Homework correcting method and system based on image recognition

Abstract

The invention relates to the technical field of digital data processing, in particular to a homework correction method and system based on image recognition, and the method comprises the steps: recognizing the identity of a student who submits a homework based on an NFC technology; after completing the homework, the student shoots and uploads a homework image, and checks the integrity of the homework image; recognizing character information in the homework image by adopting an image recognition technology to obtain question doing data; matching the recognized question-answering data with answers in a system question bank to obtain a correction result; and obtaining a knowledge graph of the question chapter, and analyzing the correction result based on the knowledge graph to obtain the chapter mastery degree. Therefore, the method can achieve the quick and accurate recognition of the homework images of the students, and achieves the intelligent matching and correction of a question bank.

Description

Technical Field

[0001] The present invention relates to the technical field of digital data processing, and in particular to a homework marking method based on image recognition and a system thereof. Background Art

[0002] Currently, the work of grading students' homework mainly relies on the manual operation of teachers. This process is not only time-consuming and labor-intensive, but also easily affected by the subjective judgment of teachers, resulting in inconsistent evaluation standards. In addition, the traditional way of grading homework cannot achieve large-scale processing and rapid feedback of students' homework, which to some extent limits the improvement of teaching efficiency and teaching quality.

[0003] Specifically, the prior art has the following disadvantages:

[0004] 1) Inefficiency: Teachers need to check students’ homework one by one, which is especially obvious when there are a large number of students, resulting in a longer time-consuming marking work.

[0005] 2) High subjectivity: Due to the teacher’s personal experience and preference, the evaluation of assignments may be subjective, making it difficult to ensure objectivity and consistency of the evaluation.

[0006] 3) Delayed feedback: Students often have to wait a long time to get feedback on their assignments, which may affect their learning enthusiasm and their ability to adjust their learning strategies in a timely manner.

[0007] 4) Insufficient data analysis: Traditional marking methods make it difficult to conduct in-depth data analysis of students’ learning situations, and it is difficult for teachers to fully understand students’ learning status and knowledge mastery.

[0008] At present, there are some question recognition and search systems, homework grading systems, etc. based on image recognition technology on the market.

[0009] However, these systems often have the following drawbacks:

[0010] 1) Limited question bank matching capability: Some systems may not be able to accurately identify students’ answers when matching with the question bank, affecting the grading effect.

[0011] 2) Insufficient data analysis capabilities: Existing systems are often simplistic in data analysis and are unable to provide in-depth learning and teaching guidance.

[0012] 3) Poor user experience: The operating interface of some systems is not user-friendly and not convenient to use, which affects the user experience of teachers. Summary of the invention

[0013] The purpose of the present invention is to provide a homework grading method and system based on image recognition, aiming to develop a system that integrates advanced image recognition technology and intelligent comparison algorithm, which can realize fast and accurate recognition of student homework images, as well as intelligent matching and grading with question banks. In addition, the system also has data analysis function, which can provide teachers with comprehensive student learning status and knowledge mastery status.

[0014] To achieve the above-mentioned object, in a first aspect, the present invention provides a homework grading method based on image recognition, comprising identifying the identity of a student submitting a homework based on NFC technology;

[0015] After completing the assignment, students take and upload the assignment image, and check the integrity of the assignment image;

[0016] Image recognition technology is used to identify text information in homework images to obtain test data;

[0017] Match the identified test data with the answers in the system question bank to obtain the grading results;

[0018] Obtain the knowledge graph of the question chapter, and analyze the correction results based on the knowledge graph to obtain the chapter mastery.

[0019] The specific steps of identifying the identity of the student submitting the homework based on NFC technology include:

[0020] Each student is assigned an NFC tag with a built-in unique ID;

[0021] Bind the NFC tag with the student’s personal information and student ID number;

[0022] When students finish their homework, they put their own NFC tag on it;

[0023] The NFC tag will be read and the student's name, student number and submission time will be recorded.

