An intelligent teaching system based on AI vision technology

Abstract

The present invention relates to the technical field of smart campuses, and discloses an intelligent teaching system based on AI vision technology, including: an image acquisition module for acquiring motion images of students; a motion data extraction module for obtaining corresponding image motion data based on the acquired motion image processing; a first image screening module for extracting key images from the motion images; a historical image mapping module for establishing a mapping relationship between the motion images of the historical first motion image set and the first motion image set of the current motion item of the current student; a qualified parameter calculation module for calculating qualified parameters; and a result evaluation module for evaluating the result of the current motion item of the current student based on the qualified parameters. The present invention uses AI vision technology to identify the motion data of students based on the acquired motion images, and performs probability comprehensive calculation of qualified parameters based on the associated historical data of the motion data to accurately evaluate the motion items of students.

Description

Technical Field

[0001] The present invention relates to the technical field of smart campuses, and more specifically, to an intelligent teaching system based on AI vision technology. Background Art

[0002] The construction of a smart campus aims to assist teachers in teaching and students in self-study through artificial intelligence technology; it is difficult to express the teaching content of sports disciplines in words, and since students' training is recorded in the form of videos, it is also difficult to score through traditional methods of text recognition and comparison. Summary of the Invention

[0003] The present invention provides an intelligent teaching system based on AI vision technology, which solves the technical problem that it is difficult to automatically score the teaching of sports disciplines based on the collected image information in the related art.

[0004] According to one aspect of the present invention, there is provided an intelligent teaching system based on AI vision technology, including:

[0005] An image acquisition module for acquiring students' motion images;

[0006] A motion data extraction module for obtaining corresponding image motion data based on the collected motion image processing;

[0007] The image motion data generated from the motion images of one student in one sports event is classified into an image motion data set and sorted according to time frames;

[0008] A first image screening module for extracting key images from the motion images;

[0009] A historical image mapping module for establishing a mapping relationship between the motion images of the historical first motion image set and the first motion image set of the current student in the current sports event;

[0010] A qualified parameter calculation module for calculating qualified parameters;

[0011] The calculation formula for the qualified parameter is as follows:

[0012]

[0013] Where X1 = 1 indicates that there is a mapping between the first motion image in the historical data and the first first motion image in the first motion image set of the current student, X i = 1 indicates that there is a mapping between the first motion image in the historical data and the i-th first motion image in the first motion image set of the current student, paX i= 1 indicates that there is a mapping between the first moving image in the historical data and all the first moving images before the i-th first moving image in the first moving image set of the current student; y = 1 indicates that the corresponding evaluation result of the historical data is qualified, and n is the number of moving images in the first moving image set of the current student.

[0014] A result evaluation module that evaluates the result of the current student's current sports event based on the qualified parameter. If the evaluation parameter is greater than the set first evaluation parameter threshold, it is judged as qualified; otherwise, it is judged as unqualified.

[0015] A teaching guidance module that is used to recommend moving images for the student to refer to when the evaluation parameter of the current student's current sports event is less than the set second evaluation parameter threshold.

[0016] Furthermore, the method for extracting key images includes:

[0017] Step 101: Select the moving image with the earliest time frame in the image motion data set as the starting moving image.

[0018] Step 102: Starting from the starting moving image, traverse the moving images in the order of time frames backward. When the first similarity between the currently traversed moving image and the starting moving image is less than the set first similarity threshold, terminate the traversal; at this time, record the currently traversed moving image as the starting moving image.

[0019] Step 103: Update the moving image at the end of the traversal as the new first moving image.

[0020] Step 104: Iteratively execute Steps 102 and 103 until all moving images are traversed.

[0021] Step 105: Generate a first moving image set from the first moving images.

[0022] Furthermore, the calculation formula for the first similarity is as follows:

[0023]

[0024] where T 1i represents the angle of the i-th limb of the human skeleton corresponding to the image motion data of a moving image, and T 2i represents the angle of the i-th limb of the human skeleton corresponding to the image motion data of another moving image, and n is the number of limbs of the human skeleton.

