An educational intelligent cloud platform and learning method based on big data processing

Abstract

The present invention discloses an educational intelligent cloud platform and a learning method based on big data processing, which relate to the technical field of painting education data processing. The method uses a brush with a pressure sensor to record the student's painting process, carefully captures information such as pressure changes, and provides a meticulous data basis for in-depth analysis. The construction of the data middle platform not only accepts painting images and recorded videos in various formats, but also makes the data easy to manage and analyze through identification and conversion. It solves the problem of restrictions on data formats in traditional teaching and improves the versatility and availability of data. The data is applied to deep learning to form multiple painting skill indicators. It not only combines data processing with deep learning, but also provides substantive technical support for subsequent evaluation and learning path generation. In S6, personalized evaluation and learning path generation are achieved by comparing skill indicators with standard thresholds, forming a set of efficient data automation processing processes.

Description

Technical Field

[0001] The present invention relates to the technical field of painting education data processing, and in particular to an education intelligent cloud platform and a learning method based on big data processing. Background Art

[0002] With the rapid development of the Internet and cloud computing technologies, the field of education has ushered in new opportunities and changes. Education cloud platforms have emerged as an important technical tool to support online courses, distance learning and virtual classrooms. This development trend provides teachers, students and educational institutions with more flexibility and diverse teaching methods, especially for students who cannot attend classes in person, such as distance education, adult education and online courses.

[0003] Painting art is an art form that requires long-term practice and cultivation. Traditional painting education often relies on teachers' subjective judgment and experience to evaluate students' paintings. This subjectivity may be affected by a variety of factors such as teachers' own aesthetic views, experience levels, personal preferences, etc., resulting in a lack of consistency and scientificity in the evaluation results. This has a negative impact on students' learning direction and growth because they cannot get accurate and objective feedback. In traditional painting education, feedback often relies on teachers' corrections to the works, which usually takes a certain amount of time. Therefore, students may have to wait for a long time to get feedback. Delayed feedback not only reduces students' real-time cognition of their own learning progress, but may also slow down their motivation to learn and affect teaching effectiveness. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention provides an educational intelligent cloud platform and a learning method based on big data processing, which solve the problems mentioned in the background technology.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: an educational intelligent learning method based on big data processing, comprising the following steps:

[0006] Step S1: During the painting process, the student uses a brush with a pressure sensor to paint a sketch image and records the whole process in video;

[0007] Step S2: The painting images, student information and painting recording videos in JPEG, PNG, BMP, TIFF or SVG format submitted by the students are stored in the data center, and the data center identifies the painting images and painting recording videos in different formats and converts them into a common format for storage;

[0008] Step S3, using the video processing tool FFmpeg to segment the student's painting recording video into a series of image frames, and after preprocessing the series of image frames, obtain a series of processed image frames and convert them into a time series data structure;

[0009] Step S4, using computer vision to track the brush motion of the time series data structure, and determining key frames from the tracked brush motion; and extracting painting skill-related features from each key frame to establish a skill feature set;

[0010] Step S5, establishing a painting skill feature model, and using the skill feature set as input into the painting skill feature model, calculating and analyzing to obtain line smoothness Phd, curve continuity Lgd, step accuracy ZqL, concentration Zzd, hand-eye coordination Xtd, and pressure motion coefficient YL;

[0011] Step S6, compare the line smoothness Phd, curve continuity Lgd, step accuracy ZqL, concentration Zzd, hand-eye coordination Xtd and pressure motion coefficient YL with the standard threshold Q respectively to obtain evaluation results, and generate a corresponding learning path according to the evaluation results.

[0012] Preferably, the data center is used to integrate and manage multi-source data types, including painting images, student information, and painting recording videos; set a common format, including JSON or XML, provide a student interface API, allow students to upload painting images in various formats, receive JPEG, PNG, BMP, TIFF, and SVG image formats, and receive painting recording videos in MP4 and AVI;

[0013] Identify painting images and convert them into PNG format for storage; identify painting recording videos and store them into AVI format.

[0014] Preferably, the step S3 includes:

[0015] S31. Install the FFmpeg tool in the data center, and use the FFmpeg tool to split the painting recording video into a series of image frames; set the output frame rate to 30 frames per second, formulate the save path and file name format of the output image frame, and use the four-digit number "%04d" to represent the frame number;

[0016] S32, preprocessing each image frame, the preprocessing including adjusting the image size, enhancing contrast and brightness, and removing noise;

[0017] S33, saving the preprocessed image frame in a specified folder, and using an image processing library tool to save the image;

[0018] S34, construct time series data, create a time series data structure, including the frame number and timestamp of each image frame, and associate the frame number and timestamp of each image frame to form a tuple of image data containing time information, and set it to timestamp, image_data format;

[0019] S35. Save the time series data structure as a binary file.