[0024] The specific steps of taking and uploading the homework image after the student completes the homework and checking the integrity of the homework image include:

[0025] Use a camera to capture images of the work;

[0026] Check whether the clarity of the uploaded job image exceeds the preset value. If not, it is considered unqualified;

[0027] Use OCR technology to identify the boundaries in the job image detected by resolution and check whether there is any missing image;

[0028] The assignment images with missing images are sent to the teacher for manual review.

[0029] The specific steps of using the OCR technology to identify the boundary in the job image detected by resolution and checking whether there is a missing image include:

[0030] Preprocess the job image;

[0031] Use OCR technology to identify boundary data in pre-processed job images;

[0032] Extract the answer outline from the boundary data and extract the homework paper boundary;

[0033] If the distance from the answer outline to the border of the assignment paper is less than the preset value, it is considered that there is a missing value.

[0034] The specific steps of preprocessing the job image include:

[0035] Convert color images to grayscale images;

[0036] Convert grayscale images to black and white images;

[0037] Use filters to remove noise points in the image;

[0038] Deskew the image to ensure the text is level.

[0039] The specific steps of using image recognition technology to recognize text information in the homework image and obtain the homework data include:

[0040] Use support vector machine to recognize and splice the text to obtain the recognized text;

[0041] The recognized text is formatted according to the format required by the question to obtain the test data;

[0042] Store the test data in the database.

[0043] The specific steps of using a support vector machine to recognize and splice text to obtain recognized text include:

[0044] Split the entire text area into lines of text blocks, and then split each line of text blocks into individual words;

[0045] Extract recognition features from the segmented individual characters;

[0046] Use the trained support vector machine to perform text recognition based on recognition features;

[0047] Apply OCR algorithm to each text block to convert it into a text string;

[0048] The recognized text strings are concatenated in their original order to obtain the recognized text.

[0049] The specific steps of matching the identified test data with the answers in the system test database to obtain the correction results include:

[0050] Establish a database containing questions and model answers;

[0051] Extract keywords and key phrases from the recognized text;

[0052] Match the keywords and key phrases with the standard answers corresponding to the questions word by word and give scores based on the degree of match.

[0053] In a second aspect, the present invention also provides a homework grading system based on image recognition, comprising an identity recognition module, an image inspection module, an image recognition module, an answer matching module and a knowledge mastery assessment module;

[0054] The identity recognition module is used to identify the identity of the student submitting the homework based on NFC technology;

[0055] The image checking module is used for students to take and upload homework images after completing their homework, and to check the integrity of the homework images;

[0056] The image recognition module is used to recognize text information in the homework image using image recognition technology to obtain the homework data;

[0057] The answer matching module is used to match the identified test data with the answers in the system test database to obtain the correction results;

[0058] The knowledge mastery evaluation module is used to obtain the knowledge graph of the question chapter, and analyze the correction results based on the knowledge graph to obtain the chapter mastery.

[0059] The present invention discloses a homework grading method and system based on image recognition, comprising: identifying the identity of a student submitting homework based on NFC technology; taking and uploading a homework image after the student completes the homework, and checking the integrity of the homework image; using image recognition technology to identify text information in the homework image to obtain question data; matching the identified question data with the answers in the system question bank to obtain the grading result; obtaining the knowledge graph of the question chapter, and analyzing the grading result based on the knowledge graph to obtain the chapter mastery. When a student submits a homework, he or she first needs to verify his or her identity through Near Field Communication (NFC) technology. After the student completes the homework, he or she uses a mobile device (such as a smart phone or a tablet) to take a photo of the paper homework. The photo should clearly cover all the answer areas. The system automatically detects whether the uploaded image contains a complete homework page to avoid scoring errors caused by partial omissions. Using image processing and OCR technology, text information is extracted from the uploaded homework image. The system compares the extracted question data with the standard answers in the question bank to determine whether the answer to each question is correct. The system calls the associated knowledge graph according to the course chapter to which the homework belongs. Knowledge graph is a structured form of knowledge representation that helps understand the connections between knowledge points. Based on the knowledge graph, the system can further analyze the students' grading results to identify which knowledge points are well mastered and which need to be strengthened. This helps teachers understand the students' learning progress and adjust the teaching plan accordingly. This method of grading homework based on image recognition not only greatly reduces the workload of teachers, but also provides more personalized and timely feedback to students, thereby promoting the improvement of learning effects. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0061] Figure 1 It is a flow chart of a homework grading method based on image recognition according to the first embodiment of the present invention.