[0025] Furthermore, one moving image corresponds to one image motion data obtained through processing. A student will collect multiple moving images sorted by time frames in a sports event, and multiple image motion data sorted by time frames can be obtained through processing.

[0026] Furthermore, the method for establishing a mapping relationship between the historical set of first motion images and the first motion images of the current student's current sports event includes:

[0027] Step 201, extract a first motion image with the earliest time frame from the set of first motion images of the current student as the reference motion image;

[0028] Step 202, sequentially traverse the first motion images in the historical set of first motion images, and the termination condition for traversal is:

[0029] The first similarity between the currently traversed first motion image and the reference motion image is greater than the set second similarity threshold;

[0030] Establish a mapping relationship between the first motion image at the end of traversal and the first motion image of the current student;

[0031] Step 203, after the traversal ends, extract the first motion image after the current reference motion image from the set of first motion images of the current student and update it as the new reference motion image;

[0032] Step 204, iteratively execute Steps 202 and 203 until all the first motion images in the set of first motion images of the current student are extracted.

[0033] Furthermore, if the first similarity between all the first motion images in the historical data and the reference motion image is less than or equal to the set second similarity threshold, the traversal is also terminated.

[0034] Furthermore, the selection range of historical data when calculating the qualification parameters is limited to the same sports event. Of course, the selection range of historical data can also be narrowed down by parameters such as the student's age and height.

[0035] Furthermore, recommending motion images for students for reference includes the following steps:

[0036] Step 301, calculate the collaborative parameters of the historical data by calculating the first motion images of the historical data;

[0037] Step 302, extract the motion images and image motion data of the top N historical data with the largest collaborative parameters;

[0038] Step 303, extract the first motion images of the current student that are mapped to the first motion images in the motion images of the extracted historical data;

[0039] Step 304, display the motion images and image motion data of the extracted historical data and the mapping relationship of the first motion images of the current student with existing mappings to the student.

[0040] Further, the method for calculating the cooperation parameter of historical data in step 301 includes:

[0041] Generating a second set of motion images based on the first motion images in the first set of motion images of historical data that are mapped to the first motion image of the current student;

[0042] The calculation formula for the cooperation parameter is as follows:

[0043]

[0044] Where Y1 = 1 indicates that there is a mapping between the first motion image in the historical data and the first first motion image in the second set of motion images of the current historical data, and Y i = 1 indicates that there is a mapping between the first motion image in the historical data and the i-th first motion image in the second set of motion images of the current historical data, and paY i = 1 indicates that there are mappings between the first motion image in the historical data and all the first motion images before the i-th first motion image in the second set of motion images of the current historical data; y = 1 indicates that the corresponding evaluation result of the historical data is qualified, and m is the number of motion images in the second set of motion images of the current historical data.

[0045] The beneficial effects of the present invention are as follows:

[0046] The present invention uses AI vision technology to identify the motion data of students based on the collected motion images, and performs probability comprehensive calculation of qualified parameters based on the associated historical data of the motion data to accurately evaluate the motion items of students, and extracts the motion images corresponding to the historical data for reference based on probability comprehensive analysis to assist students in self-learning. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 is a schematic diagram of the modules of an intelligent teaching system based on AI vision technology of the present invention;

[0048] Figure 2 is a flowchart of the method for extracting key images of the present invention;

[0049] Figure 3 is a flowchart of the method for establishing a mapping relationship between the motion images in the first set of historical motion images and the motion images in the first set of motion images of the current motion item of the current student of the present invention;

[0050] Figure 4 is a flowchart of the method for recommending motion images for students to refer to of the present invention.

