive learning mode based on human motion recognition has emerged, which can effectively improve the interactivity and fun of screen interactive learning. However, due to the randomness of human motion, a large number of interfering motions will affect the accuracy and logic of screen interaction. This interference will cause many interaction errors in the actual learning process and greatly reduce the effect of interactive learning.

[0005] Therefore, there is an urgent need to provide an artificial intelligence-based screen interactive learning method and system that can be applied to exhibition halls and eliminate human interference movements. Summary of the invention

[0006] The purpose of the present invention is to provide a screen interactive learning method and system based on artificial intelligence to solve the above-mentioned problems existing in the prior art.

[0007] In order to achieve the above object, the present invention adopts the following technical solutions:

[0008] In a first aspect, the present invention provides a screen interactive learning method based on artificial intelligence, comprising:

[0009] Based on artificial intelligence, issue a first interaction instruction, randomly select at least one question from a preset question bank as an interaction question according to the first interaction instruction, and visually display the interaction question through a screen;

[0010] Extract the human motion image of the learner through image recognition technology, and construct a human motion feature model of the learner according to the extracted human motion image of the learner;

[0011] Utilize the human motion feature model to collect mutation information and obtain a determination condition for excluding interfering actions;

[0012] According to the human motion feature model of the learner, screen out interfering actions, and based on the determination condition for excluding interfering actions, exclude the interfering actions;

[0013] Take the human motion after excluding interfering actions as the interaction feedback output of the learner, and based on artificial intelligence, issue a second interaction instruction according to the preset answer of the interaction question and the human motion features after excluding interfering actions.

[0014] In a possible design, before issuing the first interaction instruction, it further includes:

[0015] Obtain the learner's position through image recognition technology;

[0016] Utilize artificial intelligence to judge whether the learner's position is at a preset interaction position. If the judgment result is that the learner's position is not at the preset interaction position, send a reminder to the learner and make the judgment again until the judgment result is that the learner's position is at the preset interaction position, and then issue the first interaction instruction.

[0017] In a possible design, constructing a human motion feature model of the learner according to the extracted human motion image of the learner includes:

[0018] According to the extracted human motion image of the learner, extract the human skeleton key points of the learner, and establish a human skeleton diagram of the learner according to the human skeleton key points of the learner;

[0019] Perform edge detection on the human skeleton diagram of the learner to extract the contour line features of the human motion of the learner;

[0020] According to the extracted human motion image of the learner, obtain the color gradient of the human motion of the learner;

[0021] According to the contour line features of the human motion of the learner, generate an image matrix of the human motion of the learner;

[0022] Construct a human motion feature model of the learner according to the image matrix and the color gradient.

[0023] In a possible design, mutation information is collected, and the determination conditions for excluding interference actions are obtained, including:

[0024] Using the locally linear embedding algorithm, the mutation information is screened by the human body motion feature model, where the mutation information is the sudden motion features that appear in the feature space due to the randomness of the learner's human body motion or environmental interference;

[0025] Map the human body motion feature model to a two-dimensional manifold and establish a global non-significant two-dimensional manifold set matrix;

[0026] Based on the global non-significant two-dimensional manifold set matrix, with the screened mutation information as the basis, the interference action vector is dynamically updated;

[0027] Calculate the global significance color difference vector between the color map of the interference area of the human body motion feature and the surrounding color space, and generate the determination conditions for excluding interference actions based on the global significance color difference vector and the interference action vector.

[0028] In a possible design, mapping the human body motion feature model to a two-dimensional manifold and establishing a global non-significant two-dimensional manifold set matrix includes:

[0029] Map the human body motion feature model to a two-dimensional manifold through the locally linear embedding algorithm, where any point in the two-dimensional manifold corresponds to a human body motion in the human body motion feature model;

[0030] Map each point in the two-dimensional manifold to the Euclidean space, and evaluate the feature information of each point to obtain the stability and significance of each action point;

[0031] According to the stability and significance of each action point, select multiple points that conform to the human body motion logic to establish a global non-significant two-dimensional manifold set matrix.