[0062] Figure 2 This is a flow chart of identifying the identity of a student submitting an assignment based on NFC technology according to the first embodiment of the present invention.

[0063] Figure 3 This is a flow chart of the first embodiment of the present invention in which a student takes and uploads an image of the homework after completing the homework, and checks the integrity of the homework image.

[0064] Figure 4 This is a flowchart of the first embodiment of the present invention for identifying a boundary in a job image detected by resolution using the OCR technology to check whether there is a missing image.

[0065] Figure 5 It is a flowchart of pre-processing a job image according to the first embodiment of the present invention.

[0066] Figure 6 This is a flowchart of the first embodiment of the present invention, which uses image recognition technology to recognize text information in a homework image and obtain question-solving data.

[0067] Figure 7 This is a flowchart of the first embodiment of the present invention for using a support vector machine to recognize and splice text to obtain recognized text.

[0068] Figure 8 This is a flowchart of the first embodiment of the present invention, which matches the identified test data with the answers in the system test database to obtain the correction results. DETAILED DESCRIPTION

[0069] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and should not be construed as limiting the present invention.

[0070] First embodiment

[0071] See also Figures 1 to 8 The present invention provides a homework correction method based on image recognition, comprising:

[0072] S101 uses NFC technology to identify the identity of students submitting assignments;

[0073] The specific steps include:

[0074] S201 assigns each student an NFC tag with a built-in unique ID;

[0075] Schools or educational institutions need to purchase enough NFC tags to ensure that each student has one. Each NFC tag has a unique serial number that will serve as the student's identity identifier. Distribute these tags to students and make sure that each student clearly understands how to use the NFC tag.

[0076] S202 binds the NFC tag with the student's personal information and student ID number;

[0077] Create a student information database that includes basic information such as student name, student ID, and class. Bind the unique serial number of the NFC tag to the student information and store it in the database. Ensure that the serial number of each NFC tag is accurately bound to the corresponding student information and verify the accuracy of the data through testing.

[0078] S203 When the student completes the homework, he / she puts his / her own NFC tag on the homework;

[0079] Provide students with simple training on how to properly attach NFC tags to their assignments. Provide written instructions so that students can follow the instructions correctly. Ensure that the tags are properly attached to the assignments to avoid reading failures due to improper attachment.

[0080] S204 will read the NFC tag and record the student's name, student number and submission time.

[0081] Set up an NFC card reader device in the homework submission area to ensure that it can work stably. When a student submits a homework with an NFC tag, the card reader will automatically read the information in the NFC tag and record the student's name, student number, and current timestamp. The data collected by the card reader will be synchronized to the school's database system for subsequent processing by teachers. In the event that the reading cannot be performed, there should be a backup plan, such as manually entering the student's student number and other information.

[0082] S102 After completing the assignment, students take and upload the assignment image, and check the integrity of the assignment image;

[0083] The specific steps include:

[0084] S301 uses a photographing device to photograph a work image;

[0085] Choose a smartphone or tablet with enough pixels as the shooting device. Ensure that there is sufficient lighting to avoid reflections and shadows so that the work content is clearly visible. Shoot vertically from directly above to ensure that the work paper appears completely in the image.

[0086] S302 detects whether the clarity of the uploaded job image exceeds a preset value, and if not, it is deemed unqualified;

[0087] Develop an algorithm or use an existing image processing library (such as OpenCV) to evaluate the clarity of the image. Set a clarity threshold below which images are considered unsatisfactory. For images with insufficient clarity, the system should prompt students to retake and upload them.

[0088] S303 uses OCR technology to identify the boundary in the job image detected by resolution and checks whether there is a missing image;

[0089] The specific steps include:

[0090] S401 pre-processes the job image;

[0091] The specific steps include:

[0092] S501 converts the color image into a grayscale image;

[0093] Use a grayscale conversion algorithm (such as cv2.cvtColor() function in OpenCV) to convert the color image to a grayscale image to reduce computational complexity.

