WO2018006294A1 - 一种基于图像模式识别技术的阅卷系统、装置和方法 - Google Patents

一种基于图像模式识别技术的阅卷系统、装置和方法 Download PDF

Info

Publication number
WO2018006294A1
WO2018006294A1 PCT/CN2016/088814 CN2016088814W WO2018006294A1 WO 2018006294 A1 WO2018006294 A1 WO 2018006294A1 CN 2016088814 W CN2016088814 W CN 2016088814W WO 2018006294 A1 WO2018006294 A1 WO 2018006294A1
Authority
WO
WIPO (PCT)
Prior art keywords
scoring
image
result
unit
module
Prior art date
Application number
PCT/CN2016/088814
Other languages
English (en)
French (fr)
Inventor
王楚
Original Assignee
王楚
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 王楚 filed Critical 王楚
Priority to PCT/CN2016/088814 priority Critical patent/WO2018006294A1/zh
Publication of WO2018006294A1 publication Critical patent/WO2018006294A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16Z—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS, NOT OTHERWISE PROVIDED FOR
    • G16Z99/00—Subject matter not provided for in other main groups of this subclass

Definitions

  • the present invention relates to the field of image processing and pattern recognition, and in particular to a marking system, apparatus and method based on image pattern recognition technology.
  • the examination is the most common and common means to assess the mastery of students' knowledge. It is beneficial to the promotion of teachers' teaching work, the improvement of students' academic performance and the improvement of the whole education level through examinations of various scales. Objective questions can detect students' ability to judge, while subjective questions can reflect students' logical reasoning and ability to summarize and summarize. Therefore, a test paper containing both objective and subjective questions has considerable test value.
  • the subjective questions are evaluated by different scoring teachers on the computer on the computer to evaluate the electronic images of the candidates' answers, and finally by the computer system.
  • a scoring method that automatically performs verification and score verification. Online scoring has the following advantages: It uses a multi-evaluation and error control mechanism to help the scoring teachers better grasp the scoring standards, control errors, and improve the quality of scoring; but there are also scoring environments, scanning settings, and answer sheets. High quality requirements, in addition to the huge cost of marking also makes it difficult to promote the application in ordinary exams.
  • the examinations in the basic education stage in China are characterized by frequentness. In order to detect students’ knowledge in a timely manner The mastery of the point to promote the improvement of students' knowledge level, in addition to the regular mid-term and final exams, there are also increased monthly exams, small tests and other scales of the exam. Corresponding to frequent exams, exams in the basic education stage urgently require fast, accurate, and low-cost scoring. Teachers review papers in a relatively short period of time. The time is tight and the workload is so large that manual errors are inevitable in all links, and the work of scores, points, and saves requires labor, material resources, and venue resources. Therefore, in order to meet the needs of China's basic education stage examinations, the emergence of an automated scanning and scoring system is particularly important.
  • the object of the present invention is to provide a scoring device based on image pattern recognition technology, which effectively solves the problems of traditional manual scoring time, heavy task, large workload of finding paper test papers, difficulty in registering scores and verification scores, time-consuming and labor-intensive, etc.
  • the problem is to improve the quality of the teachers' reading, reduce the labor burden of the teachers, effectively reduce the errors in the manual processing and answering and turnover, save time and effort, and archive the test papers to provide guarantee for improving the quality of the marking.
  • the technical solution adopted by the invention is to design an automatic identification information scoring device including a subjective question and an objective question answering book, which not only enables the teacher to conveniently and quickly produce the answer sheet, but also efficiently recognizes the relevant information such as the answer score.
  • the invention provides a scoring device based on an image pattern recognition technology, comprising: an image recognition module for identifying a scan result image of a scanner by a pattern recognition technology; and a scoring service processing module for executing a scribing logic process to generate a scoring file a result; and a request determination module for determining whether the "question request" initiated by the user is legal.
  • the image recognition module includes: an image pre-processing unit configured to perform pre-processing on the scan result image; and a template detecting unit configured to perform template detection on the image based on the pre-processed image; And an identification unit configured to identify the image by the pattern recognition technology and store the recognition result.