[0020] Preferably, the S4 step includes:

[0021] S41. Use the optical flow method in computer vision technology to detect the movement of pixels in adjacent frames in the image and track the brush trajectory;

[0022] S42, analyzing the speed and acceleration information of the brush movement, and when it is recognized that the brush movement speed exceeds a first threshold, determining a significant change in the brush movement as a first key frame;

[0023] S43, detecting the color change on the canvas, comparing the pixel values ​​between adjacent frames, and identifying when the color change exceeds a second threshold as a second key frame;

[0024] S44, combining the first key frame and the second key frame, and extracting line smoothness features, curve continuity features, recognition of painting tool pressure change features, color distribution features, and hand-eye coordination features;

[0025] Extract hand-eye coordination features by combining brush motion and students’ visual information in the video;

[0026] S45. Establish a skill feature set based on the features extracted in S44.

[0027] Preferably, the step S5 comprises:

[0028] S51, collecting labeled painting data, including key frames, skill feature sets, and skill evaluation data;

[0029] S52. Establish a painting skill feature model, use the skill feature set as input for deep learning and training, and analyze and calculate to obtain line smoothness Phd, curve continuity Lgd, step accuracy ZqL, concentration Zzd, hand-eye coordination Xtd, and pressure motion coefficient YL.

[0030] Preferably, the line smoothness Phd and curve coherence Lgd are generated by the following formula:

[0031]

[0032] Where: N is the number of key frames, Δθ i is the angle change between the i-th frame and the i+1-th frame, Δti is the time change between the i-th frame and the i+1-th frame.

[0033] Preferably, the step accuracy ZqL, concentration Zzd and hand-eye coordination Xtd are generated by the following formula:

[0034]

[0035] Where bzsL is the number of correctly executed steps, ZBZ is the total number of painting steps, dbhh is the time the student actually takes to paint, zhhsj is the time of the painting process in the entire video, zqcs is the number of correct movements, specifically the number of times the brush movement matches the target path, and zydcs is the total number of movements, which is the total number of brush movements.

[0036] Preferably, the pressure data of the painting tool with the pressure sensor is extracted, and the time series data P1, P2, ..., P N , where P i Indicates the pressure value at the i-th time point, the pressure increment M i Indicates the amplitude of pressure change, which is calculated by the following formula:

[0037] M i =|P i+1 -P i |

[0038] Then, calculate the pressure variation coefficient YL:

[0039]

[0040] Among them, Δt i It is expressed as the time change from the ith time point to the i+1th time point.

[0041] Preferably, the standard threshold Q includes a line smoothness threshold Q1, a curve continuity threshold Q2, a step accuracy threshold Q3, a concentration threshold Q4, a hand-eye coordination threshold Q5, and a pressure motion threshold Q6; the line smoothness Phd, the curve continuity Lgd, the step accuracy ZqL, the concentration Zzd, the hand-eye coordination Xtd, and the pressure motion coefficient YL are compared and evaluated with the corresponding line smoothness threshold Q1, the curve continuity threshold Q2, the step accuracy threshold Q3, the concentration threshold Q4, the hand-eye coordination threshold Q5, and the pressure motion threshold Q6 to obtain the following evaluation results, including:

[0042] Compare the line smoothness Phd with the line smoothness threshold Q1. If the line smoothness Phd is greater than or equal to the line smoothness threshold Q1, it is qualified; if the line smoothness Phd is lower than the line smoothness threshold Q1, it is unqualified, and a first drawing training path is generated; the first drawing training path includes providing additional practice training tasks for line smoothness;

[0043] Compare the curve continuity Lgd with the curve continuity threshold Q2. If the curve continuity Lgd is greater than or equal to the curve continuity threshold Q2, it indicates that the test is qualified. If the curve continuity Lgd is lower than the curve continuity threshold Q2, it indicates that the test is unqualified. A second drawing training path is generated. The second drawing training path includes exercises and skill training tasks that emphasize curves.

[0044] Compare the step accuracy rate ZqL with the step accuracy rate threshold Q3. If the step accuracy rate ZqL is higher than or equal to the step accuracy rate threshold Q3, it indicates that the test is qualified. If the step accuracy rate ZqL is lower than the step accuracy rate threshold Q3, it indicates that the test is unqualified, and a third drawing training path is generated. The third drawing training path includes exercises and training tasks that focus on the order of steps.

[0045] Compare the concentration Zzd with the concentration threshold Q4. If the concentration Zzd is higher than or equal to the concentration threshold Q4, it means that the training is qualified. If the concentration Zzd is lower than the concentration threshold Q4, it means that the training is unqualified. A fourth painting training path is generated. The fourth painting training path includes practice tasks for improving concentration.

[0046] The hand-eye coordination degree Xtd is compared with the hand-eye coordination degree threshold Q5. If the hand-eye coordination degree Xtd is greater than or equal to the hand-eye coordination degree threshold Q5, it is qualified; if the hand-eye coordination degree Xtd is lower than the hand-eye coordination degree threshold Q5, it is unqualified, and a fifth painting training path is generated; the fifth painting training path includes practice tasks that emphasize hand-eye coordination;

[0047] Compare the pressure motion coefficient YL with the pressure motion threshold Q6. When the pressure motion coefficient YL is greater than or equal to the pressure motion threshold Q6, it indicates that the test is qualified. If the pressure motion coefficient YL is lower than the pressure motion threshold Q6, it indicates that the test is unqualified, and a sixth drawing training path is generated. The sixth drawing training path includes a practice task that focuses on uniform pressure.

