Method for synchronously correcting questions on paper screen

Abstract

The invention discloses a method for synchronously correcting questions on a paper screen, and aims to solve the efficiency bottleneck of manually inputting scores after a teacher corrects test papers in campus teaching and assist intelligent transformation of teaching. The key of the method lies in organic cooperation of an application program, a Bluetooth pen, dot matrix code paper and an OCR service. The application program is responsible for accurately loading test paper details and student score information, and creating a convenient correction environment by means of intelligent caching and handwriting rendering means; the system establishes stable Bluetooth connection with a Bluetooth pen with a built-in advanced module, receives dot matrix data written on dot matrix code paper by the pen in real time, and transmits key data to an OCR service through rigorous standardized process processing. The OCR service uses frontier technologies such as a ResNet50 model to identify written content, an application program accurately scores questions according to an identification result and a preset scientific rule, correction and score input synchronization is achieved, the questions are automatically uploaded to a cloud end, the whole process is efficient and accurate, and teaching is powerfully promoted to be intelligent.

Description

Technical Field

[0001] The present invention relates to the technical field of education and teaching, and in particular to a method for synchronously correcting questions on paper and screen. Background Art

[0002] In current campus teaching practices, when teachers summarize grades after completing the marking of test papers, they usually follow a cumbersome and inefficient process. They have to manually enter the grades of each test paper one by one into spreadsheet software such as Excel. This process not only consumes a large amount of time and energy but is also extremely prone to data entry errors due to human negligence. Even when using tools such as Excel for grade summarization and preliminary analysis, the efficiency improvement is very limited because the key bottleneck in the entire process - the manual entry of grades - has not been fundamentally solved. This traditional working method is out of place in today's context of pursuing efficient and intelligent teaching management, seriously hindering the in-depth development process of teaching informatization and intelligence, making teachers linger in repetitive low-value-added labor and unable to invest more time and energy in more valuable teaching research and student tutoring work. Therefore, there is an urgent need for an innovative solution to break this dilemma and achieve the optimization and upgrading of the teaching process. Summary of the Invention

[0003] The purpose of the present invention is to provide a method for synchronously correcting questions on paper and screen, to solve the disconnection problem between the processes of teachers' marking and entering grades. By combining the two steps of marking and entering into one step, and with the assistance of OCR (Optical Character Recognition) technology to identify the marking writing content, the marking results are directly entered into the cloud database, realizing the visualization result of information data that can be seen immediately after marking, thereby greatly improving the work efficiency of teachers and achieving the goal of teaching intelligence.

[0004] To achieve the above purpose, the present invention is realized through the following technical solutions: A method for synchronously correcting questions on paper and screen, comprising the following steps:

[0005] Including an application program, a Bluetooth pen, a dot matrix code paper, and an OCR service. The application program is used to process test paper and grade-related data. The Bluetooth pen is used to write on the dot matrix code paper and read coordinate data. The dot matrix code paper is used to provide a writing carrier with code points. The OCR service is used to identify the writing content.

[0006] As a further improvement to the technical solution of the present invention, the test paper information loading module of the application program can load test paper files in multiple formats, including but not limited to scanned test paper images in PDF, JPEG, and PNG formats, as well as text-format test paper question information. During the loading process, format conversion and compatibility processing are automatically performed to ensure that the test paper information enters the application program memory completely and accurately.

[0007] As a further improvement to the technical solution of the present invention, the score information loading module of the application program supports importing score data from multiple data sources, such as local databases, Excel spreadsheets, school academic affairs system interfaces, etc. During the loading process, the data is cleaned and verified to eliminate invalid or incorrect data records. At the same time, the score data is encrypted and stored to ensure data security.

[0008] As a further improvement to the technical solution of the present invention, the cache mechanism of the test paper data preprocessing and handwriting rendering module in the application program has an intelligent update function. When a newer version of the high-definition PDF remote resource of the test paper is detected, it automatically downloads and replaces the original cache. During the caching process, data compression technology is used to reduce storage space occupancy, and the cached data is regularly backed up to prevent data loss.

[0009] As a further improvement to the technical solution of the present invention, the historical dot matrix data processing unit of the test paper data preprocessing and handwriting rendering module in the application program can, when rendering the written handwriting, perform personalized rendering of the historical handwriting according to the display preferences set by the user, such as handwriting color, transparency, thickness, etc. At the same time, it supports operations such as local magnification, reduction, and erasure of the historical handwriting, facilitating users to view and edit.