[0032] In a possible design, based on the determination conditions for excluding interference actions, the interference actions are excluded, including:

[0033] Based on the determination conditions for excluding interference actions, select the human body motion area corresponding to the interference action from the human body motion image of the learner and mark it as the interference area;

[0034] Calculate the color difference vector between the interference area and the surrounding normal area;

[0035] Obtain the color gradient value of the surrounding normal area according to the color difference vector between the interference area and the surrounding normal area, and select the abnormal pixel points in the interference area;

[0036] Using the mean filter algorithm, correct the abnormal pixel points in the interference area according to the color gradient values of the surrounding normal areas to eliminate the interference actions.

[0037] In a possible design, based on artificial intelligence, according to the preset answers to the interaction questions and the human motion characteristics after eliminating the interference actions, send a second interaction instruction, including:

[0038] Using artificial intelligence, determine whether the human motion after eliminating the interference actions conforms to the preset answers to the interaction questions;

[0039] According to the judgment result, send a second interaction instruction to guide the learner to perform the next action interaction.

[0040] In a second aspect, the present invention provides a screen interactive learning system based on artificial intelligence, including:

[0041] An intelligent interaction module, configured to send a first interaction instruction based on artificial intelligence, randomly select at least one question from a preset question bank as an interaction question according to the first interaction instruction, and visually display the interaction question through the screen; and also configured to output the human motion after eliminating the interference actions as the interaction feedback of the learner, and based on artificial intelligence, send a second interaction instruction according to the preset answers to the interaction questions and the human motion characteristics after eliminating the interference actions;

[0042] A model construction module, configured to extract the human motion images of the learner through image recognition technology, and construct a human motion feature model of the learner according to the extracted human motion images of the learner;

[0043] A mapping calculation module, configured to collect mutation information by using the human motion feature model and obtain the determination conditions for eliminating interference actions;

[0044] An interference elimination module, configured to screen out interference actions according to the human motion feature model of the learner, and eliminate the interference actions based on the determination conditions for eliminating interference actions.

[0045] In a third aspect, the present invention provides an electronic device, including a memory, a processor, and a transceiver that are communicatively connected in sequence. Among them, the memory is used to store a computer program, the transceiver is used to send and receive messages, and the processor is used to read the computer program and execute the artificial intelligence-based screen interactive learning method described in the first aspect or any possible design of the first aspect.

[0046] In a fourth aspect, the present invention provides a computer program product containing instructions, which when run on a computer, cause the computer to execute the artificial intelligence-based screen interactive learning method described in the first aspect or any possible design of the first aspect.

[0047] Beneficial effects: The present invention provides an artificial intelligence-based screen interactive learning method and system. Firstly, a first interaction instruction is issued based on artificial intelligence, and at least one question is randomly selected from a preset question bank as an interaction question according to the first interaction instruction, and the interaction question is visually displayed through the screen. Secondly, the human action image of the learner is extracted by image recognition technology, and a human action feature model of the learner is constructed according to the extracted human action image of the learner. Then, the human action feature model is used to collect mutation information and obtain the determination condition for excluding interference actions. Then, according to the human action feature model of the learner, interference actions are screened out, and based on the determination condition for excluding interference actions, the interference actions are excluded. Finally, the human action after excluding the interference actions is output as the interaction feedback of the learner, and based on artificial intelligence, a second interaction instruction is issued according to the preset answer of the interaction question and the human action feature after excluding the interference actions, completing an interactive learning. This interactive learning method excludes the interference actions caused by the randomness of human actions, greatly improving the accuracy and logic of screen interaction, which enables the learning process of the learner to avoid many interaction errors and effectively improves the learning effect of interactive learning. Brief Description of the Drawings

[0048] Figure 1 It is a flowchart of the artificial intelligence-based screen interactive learning method in an embodiment of the present invention. Detailed Embodiments

[0049] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the present invention in combination with the drawings and the description of the embodiments or the prior art. Obviously, the following description of the structure of the drawings is only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts. It should be noted here that the description of these embodiment modes is used to help understand the present invention, but does not constitute a limitation to the present invention.