[0051] In the figure: Image acquisition module 101, motion data extraction module 102, first image screening module 103, historical image mapping module 104, qualified parameter calculation module 105, result evaluation module 106, teaching guidance module 107. Detailed implementation

[0052] Now, the subject matter described herein will be discussed with reference to exemplary embodiments. It should be understood that discussing these embodiments is only to enable those skilled in the art to better understand and thus implement the subject matter described herein. Without departing from the scope of protection of the content of this specification, changes can be made to the functions and arrangements of the elements discussed. Each example can omit, substitute, or add various processes or components as needed. Additionally, the features described relative to some examples can also be combined in other examples.

[0053] Embodiment 1

[0054] As Figures 1-4 shown, an intelligent teaching system based on AI vision technology includes:

[0055] An image acquisition module 101, which is used to acquire the motion images of students;

[0056] A motion data extraction module 102, which obtains corresponding image motion data based on the acquired motion image processing;

[0057] One image motion data is obtained by processing one corresponding motion image. A student will acquire multiple motion images sorted by time frames in a motion event. Therefore, multiple image motion data sorted by time frames can be processed and obtained;

[0058] The image motion data generated from the motion images of one motion event of one student is classified into an image motion data set and sorted by time frames;

[0059] The image motion data includes the motion parameters of the human skeleton;

[0060] Identifying the human body and human postures through image data processing is a conventional technical means in the field of image processing technology. Optionally but not limited to, the following algorithms can be used for implementation: openpose algorithm, DeepCut algorithm, MaskRCNN algorithm;

[0061] A first image screening module 103, which is used to extract key images from the motion images;

[0062] The method for extracting key images includes:

[0063] Step 101, select the motion image with the earliest time frame in the image motion data set as the starting motion image;

[0064] Step 102: Traverse the motion images backward in the order of time frames starting from the starting motion image. The traversal terminates when the first similarity between the currently traversed motion image and the starting motion image is less than the set first similarity threshold; at this time, record the currently traversed motion image as the starting motion image.

[0065] In an embodiment of the present invention, the calculation formula for the first similarity is as follows:

[0066]

[0067] where T 1i represents the angle of the i-th limb of the human skeleton corresponding to the image motion data of a motion image, and T 2i represents the angle of the i-th limb of the human skeleton corresponding to the image motion data of another motion image, and n is the number of limbs of the human skeleton;

[0068] The angle of the limb is the angle relative to the X coordinate axis in the plane;

[0069] Step 103: Update the motion image at the end of the traversal to the new first motion image;

[0070] Step 104: Iteratively execute Steps 102 and 103 until all motion images are traversed;

[0071] Step 105: Generate a first motion image set from the first motion image;

[0072] The historical image mapping module 104 is used to establish a mapping relationship between the motion images of the historical first motion image set and the first motion image set of the current student's current sports event;

[0073] The method of establishing a mapping relationship between the motion images of the historical first motion image set and the first motion image set of the current student's current sports event includes:

[0074] Step 201: Extract a first motion image with the earliest time frame from the first motion image set of the current student as the reference motion image;

[0075] Step 202: Traverse the first motion images in the historical first motion image set in sequence. The termination condition for the traversal is:

[0076] The first similarity between the currently traversed first motion image and the reference motion image is greater than the set second similarity threshold;

[0077] Establish a mapping relationship between the first motion image at the end of the traversal and the first motion image of the current student;

[0078] If the first similarity between all the first motion images in the historical data and the reference motion image is less than or equal to the set second similarity threshold, the traversal is also terminated;

[0079] Step 203, after the traversal is terminated, extract the first motion image that is the next one (in terms of time frame) of the current reference motion image from the set of the first motion images of the current student, and update it as the new reference motion image;

[0080] Step 204, iteratively execute Steps 202 and 203 until all the first motion images in the set of the first motion images of the current student are extracted;

[0081] The passing parameter calculation module 105 is used to calculate the passing parameter;

[0082] The calculation formula of the passing parameter is as follows:

[0083]