[0048] Based on the first painting training path, the second painting training path, the third painting training path, the fourth painting training path, the fifth painting training path and the sixth painting training path, in combination with public channels, search for related training feature libraries, and generate corresponding prediction content, as well as the improvement of the overall painting performance by the prediction content; and through the prediction content, set goals for students and track their progress in painting skills.

[0049] An educational intelligent cloud platform based on big data processing, including:

[0050] The data collection module is used to collect painting data from students, including painting images, student information, and painting recording videos, and provides a student interface API to allow students to upload data;

[0051] Data storage module: The data center is responsible for integrating and managing multi-source data types, identifying and converting painting images and videos into common formats, and providing redundant backup and data security mechanisms;

[0052] The data preprocessing module uses the FFmpeg tool to segment the painting recording video, generate a series of image frames, preprocess the image frames, including resizing, enhancing contrast and brightness, and removing noise, construct a time series data structure, and save it as a binary file;

[0053] The computer vision analysis module uses optical flow technology to track brush motion and extract key frames. Through brush motion and color changes, it determines key frames and extracts features related to painting skills. It provides brush motion speed and acceleration information and analyzes brush trajectory and color changes.

[0054] The skill feature model module collects marked painting data, builds a painting skill feature model, and uses deep learning and training models to calculate line smoothness Phd, curve continuity Lgd, step accuracy ZqL, concentration Zzd, hand-eye coordination Xtd, and pressure motion coefficient YL;

[0055] The evaluation and learning path generation module evaluates the line smoothness Phd, curve continuity Lgd, step accuracy ZqL, concentration Zzd, hand-eye coordination Xtd and pressure motion coefficient YL, generates evaluation results, and generates corresponding painting training paths based on the evaluation results. Each training path provides personalized exercises and training tasks for different painting skill requirements.

[0056] The present invention provides an educational intelligent cloud platform and a learning method based on big data processing, which have the following beneficial effects:

[0057] (1) This is an educational intelligent learning method based on big data processing. Through steps S1-S6, students use a brush with a pressure sensor to draw sketch images during the painting process and record the video throughout the process. Step S2 stores the painting images and recorded videos submitted by students in the data center, and uses the data center to identify and convert data in different formats. Through step S3, the video processing tool FFmpeg is used to segment the student's painting recording video into image frames, and the image frames are preprocessed to obtain a time series data structure. In step S4, the brush motion is tracked on the time series data through computer vision technology, key frames are extracted, and a skill feature set is established. In step S5, a painting skill feature model is established, and the skill feature set is deeply learned and analyzed to obtain multiple painting skill indicators, including line smoothness Phd, curve coherence Lgd, step accuracy ZqL, concentration Zzd, hand-eye coordination Xtd, and pressure motion coefficient YL. Finally, in step S6, these skill indicators are compared with the set standard thresholds to generate evaluation results, and corresponding learning paths are generated based on the evaluation results. This step enables objective assessment and personalized feedback of students' drawing skills.

[0058] (2) This is an educational intelligent cloud platform based on big data processing. Through the data middle platform and real-time processing technology, the system reduces the time students spend waiting for corrections in the traditional teaching model and realizes real-time feedback to students. At the same time, the system can track students' progress in painting skills, adjust learning plans and goals in a timely manner, and improve learning motivation and effectiveness. It integrates multiple data types such as painting images, student information, and painting recording videos, and applies data processing, computer vision, and deep learning technologies. This comprehensive application of multiple technologies makes the system more comprehensive, able to more finely analyze students' painting process and skill performance, and provide more comprehensive evaluation and guidance. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 This is a schematic diagram of the steps of an educational intelligent learning method based on big data processing according to the present invention.

[0060] Figure 2 This is a flowchart diagram of an educational intelligent cloud platform based on big data processing according to the present invention. DETAILED DESCRIPTION

[0061] 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.

[0062] Example 1

[0063] The present invention provides an educational intelligent learning method based on big data processing, please refer to Figure 1 , including the following steps:

[0064] Step S1: During the painting process, the student uses a brush with a pressure sensor to paint a sketch image and records the whole process in video;

[0065] Step S2: The painting images, student information and painting recording videos in JPEG, PNG, BMP, TIFF or SVG format submitted by the students are stored in the data center, and the data center identifies the painting images and painting recording videos in different formats and converts them into a common format for storage;

[0066] Step S3, using the video processing tool FFmpeg to segment the student's painting recording video into a series of image frames, and after preprocessing the series of image frames, obtain a series of processed image frames and convert them into a time series data structure;

[0067] Step S4, using computer vision to track the brush motion of the time series data structure, and determining key frames from the tracked brush motion; and extracting painting skill-related features from each key frame to establish a skill feature set;

[0068] Step S5, establishing a painting skill feature model, and using the skill feature set as input into the painting skill feature model, calculating and analyzing to obtain line smoothness Phd, curve continuity Lgd, step accuracy ZqL, concentration Zzd, hand-eye coordination Xtd, and pressure motion coefficient YL;

[0069] Step S6, compare the line smoothness Phd, curve continuity Lgd, step accuracy ZqL, concentration Zzd, hand-eye coordination Xtd and pressure motion coefficient YL with the standard threshold Q respectively to obtain evaluation results, and generate a corresponding learning path according to the evaluation results.