[0010] As a further improvement to the technical solution of the present invention, in addition to having the basic functions of reading and transmitting coordinate data, the Bluetooth pen is also equipped with a battery monitoring module. When the battery level is lower than the preset threshold, it automatically sends a low battery alert to the application program and supports fast charging technology, enabling it to recover sufficient power in a short time to ensure continuous use.

[0011] As a further improvement to the technical solution of the present invention, the dot matrix code points of the dot matrix code paper are made of fluorescent materials or special reflective coatings, enabling them to be clearly and accurately read by the Bluetooth pen under different lighting conditions. And during the printing process, it undergoes special surface treatment to enhance the wear resistance and stain resistance of the paper, extending the service life of the dot matrix code paper.

[0012] As a further improvement of the technical solution of the present invention, when the real-time rendering handwriting module in the application program receives the dot matrix data transmitted by the Bluetooth pen, it has a data flow control function, and can automatically adjust the data reception frequency and rate according to the network condition and device performance, avoiding data congestion and lag. At the same time, during the process of rendering handwriting, the hardware acceleration technology is utilized to improve the speed and smoothness of handwriting rendering.

[0013] As a further improvement of the technical solution of the present invention, when the dot matrix data standardization and matching module in the application program judges the handwriting landing area, it combines artificial intelligence image recognition technology to intelligently identify and locate the question area. Even if the question layout or format changes, it can accurately judge the question where the writing is located, and uses blockchain technology to encrypt and store and verify key information such as question IDs to ensure the authenticity and immutability of the data.

[0014] As a further improvement of the technical solution of the present invention, when the OCR recognition related module in the application program sends an OCR recognition request, it uses data encryption and digital signature technologies to ensure the security and integrity of data transmission. At the same time, when processing the OCR recognition result, it can be seamlessly docked with the school's grade analysis system, directly upload the marking result and grade data to the analysis system, realize the real-time statistics and analysis of grade data, and generate a visual grade report for teachers and students to view.

[0015] A method for synchronously marking questions on paper and screen according to the present invention mainly includes the following parts:

[0016] 1. Application program

[0017] (1) Loading test paper and grade information

[0018] Through a specific data interface, the test paper name, detailed test paper question information, clear test paper scan map and other contents are accurately and efficiently loaded into the memory space of the application program. At the same time, a secure and reliable data transmission protocol and verification mechanism are used to accurately import the current student's grade structure and initial results into the application program memory. For the loaded data, advanced data preprocessing algorithms such as data cleaning and format conversion are used to make it reach the best processing state and be ready to enter the subsequent process links at any time.

[0019] (2) Test paper data preprocessing and handwriting rendering

[0020] Using intelligent data integration technology, the test paper information and score information are deeply integrated and optimized to build a complete and intuitive test paper presentation model. When processing high-definition PDF resources of test papers, its caching mechanism demonstrates high intelligence and efficiency. It can quickly judge the caching status of resources. If not cached, through multi-threaded parallel download technology, combined with an efficient data compression algorithm, it quickly obtains and caches the resources and then loads them; if already cached, it uses data integrity verification technology to directly load from the local high-speed cache. After loading is completed, through advanced image rendering technology, combined with historical dot matrix data, if there is dot matrix data of the writing history, it can, based on accurate coordinate positioning and image restoration algorithms, render the writing strokes at the corresponding coordinate positions on the high-definition PDF background of the test paper with a highly realistic effect, ensuring that users can clearly and accurately view the entire test paper and marking traces.

[0021] (3) Bluetooth communication connection

[0022] Relying on a stable and low-power Bluetooth communication protocol, a two-way, real-time and reliable communication link is established with the supporting Bluetooth pen, ensuring the stability and efficiency of data transmission and laying a solid foundation for subsequent real-time data interaction.

[0023] (4) Real-time rendering of writing strokes

[0024] Receiving dot matrix data: Through an optimized Bluetooth data receiving module, using asynchronous reception and data buffering technology, the dot matrix data generated by the tip of the Bluetooth pen writing on the code paper is received in real time, stably and without omission, and stored in a carefully designed buffer area to effectively prevent data loss and overflow. At the same time, data verification algorithms are used to ensure the accuracy of the data.