  • the scoring service processing module includes: a reading unit configured to read the identification result; a query unit configured to query a standard answer; and a matching unit configured to use the standard answer and the identification node And performing a matching; and a statistical unit for performing score statistics to generate the scoring result.
  • the present invention also provides a scoring system based on image pattern recognition technology, comprising: a scoring terminal, comprising a scanner and a scoring device, wherein the scanner is configured to scan a candidate test paper that has undergone subjective review, the scoring The device is connected to the scanner for identifying a scan result image of the scanner by a pattern recognition technology, performing a scoring logic process, and generating a scoring result; a website back end connected to the scoring terminal for generating a test paper, receiving And storing the result of the marking generated by the marking terminal to generate an evaluation report; and a client connected to the back end of the website for receiving an examination result message and viewing the evaluation report.
  • a scoring terminal comprising a scanner and a scoring device, wherein the scanner is configured to scan a candidate test paper that has undergone subjective review, the scoring The device is connected to the scanner for identifying a scan result image of the scanner by a pattern recognition technology, performing a scoring logic process, and generating a scoring result
  • a website back end connected to the scoring terminal for generating
  • the scoring device comprises: an image recognition module for identifying the scan result image of the scanner by the pattern recognition technology; a scoring service processing module, configured to execute the scribing logic process, generate the a result of the scoring; and a request determination module for determining whether the "question request" initiated by the user is legal.
  • the image recognition module comprises: an image pre-processing unit for performing pre-processing on the scan result image by the pattern recognition technology; a template detection unit, configured to: based on the pre-processed image, The image is subjected to template detection; and an identification unit for identifying the image and storing the recognition result.
  • the scoring service processing module comprises: a reading unit for reading the recognition result; a query unit for querying a standard answer; and a matching unit configured to match the standard answer with the recognition result And a statistical unit for performing score statistics to generate the scoring result.
  • the website back end comprises: a group roll module for generating a test paper and an answer sheet, defining basic information, structure information, and answer and scoring standards of the test paper; and a student information module for storing student and teacher user information; And an evaluation module, configured to receive and save the scoring result of the scoring terminal, and generate the evaluation report.
  • the invention further provides a scoring method based on image pattern recognition technology, comprising: a user initiating a scoring request; determining, by the request determining module, whether the scoring request is legal; scanning a test paper by a scanner to generate a scan result image; The image recognition module receives and recognizes the image through pattern recognition technology The scan result image of the scanner is generated to generate a recognition result; the scoring logic processing is executed by the scoring service processing module to generate a scoring result; and the scoring result is received and saved by the evaluation module to generate an evaluation report.
  • said step of receiving and identifying said scan result image of said scanner by said image recognition module comprises: performing pre-processing on said scan result image by an image pre-processing unit; The image is subjected to template detection; and the image is recognized by the recognition unit by the pattern recognition technique, and the recognition result is stored.
  • the step of performing the scoring logic processing by the scoring service processing module comprises: reading the recognition result by the reading unit; the standard answer by the query unit; and the standard answer by the matching unit The recognition result is matched; and the score unit performs the score statistics to generate the result of the review.
  • the method further comprises: defining a test paper by the test volume module; the student taking the test; and the teacher reviewing the subjective question.
  • the marking system, device and method based on the image pattern recognition technology provided by the invention utilize pattern recognition, can greatly improve the efficiency of the teacher's marking, store the student examination data to the back end of the website, and generate a report of the results, the student or the teacher at any time. View historical exam data.
  • the answer sheet paper and printing requirements are low, just use A4 paper, or just use a black pen to fill the coating without necessarily requiring 2B and other pencils.