[0050] It should be understood that although terms such as first and second may be used herein to describe various units, these units should not be limited by these terms. These terms are only used to distinguish one unit from another. For example, the first unit can be called the second unit, and similarly, the second unit can be called the first unit, without departing from the scope of the exemplary embodiments of the present invention.

[0051] It should be understood that for the term "and / or" that may appear in this text, it is merely a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, B exists alone, and both A and B exist simultaneously. For the term " / and" that may appear in this text, it describes another association object relationship, indicating that there can be two relationships. For example, A / and B can represent: A exists alone, and both A and B exist. Additionally, for the character " / " that may appear in this text, it generally indicates that the front and back associated objects have an "or" relationship.

[0052] Embodiment:

[0053] As Figure 1 shown, this embodiment provides an artificial intelligence-based screen interactive learning method, which includes:

[0054] S100. Issue a first interaction instruction based on artificial intelligence, randomly select at least one question from a preset question bank as an interaction question, and visually display the interaction question on the screen;

[0055] In a possible implementation manner, in step S100, before issuing the first interaction instruction, it further includes:

[0056] S1001. Obtain the learner's position through image recognition technology;

[0057] S1002. Use artificial intelligence to determine whether the learner's position is in a preset interaction position. If the determination result is that the learner's position is not in the preset interaction position, a reminder is sent to the learner and the determination is made again until the determination result is that the learner's position is in the preset interaction position, and then the first interaction instruction is issued.

[0058] S200. Extract the learner's human motion image through image recognition technology, and construct a learner's human motion feature model based on the extracted learner's human motion image;

[0059] In a possible implementation manner, in step S200, constructing a learner's human motion feature model based on the extracted learner's human motion image includes:

[0060] S2001. According to the extracted learner's human motion image, extract the learner's human skeleton key points, and establish a learner's human skeleton diagram based on the learner's human skeleton key points;

[0061] S2002. Perform edge detection on the learner's human skeleton diagram to extract the contour line features of the learner's human motion;

[0062] S2003. Obtain the color gradient of the learner's human motion based on the extracted human motion image of the learner.

[0063] S2004. Generate an image matrix of the learner's human motion according to the contour line features of the learner's human motion.

[0064] S2005. Construct a human motion feature model of the learner based on the image matrix and the color gradient of the learner's human motion.

[0065] It should be noted that the human skeleton diagram is generated by extracting the key points of the human skeleton (such as the head, shoulders, elbows, etc.) in the learner's human motion image through image processing techniques (such as OpenPose or depth sensors) based on the requirements of human motion recognition; the Canny algorithm can be used to perform edge detection on the learner's human skeleton diagram, extract the contour line features of the human motion, and generate an image matrix corresponding to the contour line features of the human motion. On this basis, through gray conversion, gradient calculation and quantization, the color gradient information of the human motion is also represented by a matrix. Taking the image matrix and the color gradient information in each human motion image as input quantities and the human motion features of the learner as output quantities, after repeated training, a human motion feature model of the learner is finally formed.

[0066] S300. Use the human motion feature model to collect mutation information and obtain the determination conditions for excluding interfering actions.

[0067] In a possible implementation manner, in step S300, collecting mutation information and obtaining the determination conditions for excluding interfering actions includes:

[0068] S3001. Use the locally linear embedding algorithm to screen the mutation information by using the human motion feature model, where the mutation information is the sudden motion features that appear in the feature space due to the randomness of the learner's human motion or environmental interference.

[0069] S3002. Map the human motion feature model to a two-dimensional manifold and establish a global non-significant two-dimensional manifold set matrix.