[0094] S502 converts the grayscale image into a black and white image;

[0095] Use threshold processing (such as cv2.threshold() function in OpenCV) to convert the grayscale image to black and white image to enhance the contrast and facilitate subsequent processing.

[0096] S503 uses a filter to remove noise points in the image;

[0097] Use appropriate filters (such as Gaussian filters, median filters, etc.) to remove noise points in the image and improve image quality.

[0098] S504 performs tilt correction on the image to ensure that the text is level.

[0099] Use edge detection algorithms (such as Canny edge detection) to detect lines in the image and then calculate the tilt angle. Use affine transformation or perspective transformation (such as cv2.warpAffine() or cv2.warpPerspective() function in OpenCV) to correct the tilt of the image and ensure that the text is horizontal.

[0100] S402 uses OCR technology to identify boundary data in the pre-processed job image;

[0101] Use OCR (Optical Character Recognition) technology to identify text information in images, including questions, answers, etc. Use edge detection algorithms (such as Canny edge detection) to identify boundary contours in images.

[0102] S403 extracts the outline of the answer from the boundary data, and extracts the boundary of the homework paper;

[0103] Extract the outline of the answer area from the image using OpenCV's cv2.findContours() function. Determine the boundaries of the assignment paper, usually by detecting the outlines of the four corners of the paper.

[0104] S404 If the distance from the answer outline to the homework paper boundary is less than a preset value, it is considered that there is a missing value.

[0105] Calculate the distance between the answer outline and the paper border. If the distance is less than the preset value, it is considered that the part may be missing.

[0106] S304 sends the homework images with missing images to the teacher for manual review.

[0107] For the assignment images that are detected to have missing parts, the system should mark them. The system should automatically send notifications to teachers to remind them to conduct manual review. After logging into the system, teachers can view these marked images and decide whether students need to resubmit.

[0108] S103 uses image recognition technology to recognize text information in the homework image to obtain the homework data;

[0109] The specific steps include:

[0110] S601 uses support vector machine to recognize and splice text to obtain recognized text;

[0111] The specific steps include:

[0112] S701 divides the entire text area into lines of text blocks, and then divides each line of text blocks into individual characters;

[0113] Use image processing techniques (such as connected domain analysis, edge detection, etc.) to detect text areas. Use the line spacing and the blank area between words to split the entire text area into lines of text blocks. Further split each line of text blocks into individual words, which is usually achieved by analyzing the blank area between words.

[0114] S702 extracts recognition features from the segmented single characters;

[0115] Use feature extraction algorithms (such as HOG features, SIFT features, etc.) to extract key features from a single word, including shape, texture, etc. The extracted features are converted into vector form so that the support vector machine can use these features for training and prediction.

[0116] S703 uses the trained support vector machine to perform text recognition based on the recognition features;

[0117] Load the previously trained support vector machine model, which has been trained with a large number of samples and can recognize text in multiple fonts. Input the extracted features into the support vector machine model, and the model will output the most likely text category.

[0118] S704 applies an OCR algorithm to each text block to convert it into a text string;

[0119] Use OCR technology (such as Tesseract OCR) to recognize each text block and convert the text in the image into a text string. Combine the text recognized by the support vector machine with the text recognized by the OCR to improve the recognition accuracy.

[0120] S705 concatenates the recognized text strings in their original order to obtain recognized text.

[0121] According to the position information of the text blocks in the image, the recognized text strings are concatenated in the original order to form a complete recognized text, which is the preliminary form of the test data.

[0122] S602 recognizes the text and formats it according to the format required by the question to obtain the question data;

[0123] According to the requirements of the question, convert the recognized text into the corresponding format, such as converting mathematical formulas into LaTeX format, or organizing the answers in the format required by the question. Check whether the recognized text meets the requirements of the question and ensure that the format is correct.

[0124] S603 stores the test data in a database.

[0125] Store the formatted test data in the database to ensure data structure and retrievability. Create indexes for the data in the database for quick retrieval and query. Implement a data backup mechanism to ensure data security.