[0070] In this embodiment, first, through step S1, the student uses a brush with a pressure sensor to draw a sketch image during the painting process, and records the video throughout the process. This step uses the pressure sensor of modern technology to capture the student's painting process in more detail, including information such as the pressure change of the brush. Step S2 stores the painting images and recorded videos submitted by the students in the data center, and identifies and converts data in different formats through the data center. This step provides support for multiple data formats, making the data easier to manage and process. Through step S3, the video processing tool FFmpeg is used to segment the student's painting recording video into image frames, and the image frames are pre-processed to obtain a time series data structure. This step uses video processing tools to convert video data into time series data that is easier to analyze, providing a basis for subsequent brush motion tracking. In step S4, the time series data is tracked by brush motion through computer vision technology, key frames are extracted, and a skill feature set is established. This step uses advanced computer vision technology to objectively capture the characteristics related to students' painting skills. In step S5, a painting skill feature model is established, and the skill feature set is deeply learned and analyzed to obtain multiple painting skill indicators, including line smoothness Phd, curve coherence Lgd, step accuracy ZqL, concentration Zzd, hand-eye coordination Xtd, and pressure motion coefficient YL. Finally, in step S6, these skill indicators are compared with the set standard thresholds to generate evaluation results, and corresponding learning paths are generated based on the evaluation results. This step realizes the objective evaluation and personalized feedback of students' painting skills, which is expected to improve students' learning effects and motivation. Promoting students' painting skill evaluation and personalized learning paths makes up for the shortcomings of traditional painting education and provides students with more effective learning support.

[0071] Embodiment 2, this embodiment is an explanation of the embodiment 1. Specifically, the data center is used to integrate and manage multi-source data types, including painting images, student information, and painting recording videos; set a common format, including JSON or XML, provide a student interface API, allow students to upload painting images in various formats, receive JPEG, PNG, BMP, TIFF, and SVG image formats, and receive MP4 and AVI painting recording videos;

[0072] The painting images are identified and uniformly converted into PNG format for storage; the PNG format usually supports lossless compression, which helps to preserve the quality of the image. Secondly, the use of a unified format is conducive to data standardization and management. This helps to improve the stability and efficiency of the method of the present application. The painting recording videos are identified and uniformly stored in AVI format. The unified identification and storage of painting recording videos in AVI format also follows the principles of uniformity and standards. AVI is a universal multimedia container format that supports multiple encoding methods. Such a choice helps compatibility and scalability.

[0073] This embodiment helps to improve the maintainability, scalability and user-friendliness of the data center. Students and teachers can use the data center more easily, and the data processing inside the data center is more efficient. These advantages together promote the effective operation of the painting education data center.

[0074] Embodiment 3: This embodiment is an explanation of Embodiment 1. Specifically, step S3 includes:

[0075] S31. Install the FFmpeg tool in the data center, and use the FFmpeg tool to split the painting recording video into a series of image frames; set the output frame rate to 30 frames per second, formulate the save path and file name format of the output image frames, and use the four-digit number "%04d" to represent the frame number; by setting a suitable output frame rate, you can balance the image quality and data processing efficiency, and ensure that a sufficient number of image frames are extracted to capture the details of the painting process.

[0076] S32, preprocessing each image frame, the preprocessing including adjusting the image size, enhancing contrast and brightness, and removing noise;

[0077] S33, saving the preprocessed image frame in a specified folder, and using an image processing library tool to save the image;

[0078] S34, construct time series data, create a time series data structure, including the frame number and timestamp of each image frame, and associate the frame number and timestamp of each image frame to form a tuple of image data containing time information, and set it to timestamp, image_data format;

[0079] S35. Save the time series data structure as a binary file. Binary files have smaller storage space and faster reading and writing speeds, which helps improve the efficiency and response speed of the data center.

[0080] In this embodiment, the painting recording video can be processed efficiently while retaining enough information for subsequent skill analysis. This helps to improve the performance of the data center, making it more suitable for large-scale painting data processing and analysis.

[0081] Embodiment 4: This embodiment is an explanation of Embodiment 1. Specifically, the step S4 includes:

[0082] S41. Use the optical flow method in computer vision technology to detect the movement of pixels in adjacent frames in the image and track the brush trajectory. Use the optical flow method in computer vision technology to detect the movement of pixels in adjacent frames in the image, so as to track the brush trajectory. This helps to capture the movement of the brush during the painting process and provide key data for subsequent analysis.

[0083] S42, analyzing the speed and acceleration information of the brush movement, and when it is recognized that the brush movement speed exceeds a first threshold, determining a significant change in the brush movement as a first key frame; this helps to determine important nodes in the painting process.

[0084] S43, detecting the color change on the canvas, comparing the pixel values ​​between adjacent frames, and identifying when the color change exceeds the second threshold as the second key frame; this helps to identify the color changes in different areas on the canvas and provide multi-dimensional features for subsequent analysis.