[0084] Where X1 = 1 indicates that there is a mapping (as an event) between the first motion image in the historical data and the first first motion image in the set of the first motion images of the current student, X i = 1 indicates that there is a mapping between the first motion image in the historical data and the i-th first motion image in the set of the first motion images of the current student, paX i = 1 indicates that there are mappings between the first motion image in the historical data and all the first motion images before the i-th first motion image in the set of the first motion images of the current student; y = 1 indicates that the corresponding evaluation result of this historical data is qualified. For example, if i = 4, then PX i = 1paX i = 1 = PX4 = 1X1 = 1, X2 = 1, X3 = 1, and n is the number of motion images in the set of the first motion images of the current student;

[0085] The selection range of the historical data is limited to the same sports event. Of course, the selection range of the historical data can also be narrowed down by parameters such as the age and height of the student;

[0086] The result evaluation module 106 evaluates the result of the current sports event of the current student based on the passing parameter. If the evaluation parameter is greater than the set first evaluation parameter threshold, it is judged as qualified; otherwise, it is judged as unqualified.

[0087] The teaching guidance module 107 is used to recommend motion images for the student to refer to when the evaluation parameter of the current sports event of the current student is less than the set second evaluation parameter threshold;

[0088] Recommending motion images for the student to refer to includes the following steps:

[0089] Step 301: Calculate the first motion image of historical data to calculate the collaboration parameter of historical data;

[0090] Generate a second motion image set based on the first motion image in the first motion image set of historical data that maps to the first motion image of the current student;

[0091] The calculation formula of the collaboration parameter is as follows:

[0092]

[0093] Where Y1 = 1 indicates that there is a first motion image in the historical data that maps to the first first motion image in the second motion image set of the current historical data, and Y i = 1 indicates that there is a first motion image in the historical data that maps to the i-th first motion image in the second motion image set of the current historical data, and paY i = 1 indicates that there are mappings of all first motion images in the historical data before the i-th first motion image in the second motion image set of the current historical data; y = 1 indicates that the corresponding evaluation result of the historical data is qualified, and m is the number of motion images in the second motion image set of the current historical data;

[0094] Step 302: Extract the motion images and image motion data of the top N historical data with the largest collaboration parameter;

[0095] Step 303: Extract the first motion image of the current student that maps to the first motion image in the motion images of the extracted historical data;

[0096] Step 304: Display the mapping relationship between the extracted motion images and image motion data of historical data and the first motion image of the current student with mappings to the student.

[0097] The above has described the embodiments of this example, but this example is not limited to the above specific implementation manners. The above specific implementation manners are only illustrative and not restrictive. Under the inspiration of this example, those of ordinary skill in the art can also make many forms, all of which fall within the protection scope of this example.

Claims

1. An intelligent teaching system based on AI vision technology, characterized in that, Including: An image acquisition module, which is used to acquire the motion images of students; A motion data extraction module, which obtains corresponding image motion data based on the acquired motion image processing; The image motion data generated from the motion images of a single sports event of a single student is classified into an image motion data set and sorted according to time frames; A first image screening module, which is used to extract key images from the motion images; A historical image mapping module, which is used to establish a mapping relationship between the motion images of the historical first motion image set and the first motion image set of the current student's current sports event; A qualified parameter calculation module, which is used to calculate qualified parameters; The calculation formula of the qualified parameter is as follows: ； Among them It means that there is a mapping between the first motion image in the historical data and the first first motion image in the first motion image set of the current student It means that there is a mapping between the first motion image in the historical data and the i-th first motion image in the first motion image set of the current student It means that there are mappings between the first motion image in the historical data and all the first motion images before the i-th first motion image in the first motion image set of the current student; y = 1 indicates that the corresponding evaluation result of the historical data is qualified, and n is the number of motion images in the first motion image set of the current student A result evaluation module, which evaluates the result of the current student's current sports event based on the qualified parameter. If the evaluation parameter is greater than the set first evaluation parameter threshold, it is judged as qualified; otherwise, it is judged as unqualified; A teaching guidance module, which is used to recommend motion images for reference to students when the evaluation parameter of the current student's current sports event is less than the set second evaluation parameter threshold.