[0126] S104 matches the identified test data with the answers in the system test database to obtain the correction result;

[0127] The specific steps include:

[0128] S801 establishes a database containing questions and standard answers;

[0129] Collect various types of questions and their corresponding correct answers to ensure the diversity and breadth of the questions. Organize the questions and answers into a structured data format for easy computer processing. Use appropriate technology (such as relational database management system RDBMS or NoSQL database) to build a database and design a reasonable table structure to store questions and answers.

[0130] S802 extracts keywords and key phrases from the recognized text;

[0131] Preprocess the recognized text, such as removing punctuation, filtering stop words, restoring lemmas, etc. Extract keywords from the text using natural language processing (NLP) techniques, based on word frequency, TF-IDF values, or other statistical methods. Identify key phrases or expressions in the text, especially for math and physics problems containing specific terms.

[0132] Analyze the relationship between key words and key phrases, and their contextual meaning in the text to better understand student responses.

[0133] S803 matches the keywords and key phrases with the standard answers corresponding to the questions word by word, and scores them according to the degree of matching.

[0134] Compare the keywords and key phrases in the student's answer with the standard answers for the corresponding questions in the database. Compare each word to determine whether it is consistent. Calculate a match score based on the number and importance of matching words. Consider the relevance of keywords and phrases in the context and give additional points. Combine all match scores and context relevance scores to give a final match score. Set a threshold to determine whether the match score is high enough to consider the answer correct.

[0135] S105 obtains the knowledge graph of the question chapter, and analyzes the correction results based on the knowledge graph to obtain the chapter mastery.

[0136] Specific steps include collecting textbooks, reference books, online resources, etc. to obtain the information needed to build a knowledge graph; identifying knowledge points (entities) and the logical relationships between them; using appropriate tools or platforms to build a knowledge graph; obtaining the grading results of student assignments, tests, exams, etc.; for each question, identifying the types of errors made by students and mapping these errors to the associated knowledge points in the knowledge graph; and scoring students' mastery of the corresponding knowledge points based on their performance on each question.

[0137] Specifically include: collecting teaching materials, reference books, online resources, etc. to obtain the information needed to build the knowledge graph

[0138] Extract key knowledge points from the collected data, including concepts, formulas, principles, etc. Define the logical relationships between knowledge points, such as inclusion, explanation, application, etc. Construct a hierarchical structure of knowledge points to reflect the hierarchical relationships between knowledge points.

[0139] Choose a suitable tool or platform to build a knowledge graph, such as Neo4j, GraphDB and other graph databases. Take knowledge points as entities and logical relationships as edges to build a knowledge graph model. Import the extracted knowledge points and relationship data into the selected tool.

[0140] Then obtain the correction results of student assignments, tests, exams, etc., collect the correction results of student assignments, tests, exams, etc., and ensure the integrity of the data. Organize the correction results to ensure the consistency and readability of the data.

[0141] For each question, identify the types of errors made by students and map these errors to the relevant knowledge points in the knowledge graph. Based on the students' answers, identify the types of errors, such as concept confusion, calculation errors, etc. Associate the error types with the knowledge points in the knowledge graph to understand which knowledge points students have problems with.

[0142] Based on the students' scores, the students' mastery of each knowledge point is evaluated. Scoring rules are set, such as full marks for completely correct answers, lower marks for partially correct answers, and zero marks for completely wrong answers. Comprehensively consider the students' performance on different questions and give a comprehensive score for each knowledge point.

[0143] Second embodiment

[0144] The present invention also provides a homework grading system based on image recognition, comprising an identity recognition module, an image inspection module, an image recognition module, an answer matching module and a knowledge mastery assessment module; the identity recognition module is used to identify the identity of the student submitting the homework based on NFC technology; the image inspection module is used to take and upload the homework image after the student completes the homework, and check the integrity of the homework image; the image recognition module is used to use image recognition technology to recognize the text information in the homework image to obtain the question data; the answer matching module is used to match the recognized question data with the answer in the system question bank to obtain the grading result; the knowledge mastery assessment module is used to obtain the knowledge graph of the question chapter, and analyze the grading result based on the knowledge graph to obtain the chapter mastery.