[0085] S44. Combine the first key frame and the second key frame, and extract line smoothness features, curve continuity features, pressure change features for identifying painting tools, color distribution features, and hand-eye coordination features; this helps to more comprehensively understand students' performance in the painting process and provide more dimensional information for the establishment of a skill feature set.

[0086] Extract hand-eye coordination features by combining brush motion and students’ visual information in the video;

[0087] S45, establishing a skill feature set based on the features extracted in S44. This feature set contains key painting skill information and provides a basis for subsequent skill evaluation and personalized learning path generation.

[0088] In this embodiment, step S4 provides comprehensive and in-depth information on painting skills for the data center by fully considering various aspects of feature extraction, laying a solid foundation for subsequent analysis and evaluation. This helps to achieve comprehensive monitoring and analysis of the student's painting process.

[0089] Example 5: This example is an explanation of Example 1. Specifically, the step S5 includes:

[0090] S51. Collecting labeled painting data, including key frames, skill feature sets, and skill evaluation data; By collecting labeled painting data, including key frames, skill feature sets, and skill evaluation data, a rich dataset is established. These labeled data are the basis for deep learning model training and evaluation, ensuring the effectiveness of the model in real painting scenarios.

[0091] S52, establish a painting skill feature model, use the skill feature set as input for deep learning and training, and analyze and calculate to obtain line smoothness Phd, curve coherence Lgd, step accuracy ZqL, concentration Zzd, hand-eye coordination Xtd and pressure motion coefficient YL. Design and establish a deep learning model, which uses the skill feature set as input for training. This model can learn and understand the multi-dimensional features extracted from the painting process, so as to more accurately capture and evaluate students' painting skills. Use the collected labeled data for deep learning and model training. Deep learning technology has a strong model learning ability and can learn effective feature representations from a large amount of painting data, thereby improving the generalization performance of the model. After the training is completed, the established painting skill feature model is analyzed and calculated. This step extracts painting skill indicators such as line smoothness Phd, curve coherence Lgd, step accuracy ZqL, concentration Zzd, hand-eye coordination Xtd and pressure motion coefficient YL from the skill feature set through a deep learning model, thereby realizing the automated evaluation of students' painting ability.

[0092] The line smoothness Phd and curve continuity Lgd are generated by the following formula:

[0093]

[0094] Where: N is the number of key frames, Δθ i is the angle change between the i-th frame and the i+1-th frame, Δt i It is the time change between the i-th frame and the i+1-th frame. The calculation of line smoothness Phd is to sum and average the ratio of the angle change between each two adjacent key frames divided by the time change, thus reflecting the smoothness of the line in time. The larger the calculated result, the smoother the line change.

[0095] The calculation method of curve coherence Lgd is similar to that of line smoothness. It is also used to evaluate the coherence of the curve by summing and averaging the ratio of the angle change divided by the time change between each two adjacent keyframes. The larger the calculated result, the more coherent the curve change. The generation formula of line smoothness Phd and curve coherence Lgd uses the measurement of angle change and time change to quantify the smoothness of the line and the coherence of the curve during the drawing process.

[0096] The accuracy rate ZqL, concentration Zzd and hand-eye coordination Xtd of the step are generated by the following formula:

[0097]

[0098] In the formula, bzsL represents the number of steps executed correctly, ZBZ represents the total number of painting steps; dbhh represents the time when students actually pick up the pen to paint, zhhsj represents the time of the painting process in the entire video; zqcs represents the number of correct movements, specifically the number of times the brush movement matches the target path; zydcs represents the total number of movements, which is the total number of brush movements. The step accuracy ZqL is calculated by dividing the number of steps executed correctly by the total number of painting steps to get a percentage. This percentage indicates the degree to which students correctly execute the steps in the entire painting process and is used to evaluate the accuracy of students' steps. The concentration Zzd is calculated by dividing the time when students actually pick up the pen to paint by the time of the painting process in the entire video to get a percentage. This percentage indicates the degree of concentration of students in the painting process and is used to evaluate students' concentration. The hand-eye coordination Xtd is calculated by dividing the number of correct movements by the total number of movements to get a percentage. This percentage indicates the degree of hand-eye coordination of students in the painting process and is used to evaluate students' hand-eye coordination ability.

[0099] Specifically, the pressure data of the painting tool with the pressure sensor is extracted, and the time series data P1, P2, ..., P N , where P i Indicates the pressure value at the i-th time point, the pressure increment M i Indicates the amplitude of pressure change, which is calculated by the following formula:

[0100] M i =|P i+1 -P i |

[0101] Then, calculate the pressure variation coefficient YL:

[0102]

[0103] Among them, Δt i It is expressed as the time change from the ith time point to the i+1th time point.

[0104] The pressure change coefficient YL is calculated by dividing the pressure change by the corresponding time change. This coefficient can be used to evaluate the speed of pressure change during the painting process of students, providing information about pressure changes for the analysis of painting skills.