[0025] Rendering writing strokes: Based on advanced graphics and image processing algorithms, according to the dot matrix coordinate data on the real code paper, combined with parameters such as the resolution, pixel density and display ratio of the device screen, using precise coordinate conversion and graphics drawing technology, the writing strokes are rendered completely, smoothly and clearly on the layer, realizing the accurate display of the writing strokes on the screen and providing users with a natural experience as if writing on paper.

[0026] (5) Standardization and matching of dot matrix data

[0027] Judging the landing area of the writing strokes: Using intelligent image recognition and analysis technology, combined with predefined templates and machine learning algorithms for the question areas, when the first stroke landing point of the writing or 80% of the dot matrix data of the total writing length is within the area circled by the question, it can quickly and accurately determine the question where the writing is located, obtain key information such as the question id through an efficient indexing algorithm, and at the same time use data marking technology to accurately mark the content of the writing strokes for subsequent data processing to provide a clear direction.

[0028] Normalize dot matrix data: By comprehensively applying advanced image processing techniques and mathematical models, considering various factors such as pen-lifting actions, pen-dropping actions, the number of writing strokes, writing intervals, and writing areas, through complex coordinate transformation and data normalization algorithms, all coordinate information of the writing content for this time is obtained, and it is proportionally converted into standard coordinate data with a resolution of 80x80. At the same time, efficient data compression and encryption technologies are adopted to optimize the processed data, reduce the data transmission volume, and improve data security to ensure the efficient utilization of data in subsequent processes.

[0029] (6) Send an OCR recognition request

[0030] Send the normalized coordinate data and the encrypted relevant core parameters to the OCR data interface for recognition through a secure, stable, and efficient network transmission protocol. The recognition process includes:

[0031] Preprocessing: Based on the coordinate data, use advanced vector graphics generation algorithms to generate high-quality vector images, and then through image resolution adjustment and grayscale processing techniques, convert them into grayscale images with a resolution of 80x80 pixels, providing high-quality image data basis for subsequent vector conversion.

[0032] Vector conversion: Use the ResNet50 model in the field of deep learning, combined with high-performance computing hardware resources, to perform vectorization processing on the preprocessed images, and convert the image information into a vector data form that can be efficiently processed by the computer for subsequent vector retrieval operations.

[0033] Vector retrieval: Based on a large-scale vector database and efficient retrieval algorithms, quickly and accurately retrieve similar vectors in the database according to the obtained image vectors, and obtain a set of similar vector data with a length of 100, providing rich data resources for subsequent screening algorithms.

[0034] Screening algorithm: Perform a fine filtering operation on the above data set. Using an intelligent character stroke analysis algorithm, accurately remove the data in the data set where the marked character stroke count is inconsistent with the written stroke count to be recognized. On this basis, perform a distribution statistics on the data set through a complex statistical analysis algorithm. If the proportion of the character with the highest proportion is greater than 10%, then further use an intelligent screening strategy for judgment. If there is only one type of character in the result set, then use this character as a highly credible recognition result; if there are multiple types of characters in the result set, when the number of written strokes is less than or equal to 2, according to statistical principles, use the character with the largest proportion in the result set as a reasonable recognition result; when the number of written strokes is greater than 2, use a complex character matching and combination algorithm to perform single-character matching and combination on the strokes of the written content, and through precise comparison and intelligent reasoning with a predefined character library, obtain the most accurate recognition result to ensure the accuracy and reliability of OCR recognition.

[0035] (7) Process the OCR recognition result

[0036] According to the OCR recognition result, use an intelligent score evaluation algorithm to score the questions, specifically as follows:

[0037] When the user writes √, based on the preset scoring rules, the corresponding question is directly judged as full marks; when the user writes ×, the corresponding question is judged as 0 points. This simple and clear symbol scoring method can quickly and accurately handle common marking situations.

[0038] For numerical scoring, the system supports the recognition of positive integers and half points (1 / 2) within 0 - 150. Through precise character recognition and numerical conversion algorithms, accurately convert the recognized characters into corresponding scores, and score the questions accordingly to meet the diverse score evaluation requirements.

[0039] At the same time, the system also supports the recognition of negative integers and half points (1 / 2) within 0 - 60 to adapt to special deduction situation scoring. Similarly, through rigorous numerical conversion and logical judgment, accurate score evaluation is achieved.