[0070] S3003. Based on the global non-significant two-dimensional manifold set matrix, dynamically update the interfering action vector based on the screened mutation information.

[0071] S3005. Calculate the global significant color difference vector between the color map of the interfering area of the human motion feature and the surrounding color space, and generate the determination conditions for excluding interfering actions based on the global significant color difference vector and the interfering action vector.

[0072] It should be noted that a two-dimensional manifold is a curved surface. Specifically, a point on a two-dimensional manifold has no overall coordinates, but a homeomorphic mapping can be established with the Euclidean space in the neighborhood of the point, so as to obtain the image point of the point in the Euclidean space as the coordinate of the point; and mapping the human motion feature model to a two-dimensional manifold is actually mapping the points on the feature space where the human motion features are located to a two-dimensional manifold, where each point corresponds to a human motion sample on the feature space, which is essentially a dimensionality reduction calculation, so as to expand and describe the data that cannot be described in the high-dimensional feature space in a low-dimensional manner, and establish a global non-significant two-dimensional manifold set matrix based on this; the global non-significant two-dimensional manifold set matrix mentioned here is actually a matrix that aggregates the mapped points after dimensionality reduction, so as to ignore the complex high-dimensional relationship between the points and consider them to be non-significant, and by eliminating these non-significant features, the model structure can be simplified and the calculation efficiency can be improved, which eliminates unnecessary influences for subsequent interference motion recognition and greatly improves its recognition efficiency.

[0073] The local linear embedding algorithm (LLE) is a nonlinear dimensionality reduction technology. Its core idea is that in high-dimensional space, each data point can be linearly reconstructed from several neighboring points, and this reconstruction relationship should remain unchanged in low-dimensional space. Therefore, when the points in the feature space are mapped to the two-dimensional manifold, their reconstruction relationship (local neighborhood structure) is also maintained. The local neighborhood of the points on the two-dimensional manifold can be selected to update the local linear embedding weights to achieve dynamic updates of the connecting action (interference action vector) between two action points. Through this step, a closed-loop process from feature dimensionality reduction, vector update to interference judgment is realized, which effectively improves the accuracy of human motion feature recognition.

[0074] In step S3002, the human motion feature model is mapped to a two-dimensional manifold, and a global non-significant two-dimensional manifold set matrix is established, including:

[0075] S30021. Mapping the human motion feature model to a two-dimensional manifold by a local linear embedding algorithm, wherein any point in the two-dimensional manifold corresponds to a human motion in the human motion feature model;

[0076] S30022. Map each point in the two-dimensional manifold to the Euclidean space, and evaluate the feature information of each point to obtain the stability and significance of each action point;

[0077] S30023. According to the stability and significance of each action point, select multiple points that conform to the logic of human body action to establish a global non-significant two-dimensional manifold set matrix.

[0078] Among them, the Euclidean space here refers to the Euclidean space, whose dimension is determined by the image matrix, and any point on it is used to represent a human body movement.

[0079] S400. According to the human body movement feature model of the learner, filter out interfering movements, and based on the determination conditions for excluding interfering movements, exclude the interfering movements.

[0080] In a possible implementation manner, in step S400, excluding the interfering movements based on the determination conditions for excluding interfering movements includes:

[0081] S4001. Based on the determination conditions for excluding interfering movements, select the human body movement area corresponding to the interfering movement from the human body movement images of the learner, and mark it as the interfering area.

[0082] S4002. Calculate the color difference vector between the interfering area and the surrounding normal areas.

[0083] S4003. Obtain the color gradient value of the surrounding normal areas according to the color difference vector between the interfering area and the surrounding normal areas, and select the abnormal pixel points in the interfering area.

[0084] S4004. Use the mean filter algorithm to correct the abnormal pixel points in the interfering area according to the color gradient value of the surrounding normal areas, so as to achieve the exclusion of interfering movements.