[0105] Example 6. This example is an explanation of Example 5. Specifically, the standard threshold Q includes a line smoothness threshold Q1, a curve continuity threshold Q2, a step accuracy threshold Q3, a concentration threshold Q4, a hand-eye coordination threshold Q5, and a pressure motion threshold Q6. The line smoothness Phd, the curve continuity Lgd, the step accuracy ZqL, the concentration Zzd, the hand-eye coordination Xtd, and the pressure motion coefficient YL are compared and evaluated with the corresponding line smoothness threshold Q1, the curve continuity threshold Q2, the step accuracy threshold Q3, the concentration threshold Q4, the hand-eye coordination threshold Q5, and the pressure motion threshold Q6 to obtain the following evaluation results, including:

[0106] Compare the line smoothness Phd with the line smoothness threshold Q1. If the line smoothness Phd is greater than or equal to the line smoothness threshold Q1, it is qualified; if the line smoothness Phd is lower than the line smoothness threshold Q1, it is unqualified, and a first drawing training path is generated; the first drawing training path includes providing additional practice training tasks for line smoothness;

[0107] Compare the curve continuity Lgd with the curve continuity threshold Q2. If the curve continuity Lgd is greater than or equal to the curve continuity threshold Q2, it indicates that the test is qualified. If the curve continuity Lgd is lower than the curve continuity threshold Q2, it indicates that the test is unqualified. A second drawing training path is generated. The second drawing training path includes exercises and skill training tasks that emphasize curves.

[0108] Compare the step accuracy rate ZqL with the step accuracy rate threshold Q3. If the step accuracy rate ZqL is higher than or equal to the step accuracy rate threshold Q3, it indicates that the test is qualified. If the step accuracy rate ZqL is lower than the step accuracy rate threshold Q3, it indicates that the test is unqualified, and a third drawing training path is generated. The third drawing training path includes exercises and training tasks that focus on the order of steps.

[0109] Compare the concentration Zzd with the concentration threshold Q4. If the concentration Zzd is higher than or equal to the concentration threshold Q4, it means that the training is qualified. If the concentration Zzd is lower than the concentration threshold Q4, it means that the training is unqualified. A fourth painting training path is generated. The fourth painting training path includes practice tasks for improving concentration.

[0110] The hand-eye coordination degree Xtd is compared with the hand-eye coordination degree threshold Q5. If the hand-eye coordination degree Xtd is greater than or equal to the hand-eye coordination degree threshold Q5, it is qualified; if the hand-eye coordination degree Xtd is lower than the hand-eye coordination degree threshold Q5, it is unqualified, and a fifth painting training path is generated; the fifth painting training path includes practice tasks that emphasize hand-eye coordination;

[0111] The pressure movement coefficient YL is compared with the pressure movement threshold Q6. When the pressure movement coefficient YL is greater than or equal to the pressure movement threshold Q6, it indicates that it is qualified; if the pressure movement coefficient YL is lower than the pressure movement threshold Q6, it indicates that it is unqualified, and the sixth painting training path is generated; the sixth painting training path includes practice tasks that focus on uniform pressure sensitivity.

[0112] In this embodiment, through such personalized evaluation and training path generation, the method can better meet the painting needs of each student, provide targeted practice tasks, and thus improve the learning effect. This also reflects the advantages of introducing data processing and personalized analysis in painting education.

[0113] Embodiment 7, this embodiment is explained in Embodiment 6. Specifically, according to the first painting training path, the second painting training path, the third painting training path, the fourth painting training path, the fifth painting training path and the sixth painting training path, combined with public channels, the training library and resources related to each training feature are searched through public channels. This may include training resources in terms of line smoothness, curve continuity, step accuracy, concentration, hand-eye coordination, pressure motion coefficient, etc. Search the associated training feature library, and generate corresponding prediction content, as well as the improvement of the overall painting performance by the prediction content; the data middle station analyzes the generated prediction content and evaluates its improvement on the overall painting performance. This can be measured by comparing the students' paintings at different time points, the degree of improvement of skill characteristics, etc. And through the prediction content, set goals for students, and based on the analysis results, the data middle station sets specific painting goals for students. These goals can be to improve specific skill levels, complete a certain number of practice tasks, and track students' progress in painting skills.

[0114] In this embodiment, this step tracks the progress of students in painting skills in real time by continuously monitoring students' paintings, recording skill characteristics, and analyzing students' performance during the painting process. This helps to adjust the training plan in a timely manner and ensure a personalized learning path.

[0115] Example 8, please refer to Figure 2 , an educational intelligent cloud platform based on big data processing, including,

[0116] The data collection module is used to collect painting data from students, including painting images, student information, and painting recording videos, and provides a student interface API to allow students to upload data;

[0117] Data storage module: The data center is responsible for integrating and managing multi-source data types, identifying and converting painting images and videos into common formats, and providing redundant backup and data security mechanisms;

[0118] The data preprocessing module uses the FFmpeg tool to segment the painting recording video, generate a series of image frames, preprocess the image frames, including resizing, enhancing contrast and brightness, and removing noise, construct a time series data structure, and save it as a binary file;

[0119] The computer vision analysis module uses optical flow technology to track brush motion and extract key frames. Through brush motion and color changes, it determines key frames and extracts features related to painting skills. It provides brush motion speed and acceleration information and analyzes brush trajectory and color changes.