[0040] When the input content written by the user does not meet the above conditions, the system uses natural language processing technology and an intelligent error diagnosis algorithm to deeply analyze and classify the unrecognized content, record relevant information for subsequent manual review and processing, and at the same time feedback clear, friendly and guiding prompt information to the user to help the user understand the reason for the recognition failure and possible solutions.

[0041] 2. Bluetooth pen

[0042] Using advanced microelectromechanical system (MEMS) technology and high-performance Bluetooth chipset, a ballpoint pen with a built-in Bluetooth data processing module and an intelligent operation logic processor is created. During the writing process, its high-precision sensors can read the coordinate data on the dot matrix paper in real time with extremely high frequency and accuracy, and through an optimized Bluetooth communication protocol and data transmission algorithm, transmit the data stably, quickly and accurately to the application program. At the same time, it has low-power standby and fast wake-up functions to extend the battery life and improve the usability.

[0043] 3. Dot matrix paper

[0044] Using professional coding software and a unique dot matrix generation algorithm, a special paper is made, which is printed by a high-precision printer after being carefully designed and specially rendered by a software program. The dot matrix code points on its surface are evenly distributed, with high precision and good stability and readability, and can be quickly and accurately identified and read by the Bluetooth pen, providing a reliable physical basis for the acquisition of coordinate data. At the same time, in the selection of paper materials and surface treatment, materials with good writing performance and durability are used to ensure that the paper is not easily damaged and deformed during the writing process, guaranteeing the stability and reliability of data acquisition.

[0045] 4. OCR service

[0046] Build a powerful OCR service system based on the cloud computing platform, provide an efficient, accurate and stable network interface service for OCR recognition, and through an optimized network architecture and data transmission protocol, seamlessly dock with the application program of the present invention, receive the standardized coordinate data sent by the application program for recognition, and use advanced deep learning algorithms and a large-scale character sample library to continuously improve the recognition accuracy and generalization ability, and return the recognition result to the application program quickly and reliably, providing core character recognition support for the entire paper-screen synchronous marking method and promoting the efficient operation of the entire system.

[0047] The present invention has the following beneficial effects:

[0048] 1. Significantly improve work efficiency:

[0049] In the traditional scenario of teachers grading test papers, it is usually necessary to sequentially complete grading N test papers, manually inputting the scores of N test papers into computer tools (such as Excel, WPS Spreadsheets, etc.), and the cumbersome data aggregation work. The entire process is cumbersome and time-consuming. By using the application program of the present invention, through innovatively merging the two key steps of grading test papers and inputting scores into one synchronous step, with the help of efficient data transmission and processing technologies, the grading results are uploaded to the cloud database in real time. Utilizing the powerful computing power of the cloud and the automated data aggregation function, the relevant data aggregation work can be completed in an extremely short time. In the same working environment and task volume, theoretically, the present invention can significantly reduce the originally cumbersome process steps from 2N to N to complete. Through actual testing and theoretical analysis, the work efficiency can be effectively increased by 50% or reach a higher improvement ratio according to the actual situation, greatly reducing the workload of teachers and enabling teachers to invest more time and energy in more valuable work such as teaching research and student tutoring.

[0050] 2. Strongly promote the development of teaching intelligence:

[0051] By innovatively realizing paper-screen synchronous grading and the application of high-precision OCR technology, the present invention successfully breaks the gap between traditional teaching grading and score input, achieving automated and intelligent input of grading results and instant visualization of data. While grading test papers, teachers can simultaneously see the input and statistics of scores in real time, facilitating immediate evaluation and analysis of teaching effects and providing strong data support for subsequent adjustment of teaching strategies. This intelligent grading and data processing method, deeply integrated with modern information technology, helps to comprehensively promote the process of electrification and intelligence in teaching, making teaching management more efficient, accurate, and scientific, and injecting new vitality into the innovative development of the education field. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Other features, objectives, and advantages of the present invention will become more apparent by reading the detailed description of the non-limiting embodiments with reference to the following drawings:

[0053] Figure 1 is a flowchart framework of a method for paper-screen synchronous grading of questions in an embodiment of the present invention;

[0054] Figure 2 is Figure 1 one of the partial enlarged schematic diagrams of

[0055] Figure 3 is Figure 1 one of the partial enlarged schematic diagrams of

[0056] Figure 4 is Figure 1 one of the partial enlarged schematic diagrams of

[0057] Figure 5 is Figure 1 The fourth partial enlarged schematic diagram. Specific implementation manners

[0058] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. Here, the schematic embodiments and descriptions of the present invention are used to explain the present invention, but do not limit the present invention.