2. The intelligent teaching system based on AI vision technology according to claim 1, wherein The method for extracting key images includes: Step 101, select the motion image with the earliest time frame in the image motion data set as the starting motion image; Step 102, start traversing the motion images backward in the order of time frames from the starting motion image. When the first similarity between the currently traversed motion image and the starting motion image is less than the set first similarity threshold, terminate the traversal; at this time, record the currently traversed motion image as the starting motion image; Step 103, update the motion image at the end of the traversal as the new first motion image; Step 104, iteratively execute Steps 102 and 103 until all motion images are traversed; Step 105, generate a first motion image set from the first motion images.

3. An intelligent teaching system based on AI vision technology according to claim 2, characterized in that, The calculation formula of the first similarity is as follows: ； wherein, represents the included angle of the i-th limb of the human skeleton corresponding to the image motion data of a moving image, represents the included angle of the i-th limb of the human skeleton corresponding to the image motion data of another moving image, and n is the number of limbs of the human skeleton.

4. An intelligent teaching system based on AI vision technology according to claim 1, characterized in that, One motion image corresponds to one processed image motion data. A single student will acquire multiple motion images sorted by time frames in a single sports event, and multiple image motion data sorted by time frames can be processed and obtained.

5. An intelligent teaching system based on AI vision technology according to claim 1, characterized in that, The method for establishing a mapping relationship between the motion images of the historical first motion image set and the first motion image set of the current student's current sports event includes: Step 201, extract the earliest first motion image in the first motion image set of the current student as the reference motion image; Step 202, sequentially traverse the first motion images in the historical first motion image set. The termination condition for traversal is: The first similarity between the currently traversed first motion image and the reference motion image is greater than the set second similarity threshold; Establish a mapping relationship between the first motion image at the end of the traversal and the first motion image of the current student; Step 203, after the traversal ends, extract the first motion image after the current reference motion image in the first motion image set of the current student and update it as the new reference motion image; Step 204, iteratively execute Steps 202 and 203 until all first motion images in the first motion image set of the current student are extracted.

6. An intelligent teaching system based on AI vision technology according to claim 5, characterized in that, If the first similarity between all the first motion images obtained by traversing the historical data and the reference motion image is less than or equal to the set second similarity threshold, the traversal is also terminated.

7. An intelligent teaching system based on AI vision technology according to claim 1, characterized in that, The selection range of historical data when calculating the qualified parameters is limited to the same sports event, or the selection range of historical data is narrowed down by the age and height parameters of the student.

8. An intelligent teaching system based on AI vision technology according to claim 1, characterized in that, Recommending motion images for students for reference includes the following steps: Step 301, calculating the first motion images of the historical data to calculate the collaborative parameters of the historical data; Step 302, extracting the motion images and image motion data of the first N historical data with the largest collaborative parameters; Step 303, extracting the first motion image of the current student mapped to the first motion image in the motion images of the historical data extracted; Step 304, presenting the extracted motion images and image motion data of the historical data and the mapping relationship of the first motion image of the current student with the mapping to the student.

9. An intelligent teaching system based on AI vision technology according to claim 8, characterized in that, The method for calculating the collaborative parameters in Step 301 includes: Generating a second motion image set based on the first motion images in the first motion image set of the historical data that are mapped to the first motion image of the current student; The calculation formula for the collaborative parameters is as follows: ； Among them indicates that there is a mapping between the first moving image in the historical data and the first first moving image in the second moving image set of the current historical data indicates that there is a mapping between the first moving image in the historical data and the i-th first moving image in the second moving image set of the current historical data indicates that there are mappings between the first moving image in the historical data and all the first moving images before the i-th first moving image in the second moving image set of the current historical data; y = 1 indicates that the corresponding evaluation result of the historical data is qualified, and m is the number of moving images in the second moving image set of the current historical data

Images