[0085] Among them, the determination conditions for excluding interfering movements are determined by the color map of the interfering area of the human body movement features and the global significant color difference vector of the surrounding color space. Specifically, when the color difference vector exceeds the preset threshold, it is determined that this area is a significant interfering movement feature area, that is: the interfering area (non-significant area); the movement in the interfering area deviates from the normal movement pattern. For the interfering area, local smoothing processing of the abnormal pixel points is performed using the color gradient change, that is: mean filtering, to eliminate noise and restore the logical continuity of the movement contour, so as to achieve the exclusion of interfering movements.

[0086] S500. Use the human body movement after excluding the interfering movements as the interactive feedback output of the learner, and based on artificial intelligence, issue a second interactive instruction according to the preset answer to the interactive question and the human body movement features after excluding the interfering movements.

[0087] In a possible implementation manner, in step S500, issuing a second interactive instruction based on artificial intelligence according to the preset answer to the interactive question and the human body movement features after excluding the interfering movements includes:

[0088] S5001. Use artificial intelligence to determine whether the human body movement after excluding the interfering movements conforms to the preset answer to the interactive question.

[0089] S5002. Issue a second interaction instruction according to the judgment result to guide the learner to perform the next action interaction.

[0090] It should be noted that it is necessary to judge whether the human body movement after excluding the interference movement conforms to the preset answer of the interaction problem. This preset answer is generally a human body movement image. The similarity between the human body movement image after excluding the interference movement and the preset answer is judged. When the similarity is higher than the preset threshold, it is considered that: it conforms to the answer, otherwise it is: does not conform to the answer. This preset threshold can generally be set to 90%; according to the judgment result, a second interaction instruction is issued. Here, according to the different judgment results, the content of the second interaction instruction is also different. When the judgment result is "conforms to the answer", the second interaction instruction is generally set to: guide the learner to learn the next question, and when the judgment result is "does not conform to the answer", the second interaction instruction is generally set to: answer again / announce the answer and learn the next question.

[0091] Finally, it should be noted that: the above are only the preferred embodiments of the present invention and are not used to limit the protection scope of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An artificial intelligence-based screen interactive learning method, characterized in that, Including: Issuing a first interaction instruction based on artificial intelligence, randomly selecting at least one question from a preset question bank as an interaction question according to the first interaction instruction, and visually displaying the interaction question through a screen; Extracting the human action image of the learner through image recognition technology, and constructing a human action feature model of the learner according to the extracted human action image of the learner; Using the human action feature model to collect mutation information and obtaining a determination condition for excluding interfering actions; Filtering out interfering actions according to the human action feature model of the learner, and excluding the interfering actions based on the determination condition for excluding interfering actions; Taking the human action after excluding the interfering actions as the interaction feedback output of the learner, and based on artificial intelligence, issuing a second interaction instruction according to the preset answer of the interaction question and the human action features after excluding the interfering actions.

2. The screen interactive learning method based on artificial intelligence according to claim 1, wherein Before issuing the first interaction instruction, it further includes: Obtaining the learner's position through image recognition technology; Using artificial intelligence to judge whether the learner's position is in a preset interaction position. If the judgment result is that the learner's position is not in the preset interaction position, a reminder is sent to the learner and the judgment is made again until the judgment result is that the learner's position is in the preset interaction position, and then the first interaction instruction is issued.

3. The screen interactive learning method based on artificial intelligence according to claim 1, characterized in that Constructing a human action feature model of the learner according to the extracted human action image of the learner, including: According to the extracted human action image of the learner, extracting the human skeleton key points of the learner, and establishing a human skeleton map of the learner according to the human skeleton key points of the learner; Performing edge detection on the human skeleton map of the learner to extract the contour line features of the human action of the learner; Obtaining the color gradient of the human action of the learner according to the extracted human action image of the learner; Generating an image matrix of the human action of the learner according to the contour line features of the human action of the learner; Constructing a human action feature model of the learner according to the image matrix and the color gradient.