[0145] In this embodiment, the identity recognition module uses NFC (near field communication) technology to authenticate the student's submitted homework. The student uses a student card with an NFC tag to approach the card reader. The card reader reads the information in the NFC tag and sends it to the system for identity authentication. The system verifies the student information and allows the student to continue submitting homework after confirming that it is correct.

[0146] After students complete their assignments, they take photos of their assignments and upload them to the system through the image check module. This module is responsible for checking whether the uploaded images are complete and clear. Students use a mobile phone application to take photos and upload images of their assignments. The system automatically detects the size, resolution, and brightness of the images to ensure image quality. It checks whether the images contain all necessary pages of the assignment and there are no missing parts. If the image is incomplete or blurry, the student is prompted to take a new photo and upload it.

[0147] The image recognition module uses OCR (Optical Character Recognition) technology to extract text information from the work image. The system receives the work image that has passed the inspection. The OCR technology is applied to recognize the text in the image and convert it into an editable text format. The recognized text is corrected to reduce OCR errors. The recognized text is saved as the test data.

[0148] The answer matching module compares the identified test data with the standard answers in the system to obtain the grading results.

[0149] The system compares the recognized test data with the standard answers stored in the question bank. For multiple-choice questions, the system directly compares whether the options are consistent. For essay questions or discussion questions, the system uses natural language processing technology to compare whether the key information points of the answers meet the requirements. The system calculates the score of each question and gives an overall evaluation. The system feeds back the correction results to the students.

[0150] The knowledge mastery assessment module evaluates students' mastery of knowledge points by analyzing the matching of answers in their assignments.

[0151] The system first collects textbooks, reference books, online resources, etc., extracts knowledge points and their logical relationships, and constructs a knowledge graph. Based on the correction results, the student's answer is mapped to the relevant knowledge point in the knowledge graph. The student's performance on each knowledge point is analyzed, the error type is identified and located to a specific knowledge point. Based on the student's performance, a mastery score is assigned to each knowledge point. Combining all the information, a detailed mastery report is generated for the student, indicating strengths and weaknesses.

[0152] The detection and experimental verification of the present invention mainly simulates actual teaching scenarios, collects student homework images, conducts system testing and data analysis to verify the effectiveness and accuracy of the technical solution. The following are specific test methods and results:

[0153] Test method:

[0154] Collect images of student homework of different grades and subjects, including mathematics, Chinese, English and other subjects.

[0155] 2) Use a camera to take photos of student assignments to ensure image clarity and quality.

[0156] 3) Upload the collected homework images to the system for text recognition and automatic correction.

[0157] 4) Compare the system’s grading results with the teacher’s manual grading results, and calculate the accuracy and consistency.

[0158] 5) Generate a data analysis report based on the marking results to evaluate students’ learning situation and knowledge mastery.

[0159] Test data range:

[0160] Number of homework images: 100-500, covering questions of different difficulty and types.

[0161] Text recognition accuracy: 95%-99%, adjusted according to image quality and system algorithm.

[0162] Automatic grading accuracy: 90%-95%, taking into account different teachers' grading standards and subjective factors.

[0163] Coverage of data analysis report: 80%-100%, including students’ homework completion status, knowledge mastery, etc.

[0164] Test results:

[0165] Text recognition: Within the test range, the system's text recognition accuracy is stable at around 97%, which can meet actual application needs.

[0166] Automatic grading: The accuracy of the system's automatic grading is over 92%, which is highly consistent with the manual grading results of teachers.

[0167] Data analysis: The data analysis report generated by the system can comprehensively reflect the students’ learning situation, with a coverage of more than 95%.

[0168] Key data disclosure: During the test, we found that certain types of questions (such as math formulas, charts, etc.) had certain errors in text recognition and automatic correction. In response to these situations, we optimized and adjusted the system to increase the recognition and correction accuracy of relevant questions to more than 90%.

[0169] Conclusion: Through the above tests and data analysis, we have verified the effectiveness and accuracy of the technical solution of the present invention in actual teaching scenarios. The system can achieve fast and accurate homework grading and data analysis, providing valuable reference information for teachers and students. At the same time, we will continue to optimize the system algorithm to improve the accuracy of recognition and grading to meet the needs of different users.