[0059] It should be noted that all directional indications (such as up, down, left, right, front, back, upper end, lower end, top, bottom...) in the embodiments of the present invention are only used to explain the relative positional relationship and movement conditions between components in a specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indications will also change accordingly.

[0060] In the present invention, unless otherwise clearly defined and limited, the term "connection" should be understood in a broad sense. For example, "connection" can be a fixed connection, a detachable connection, or integrated; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and can be the communication inside two components or the interaction relationship between two components, unless otherwise clearly limited. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific situations.

[0061] In addition, in the present invention, descriptions such as "first" and "second" are only for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the quantity of the indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the technical solutions between various embodiments can be combined with each other, but it must be based on the fact that those of ordinary skill in the art can implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the protection scope required by the present invention.

[0062] The present invention will be further described in detail below with reference to the accompanying drawings.

[0063] Refer to Figure 1 and Figures 2 to 5 , the present invention carefully constructs a method for synchronously correcting questions on a paper screen, including the following steps:

[0064] I. Application program operations

[0065] (1) Loading of test papers and scores

[0066] When the correction system of the present invention is started, the application will automatically start the test paper and score loading program. First, it will send an encrypted request to the school's test paper resource library, and obtain the test paper information of the corresponding course, including the test paper name, detailed question content, high-definition scan map, etc. through a secure network transmission protocol, and use advanced data decompression and parsing algorithms to accurately load this information into memory. At the same time, preliminary data verification and format conversion are carried out to ensure the integrity and availability of the data. For the score information, the application will interface with the school's score management system. After authentication and authorization, incremental data synchronization technology is used to load the score structure and initial score data of the current students into memory, and establish an associated index with the test paper information to facilitate data update and query during the subsequent correction process.

[0067] (2) Data integration and rendering

[0068] The data integration module in the application will deeply integrate the loaded test paper and score data, and build an intuitive and clear test paper correction interface model through intelligent layout algorithms. When processing the high-definition PDF resources of the test paper, its cache management system will first query the local cache database. If the corresponding cache file is not found, it will immediately start a multi-threaded download task, and at the same time perform real-time integrity verification and data chunk storage on the downloaded resources to improve the read and write speed of the cache. If there is a cache, it will be directly read and the test paper content and writing strokes will be accurately rendered on the screen through a high-speed graphics rendering engine, combined with possible historical dot matrix data, providing a familiar and convenient correction environment for teachers.

[0069] (3) Bluetooth connection and data processing

[0070] When the application is started, it will automatically search for nearby available Bluetooth devices, and establish a stable connection with the Bluetooth pen through a secure pairing protocol according to the preset Bluetooth pen device identifier. After the connection is established, the application will start a dedicated Bluetooth data listening thread, and use efficient data reception and parsing algorithms to receive the dot matrix data transmitted by the Bluetooth pen in real time. Once the data is received, it will immediately be passed to the subsequent data processing module for data preparation work related to handwriting rendering and OCR recognition to ensure the real-time and smoothness of the entire correction process.

[0071] II. Bluetooth pen usage

[0072] When using the Bluetooth pen, the teacher only needs to correct and write on the dot matrix code paper just like using an ordinary ballpoint pen. The high-precision pressure sensor and optical sensor inside the Bluetooth pen work together to accurately collect the coordinate data of the pen tip on the code paper and the writing pressure information hundreds of times per second. These data will be quickly and stably transmitted to the application in the form of data packets through the low-power Bluetooth chip built into the Bluetooth pen, using an optimized Bluetooth communication protocol. At the same time, the Bluetooth pen also has an intelligent power monitoring function. When the battery level is lower than 20%, it will remind the teacher to charge in time by vibrating and flashing the indicator light, and supports fast charging through a common USB Type-C interface, and can recover enough power in a short time to ensure the continuity of the correction work.

[0073] III. Preparation of Dot Matrix Code Paper

[0074] The school's teaching management department can use specialized dot matrix code paper generation software to typeset and render the content of the test papers to be corrected with specific dot matrix patterns. These dot matrix patterns are generated through complex coding algorithms, and each code point has a unique position identifier and feature information, which can be accurately recognized by the Bluetooth pen. When printing the dot matrix code paper, use a high-precision laser printer or inkjet printer to ensure the clarity and integrity of the dot matrix code points. After printing, distribute the test papers to students for the exam according to the normal teaching process. After the exam, the teacher can use the system of the present invention for correction.