[0120] The skill feature model module collects marked painting data, builds a painting skill feature model, and uses deep learning and training models to calculate line smoothness Phd, curve continuity Lgd, step accuracy ZqL, concentration Zzd, hand-eye coordination Xtd, and pressure motion coefficient YL;

[0121] The evaluation and learning path generation module evaluates the line smoothness Phd, curve continuity Lgd, step accuracy ZqL, concentration Zzd, hand-eye coordination Xtd and pressure motion coefficient YL, generates evaluation results, and generates corresponding painting training paths based on the evaluation results. Each training path provides personalized exercises and training tasks for different painting skill requirements.

[0122] This educational intelligent cloud platform provides preliminary education, accurate painting skills assessment, supports personalized teaching, and significantly improves quality through real-time feedback and generation of learning paths.

[0123] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that many changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the invention.

Claims

1. An educational intelligent learning method based on big data processing, characterized in that: The following steps are involved: Step S1: During the painting process, the student uses a brush with a pressure sensor to paint a sketch image and records the whole process in video; Step S2: The painting images, student information and painting recording videos in JPEG, PNG, BMP, TIFF or SVG format submitted by the students are stored in the data center, and the data center identifies the painting images and painting recording videos in different formats and converts them into a common format for storage; Step S3, using the video processing tool FFmpeg to segment the student's painting recording video into a series of image frames, and after preprocessing the series of image frames, obtain a series of processed image frames and convert them into a time series data structure; Step S4, using computer vision to track the brush motion of the time series data structure, and determining key frames from the tracked brush motion; And extract painting skill-related features from each key frame to establish a skill feature set; Step S5, establishing a painting skill feature model, and using the skill feature set as input into the painting skill feature model, calculating and analyzing to obtain line smoothness Phd, curve continuity Lgd, step accuracy ZqL, concentration Zzd, hand-eye coordination Xtd, and pressure motion coefficient YL; The line smoothness Phd and curve continuity Lgd are generated by the following formula: Where: N is the number of key frames, Δθ i is the angle change between the i-th frame and the i+1-th frame, Δt i is the time change between the i-th frame and the i+1-th frame; The accuracy rate ZqL, concentration Zzd and hand-eye coordination Xtd of the step are generated by the following formula: Where bzsL is the number of correctly executed steps, ZBZ is the total number of painting steps; dbhh is the time when students actually pick up the pen to paint, zhhsj is the time of the painting process in the entire video; zqcs is the number of correct movements, specifically the number of times the brush movement matches the target path; zydcs is the total number of movements, which is the total number of brush movements; Extract the pressure data of the painting tool with a pressure sensor, when the time series data P1, P2, ..., P N , where P i Indicates the pressure value at the i-th time point, the pressure increment M i Indicates the amplitude of pressure change, which is calculated by the following formula: M i =|P i+1 -P i | Then, calculate the pressure variation coefficient YL: Among them, Δt i It is expressed as the time change from the i-th time point to the i+1-th time point; Step S6, compare the line smoothness Phd, curve coherence Lgd, step accuracy ZqL, concentration Zzd, hand-eye coordination Xtd and pressure motion coefficient YL with the standard threshold Q respectively to obtain the evaluation results, and generate the corresponding learning path according to the evaluation results; The standard threshold Q includes a line smoothness threshold Q1, a curve continuity threshold Q2, a step accuracy threshold Q3, a concentration threshold Q4, a hand-eye coordination threshold Q5, and a pressure motion threshold Q6; the line smoothness Phd, the curve continuity Lgd, the step accuracy ZqL, the concentration Zzd, the hand-eye coordination Xtd, and the pressure motion coefficient YL are compared and evaluated with the corresponding line smoothness threshold Q1, the curve continuity threshold Q2, the step accuracy threshold Q3, the concentration threshold Q4, the hand-eye coordination threshold Q5, and the pressure motion threshold Q6 to obtain the following evaluation results, including: Compare the line smoothness Phd with the line smoothness threshold Q1. If the line smoothness Phd is greater than or equal to the line smoothness threshold Q1, it is qualified; if the line smoothness Phd is lower than the line smoothness threshold Q1, it is unqualified, and a first drawing training path is generated; the first drawing training path includes providing additional practice training tasks for line smoothness; Compare the curve continuity Lgd with the curve continuity threshold Q2. If the curve continuity Lgd is greater than or equal to the curve continuity threshold Q2, it indicates that the test is qualified. If the curve continuity Lgd is lower than the curve continuity threshold Q2, it indicates that the test is unqualified. A second drawing training path is generated. The second drawing training path includes exercises and skill training tasks that emphasize curves. Compare the step accuracy rate ZqL with the step accuracy rate threshold Q3. If the step accuracy rate ZqL is greater than or equal to the step accuracy rate threshold Q3, it indicates that the test is qualified. If the step accuracy rate ZqL is lower than the step accuracy rate threshold Q3, it indicates that the test is unqualified, and a third drawing training path is generated. The third drawing training path includes exercises and training tasks that focus on the step sequence. Compare the concentration Zzd with the concentration threshold Q4. If the concentration Zzd is higher than or equal to the concentration threshold Q4, it means that the training is qualified. If the concentration Zzd is lower than the concentration threshold Q4, it means that the training is unqualified. A fourth painting training path is generated. The fourth painting training path includes practice tasks for improving concentration. The hand-eye coordination degree Xtd is compared with the hand-eye coordination degree threshold Q5. If the hand-eye coordination degree Xtd is greater than or equal to the hand-eye coordination degree threshold Q5, it is qualified; if the hand-eye coordination degree Xtd is lower than the hand-eye coordination degree threshold Q5, it is unqualified, and a fifth painting training path is generated; the fifth painting training path includes practice tasks that emphasize hand-eye coordination; Compare the pressure motion coefficient YL with the pressure motion threshold Q6. When the pressure motion coefficient YL is greater than or equal to the pressure motion threshold Q6, it indicates that the test is qualified. If the pressure motion coefficient YL is lower than the pressure motion threshold Q6, it indicates that the test is unqualified, and a sixth drawing training path is generated. The sixth drawing training path includes a practice task that focuses on uniform pressure. Based on the first painting training path, the second painting training path, the third painting training path, the fourth painting training path, the fifth painting training path and the sixth painting training path, in combination with public channels, search for related training feature libraries, and generate corresponding prediction content, as well as the improvement of the overall painting performance by the prediction content; and through the prediction content, set goals for students and track their progress in painting skills.