[0170] What is disclosed above is only a preferred embodiment of the present invention, and it certainly cannot be used to limit the scope of rights of the present invention. A person skilled in the art understands that all or part of the processes of the above embodiment and equivalent changes made according to the claims of the present invention still fall within the scope of the invention.

Claims

1. A homework marking method based on image recognition, characterized in that: include: Identify the identity of students submitting assignments based on NFC technology; After completing the assignment, students take and upload the assignment image, and check the integrity of the assignment image; Image recognition technology is used to identify text information in homework images to obtain test data; Match the identified test data with the answers in the system question bank to obtain the grading results; Obtain the knowledge graph of the question chapter, and analyze the correction results based on the knowledge graph to obtain the chapter mastery.

2. A homework marking method based on image recognition as claimed in claim 1, characterized in that: The specific steps of identifying the identity of the student submitting the homework based on NFC technology include: Each student is assigned an NFC tag with a built-in unique ID; Bind the NFC tag with the student’s personal information and student ID number; When students finish their homework, they put their own NFC tag on it; The NFC tag will be read and the student's name, student number and submission time will be recorded.

3. A homework marking method based on image recognition as claimed in claim 2, characterized in that: After the student completes the homework, the specific steps of taking and uploading the homework image and checking the integrity of the homework image include: Use a camera to capture images of the work; Check whether the clarity of the uploaded job image exceeds the preset value. If not, it is considered unqualified; Use OCR technology to identify the boundaries in the job image detected by resolution and check whether there is any missing image; The assignment images with missing images are sent to the teacher for manual review.

4. A homework marking method based on image recognition as claimed in claim 3, characterized in that: The specific steps of using the OCR technology to identify the boundary in the job image detected by resolution and checking whether there is a missing image include: Preprocess the job image; Use OCR technology to identify boundary data in pre-processed job images; Extract the answer outline from the boundary data and extract the homework paper boundary; If the distance from the answer outline to the border of the assignment paper is less than the preset value, it is considered that there is a missing value.

5. A homework marking method based on image recognition as claimed in claim 4, characterized in that: The specific steps of preprocessing the job image include: Convert color images to grayscale images; Convert grayscale images to black and white images; Use filters to remove noise points in the image; Deskew the image to ensure the text is level.

6. A homework marking method based on image recognition as claimed in claim 5, characterized in that: The specific steps of using image recognition technology to recognize text information in the homework image and obtain the homework data include: Use support vector machine to recognize and splice the text to obtain the recognized text; The recognized text is formatted according to the format required by the question to obtain the test data; Store the test data in the database.

7. A homework marking method based on image recognition as claimed in claim 6, characterized in that: The specific steps of using a support vector machine to recognize and splice text to obtain recognized text include: Split the entire text area into lines of text blocks, and then split each line of text blocks into individual words; Extract recognition features from the segmented individual characters; Use the trained support vector machine to perform text recognition based on recognition features; Apply OCR algorithm to each text block to convert it into a text string; The recognized text strings are concatenated in their original order to obtain the recognized text.

8. A homework marking method based on image recognition as claimed in claim 7, characterized in that: The specific steps of matching the identified test data with the answers in the system test database to obtain the correction results include: Establish a database containing questions and model answers; Extract keywords and key phrases from the recognized text; Match the keywords and key phrases with the standard answers corresponding to the questions word by word and give scores based on the degree of match.

9. A homework grading system based on image recognition, characterized in that: It includes identity recognition module, image inspection module, image recognition module, answer matching module and knowledge mastery assessment module; The identity recognition module is used to identify the identity of the student submitting the homework based on NFC technology; The image checking module is used for students to take and upload homework images after completing their homework, and to check the integrity of the homework images; The image recognition module is used to recognize text information in the homework image using image recognition technology to obtain the homework data; The answer matching module is used to match the identified test data with the answers in the system test database to obtain the correction results; The knowledge mastery evaluation module is used to obtain the knowledge graph of the question chapter, and analyze the correction results based on the knowledge graph to obtain the chapter mastery.

Images