[0075] IV. OCR Service Processing

[0076] The OCR service runs on a high-performance cloud computing platform, adopting a distributed computing architecture and a large-scale GPU cluster to improve the speed and accuracy of recognition. After the application sends the standardized coordinate data and related parameters to the OCR service, the service first preprocesses the data, including using advanced image enhancement algorithms to improve the clarity and contrast of the image, then performs feature extraction and vectorization processing through the ResNet50 model based on convolutional neural networks, and then uses a large-scale vector database for similar vector retrieval, and performs character recognition and judgment through complex screening algorithms and deep learning models. Finally, the recognition results are quickly returned to the application in the form of standardized JSON format. The application calculates and updates the scores based on these results, and uploads the final score data to the cloud database to achieve centralized management and sharing of data.

[0077] V. Implementation Case

[0078] In the marking of a monthly math exam paper in a certain middle school, a total of 45 students took the exam, and the exam paper included 23 types of various questions. According to the traditional way of marking exam papers, from marking the papers to manually entering the scores into an Excel spreadsheet, and completing the work of statistics on the "three rates and one average" (excellent rate, passing rate, low score rate, and average score, the same below) and error question statistics, the whole process takes about 6 hours, and errors are likely to occur during the manual entry process. After using the paper-screen synchronous marking method of the present invention, during the process of marking exam papers, every time a question is marked, the system automatically identifies the marking result through a Bluetooth pen and OCR technology, and the marked data is automatically uploaded to the cloud database. After the marking is completed, comprehensive data statistics and analysis work including the statistics on the "three rates and one average" and error question statistics can be generated. The whole process takes about 2 hours. Just comparing the work content of only marking the papers and completing the statistics on the "three rates and one average" and error question statistics, the work efficiency has increased by 2 times, and the accuracy of the data is fully guaranteed. At the same time, teachers can view the score distribution and the analysis of students' answering situations at any time through the application program, providing strong data support for subsequent teaching explanations and tutoring, and significantly improving the teaching effect and quality. In subsequent multiple exam markings, this method has shown stable and efficient performance, received unanimous praise and recognition from teachers, and strongly promoted the intelligent development process of the school's teaching work.

[0079] In summary, using the paper-screen synchronous marking method of the present invention has many significant advantages:

[0080] Efficient and time-saving: Taking the marking of math exam papers in a certain middle school as an example, for the same work content comparison, the traditional marking method takes about 6 hours to process the exam paper data of 45 students, while using this method takes about 2 hours, and the work efficiency has increased by 2 times, freeing teachers from the cumbersome data entry work and allowing them to invest time in more valuable teaching activities.

[0081] Accurate and error-free: Abandoning the error-prone manual entry link, using a Bluetooth pen and OCR technology to automatically identify the marking result and enter the score, greatly reducing the risk of human error, ensuring the high accuracy of the score data, and providing a reliable basis for teaching evaluation and student assessment.

[0082] Instant data support: The marked data is automatically uploaded to the cloud for processing. Teachers can view the marking situation, score distribution, and analysis of students' answering situations in real time through the application program, timely grasp the teaching effectiveness, provide strong data guidance for subsequent targeted teaching explanations and tutoring, and optimize teaching strategies.

[0083] Promote teaching intelligence: It runs stably and efficiently in multiple marking practices, is widely recognized by teachers, strongly promotes the school's teaching work towards intelligence, improves the overall teaching quality and management level, meets the needs of modern education development, and helps educational innovation and reform.

[0084] The technical solutions provided by the embodiments of the present invention have been introduced in detail above. Specific examples are used herein to elaborate on the principles and implementation manners of the embodiments of the present invention. The descriptions of the above embodiments are only applicable to helping understand the principles of the embodiments of the present invention. At the same time, for those of ordinary skill in the art, according to the embodiments of the present invention, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation to the present invention.