2. The educational intelligent learning method based on big data processing according to claim 1 is characterized by: The data center is used to integrate and manage multi-source data types, including painting images, student information, and painting recording videos; set common formats, including JSON or XML, provide student interface APIs, allow students to upload painting images in various formats, receive JPEG, PNG, BMP, TIFF, and SVG image formats, and receive painting recording videos in MP4 and AVI; Identify painting images and convert them into PNG format for storage; identify painting recording videos and store them into AVI format.

3. The educational intelligent learning method based on big data processing according to claim 1 is characterized by: The step S3 includes: S31, install FFmpeg tool in the data center, and use FFmpeg tool to split the painting recording video into a series of image frames; set the output frame rate to 30 frames per second, formulate the saving path and file name format of the output image frame, and use four-digit "%04d" to represent the frame number; S32, preprocessing each image frame, the preprocessing including adjusting the image size, enhancing contrast and brightness, and removing noise; S33, saving the preprocessed image frame in a specified folder, and using an image processing library tool to save the image; S34, construct time series data, create a time series data structure, including the frame number and timestamp of each image frame, and associate the frame number and timestamp of each image frame to form a tuple of image data containing time information, and set it to timestamp, image_data format; S35. Save the time series data structure as a binary file.

4. The educational intelligent learning method based on big data processing according to claim 1 is characterized by: The S4 step includes: S41. Use the optical flow method in computer vision technology to detect the movement of pixels in adjacent frames in the image and track the brush trajectory; S42, analyzing the speed and acceleration information of the brush movement, and when it is recognized that the brush movement speed exceeds a first threshold, determining a significant change in the brush movement as a first key frame; S43, detecting the color change on the canvas, comparing the pixel values ​​between adjacent frames, and identifying when the color change exceeds a second threshold as a second key frame; S44, combining the first key frame and the second key frame, and extracting line smoothness features, curve continuity features, recognition of painting tool pressure change features, color distribution features, and hand-eye coordination features; Extract hand-eye coordination features by combining brush motion and students’ visual information in the video; S45. Establish a skill feature set based on the features extracted in S44.

5. The educational intelligent learning method based on big data processing according to claim 1 is characterized by: The S5 step includes: S51, collecting labeled painting data, including key frames, skill feature sets, and skill evaluation data; S52. Establish a painting skill feature model, use the skill feature set as input for deep learning and training, and analyze and calculate to obtain line smoothness Phd, curve continuity Lgd, step accuracy ZqL, concentration Zzd, hand-eye coordination Xtd, and pressure motion coefficient YL.

6. An educational intelligent cloud platform based on big data processing, applied to an educational intelligent learning method based on big data processing as claimed in any one of claims 1 to 5, characterized in that: include, The data collection module is used to collect painting data from students, including painting images, student information, and painting recording videos, and provides a student interface API to allow students to upload data; Data storage module: The data center is responsible for integrating and managing multi-source data types, identifying and converting painting images and videos into common formats, and providing redundant backup and data security mechanisms; The data preprocessing module uses the FFmpeg tool to segment the painting recording video, generate a series of image frames, preprocess the image frames, including resizing, enhancing contrast and brightness, and removing noise, construct a time series data structure, and save it as a binary file; The computer vision analysis module uses optical flow technology to track brush motion and extract key frames. Through brush motion and color changes, it determines key frames and extracts features related to painting skills. It provides brush motion speed and acceleration information and analyzes brush trajectory and color changes. The skill feature model module collects marked painting data, builds a painting skill feature model, and uses deep learning and training models to calculate line smoothness Phd, curve continuity Lgd, step accuracy ZqL, concentration Zzd, hand-eye coordination Xtd, and pressure motion coefficient YL; The evaluation and learning path generation module evaluates the line smoothness Phd, curve continuity Lgd, step accuracy ZqL, concentration Zzd, hand-eye coordination Xtd and pressure motion coefficient YL, generates evaluation results, and generates corresponding painting training paths based on the evaluation results. Each training path provides personalized exercises and training tasks for different painting skill requirements.

Images