4. The screen interactive learning method based on artificial intelligence according to claim 1, characterized in that Collecting mutation information and obtaining a determination condition for excluding interfering actions, including: Using the locally linear embedding algorithm to screen the mutation information by using the human action feature model, where the mutation information is the sudden action feature that appears in the feature space due to the randomness of the learner's human action or environmental interference; Mapping the human action feature model to a two-dimensional manifold and establishing a global non-significant two-dimensional manifold set matrix; Based on the global non-significant two-dimensional manifold set matrix, dynamically updating the interfering action vector based on the screened mutation information; Calculating the global significance color difference vector between the color map of the interfering area of the human action feature and the surrounding color space, and generating a determination condition for excluding interfering actions based on the global significance color difference vector and the interfering action vector.

5. The screen interactive learning method based on artificial intelligence according to claim 4, characterized in that Mapping the human action feature model to a two-dimensional manifold and establishing a global non-significant two-dimensional manifold set matrix, including: Mapping the human action feature model to a two-dimensional manifold through the locally linear embedding algorithm, where any point in the two-dimensional manifold corresponds to a human action in the human action feature model; Map each point in the two-dimensional manifold to the Euclidean space, and evaluate the feature information of each point to obtain the stability and significance of each action point; According to the stability and significance of each action point, select multiple points that conform to the human action logic to establish a global non-significant two-dimensional manifold set matrix.

6. The screen interactive learning method based on artificial intelligence according to claim 1, wherein Based on the determination conditions for interference action exclusion, exclude interference actions, including: Based on the determination conditions for interference action exclusion, select the human action area corresponding to the interference action from the human action image of the learner and mark it as the interference area; Calculate the color difference vector between the interference area and the surrounding normal area; Obtain the color gradient value of the surrounding normal area according to the color difference vector between the interference area and the surrounding normal area, and select abnormal pixel points in the interference area; Use the mean filter algorithm to correct the abnormal pixel points in the interference area according to the color gradient value of the surrounding normal area to achieve the exclusion of interference actions.

7. The artificial intelligence-based screen interactive learning method according to claim 1, characterized in that, Based on artificial intelligence, according to the preset answer of the interaction question and the human action characteristics after excluding interference actions, issue a second interaction instruction, including: Use artificial intelligence to determine whether the human action after excluding interference actions conforms to the preset answer of the interaction question; According to the judgment result, issue a second interaction instruction to guide the learner to perform the next action interaction.

8. An artificial intelligence-based screen interactive learning system, characterized in that, Including: An intelligent interaction module for issuing a first interaction instruction based on artificial intelligence, randomly selecting at least one question from a preset question bank as an interaction question according to the first interaction instruction, and visually displaying the interaction question through a screen; It is also used to output the human action after excluding interference actions as the interaction feedback of the learner, and based on artificial intelligence, issue a second interaction instruction according to the preset answer of the interaction question and the human action characteristics after excluding interference actions; A model construction module for extracting the human action image of the learner through image recognition technology and constructing a human action feature model of the learner according to the extracted human action image of the learner; A mapping calculation module for collecting mutation information by using the human action feature model and obtaining the determination conditions for interference action exclusion; An interference exclusion module for screening out interference actions according to the human action feature model of the learner and excluding the interference actions based on the determination conditions for interference action exclusion.

9. An electronic device, characterized in that, It includes a memory, a processor, and a transceiver that are communicatively connected in sequence. Among them, the memory is used to store computer programs, the transceiver is used to send and receive messages, and the processor is used to read the computer programs and execute the artificial intelligence-based screen interactive learning method according to any one of claims 1 to 7.

10. A computer program product, comprising a computer program or instructions, characterized in that, The computer program or the instruction, when executed by a computer, implements the artificial intelligence-based screen interactive learning method according to any one of claims 1 to 7.

Images