Claims

1. A method for grading questions on paper and screen simultaneously, characterized in that: The following steps are involved: Step S1, application operation: Step S11, loading test paper and score information: loading the test paper name, test paper title information, and test paper scan image into the application memory, and loading the current student's score structure and results into the application memory; Step S12, test paper data preprocessing and handwriting rendering: perform data integration on the loaded test paper information and score information. If the remote resource of the test paper high-definition PDF is not cached, it is downloaded and cached before loading. If it is cached, it is directly displayed using the cache. If there is dot matrix data of the writing history, the writing handwriting is rendered at the corresponding coordinate position based on the test paper high-definition PDF base map. Step S13, Bluetooth communication connection: bidirectional communication with the matching Bluetooth pen through the Bluetooth communication interface; real-time rendering of handwriting: Step S14, receiving dot matrix data: receiving the dot matrix data written by the tip of the Bluetooth pen on the code paper in real time; Step S15, rendering the handwriting: after calculating the relative position according to the dot matrix coordinate data on the real code paper and the screen size ratio of the mobile device, the handwriting is completely rendered on the layer; Step S2: standardization and matching of dot matrix data: Step S21, determine the handwriting landing area: if the first stroke of the writing or 80% of the dot matrix data of the total length of the writing is within the question circled area, determine the question where the writing is located, and obtain the key information of the question ID to mark the handwriting content of the writing; Step S22, standardize the dot matrix data: comprehensively analyze the pen lifting action, pen dropping action, number of strokes, writing interval and writing area to obtain all coordinate information of the writing content, and convert it geometrically into coordinate data with a resolution of 80x80; Step S3, sending an OCR recognition request: sending the standardized coordinate data and related core parameters to the OCR data interface for recognition, the recognition includes: Step S31, pre-processing: generating a vector image according to the coordinate data, converting the resolution into an 80x80 pixel image and performing grayscale processing; Step S32, vector conversion: use the ResNet50 model to vectorize the preprocessed image; Step S33, vector retrieval: retrieve similar vectors in the vector database according to the obtained picture vector, and obtain a similar vector data set with a length of 100; Step S34, screening algorithm: filter the data set, remove the data whose number of strokes of the marked character is inconsistent with the number of strokes of the writing requested for recognition, perform distribution statistics on the filtered data set, and if the character with the highest proportion is greater than 10%, further screen, if the result set has and only has one character, use this character as the recognition result, if the result set has multiple characters, when the number of writing strokes is less than or equal to 2, use the character with the largest proportion in the result set as the recognition result, and when the number of writing strokes is greater than 2, perform single-word matching on the strokes of the writing content and then combine them to return the result; Step S4, processing the OCR recognition result: scoring the question according to the OCR recognition result, wherein: when the user writes √, the corresponding question gets full marks, and when the user writes ×, the corresponding question gets 0 points; it supports recognition of positive integers within 0-150 and half-point (1 / 2) mark characters, and scores the question accordingly; Supports the recognition of negative integers within 0-60 and half-cent (1 / 2) mark characters, and scores the questions accordingly; When the input content written by the user does not meet the above conditions, an unrecognizable result is returned; Step S5, Bluetooth pen operation: using a ballpoint pen with a built-in Bluetooth data processing module and a related operation logic processor as a Bluetooth pen, reading the coordinate data on the dot matrix code paper when writing and transmitting it to the application via Bluetooth; Step S6, dot matrix code paper preparation: use a paper specially rendered by a software program and printed as dot matrix code paper, with dot matrix code points printed on its surface for the Bluetooth pen to read coordinate data; Step S7, OCR service operation: using a network interface service that provides OCR recognition as an OCR service, receiving the standardized coordinate data sent by the application for recognition, and returning the recognition result to the application.

2. A method for correcting questions on paper and screen simultaneously according to claim 1, characterized in that: The application updates the relevant status of the grade structure during the user's grading process and uploads the grade data to the cloud database at a specific time.

3. The method for grading questions on paper and screen simultaneously according to claim 1, characterized in that: The dot matrix data is transmitted between the Bluetooth pen and the application program via Bluetooth communication.

4. The method for grading questions on paper and screen simultaneously according to claim 1, characterized in that: The dot matrix code points of the dot matrix code paper are read by a Bluetooth pen.

5. The method for grading questions on paper and screen simultaneously according to claim 1, characterized in that: The recognition algorithm of the OCR service recognizes the written content and gives a score judgment or an unrecognizable result based on the written content.

6. A method for correcting questions on paper and screen simultaneously according to claim 1, characterized in that: In the scenario where teachers mark test papers, the application combines marking test papers and entering grades into one step, and uploads grade data to the cloud database for data aggregation.

Images