  • the system can allow the answer sheet to be partially tilted, and the system automatically corrects the tilted image when recognized.
  • FIG. 1 is a schematic structural diagram of a scoring system based on an image pattern recognition technology according to an embodiment of the present invention.
  • FIG. 2 is a schematic structural diagram of a scoring device based on an image pattern recognition technology according to an embodiment of the present invention.
  • FIG. 3 is a specific flowchart of a scoring method based on an image pattern recognition technology according to an embodiment of the present invention.
  • FIG. 4 is a specific flowchart of an image recognition method according to an embodiment of the present invention.
  • FIG. 5 is a specific flowchart of a method for processing a marking service according to an embodiment of the present invention.
  • FIG. 6 is a block diagram showing the elements of the answer sheet adopted by the scoring system based on the image pattern recognition technology according to an embodiment of the present invention.
  • FIG. 1 is a schematic structural diagram of a scoring system 100 based on an image pattern recognition technology according to an embodiment of the present invention.
  • the scoring system 100 includes a scoring terminal 110, a website back end 120 connected to the scoring terminal 110, and a client 130 connected to the website back end 120.
  • the scoring terminal 110 includes a scanner 112 and a scoring device 114.
  • the scanner 112 is used to scan candidate test papers that have been reviewed by subjective questions.
  • the scoring device 114 is configured to recognize the scan result image of the scanner 112, execute the scribing logic process, and generate a scoring result.
  • the website back end 120 includes a test volume module 122, a student information module 124, and an evaluation module 126.
  • the group volume module 122 is configured to generate a test paper and an answer sheet, define basic information of the test paper, structural information, and an answer and scoring standard, wherein the test paper and the answer sheet downloaded by the test volume module 122 according to the present invention have a specified style. Only the test papers and answer sheets that conform to this style can be reviewed by the marking terminal 110.
  • Student information module 124 is used to store student and teacher user information.
  • the evaluation module 126 is configured to receive and save the marking result of the marking terminal 110, and analyze and generate various evaluation reports, wherein the evaluation report includes basic information of the student, a minimum score warning, and a recommendation suggestion.
  • the client 130 is used for the user to log in using an account, receive an examination result message, view an evaluation report, and the like, and may be an Android or an IOS client.
  • FIG. 2 is a schematic structural diagram of a scoring device 114 based on an image pattern recognition technology according to an embodiment of the present invention.
  • FIG. 6 is a block diagram showing the elements of the answer sheet adopted by the scoring system based on the image pattern recognition technology according to an embodiment of the present invention.
  • the scoring device 200 includes an image recognition module 210, a scoring service processing module 220, and a request determination module 230.
  • the image recognition module 210 is configured to identify scan results of the scanner 112, including an image pre-processing unit 202, a template detection unit 204, and an identification unit 206.
  • the image pre-processing unit 202 is configured to perform pre-processing on the scan result image, including image scale normalization and binarization.
  • the scale normalization refers to scaling the image to a uniform size resolution, for example, 1600*1200, and the invention is not limited thereto.
  • Binarization refers to converting an image into a binary image according to a set threshold 200.
  • the template detecting unit 204 is configured to perform template detection on the image based on the binary image obtained through the preprocessing. Specifically, the template detecting unit 204 first detects a barcode area (such as the test volume barcode A shown in FIG. 6), whether or not there is a sub-area between the black and white strips, and if so, cuts out the section for identification. Next, check the rectangle: first find the right line of the rectangle, then search for the upper line connected to the right line, and finally verify the left and bottom lines.
  • a barcode area such as the test volume barcode A shown in FIG. 6
  • the identification unit 206 is configured to identify the test volume barcode A, the student number B, the objective question area C, the score area D of the subjective question, and the answer area screenshot E of the subjective question shown in FIG. 6, and store the recognition result. Specifically, for the test volume barcode subinterval (for example, the test volume barcode A shown in FIG. 6), the recognition unit 206 recognizes according to the code 39 code rule. First, the training is performed according to the sample pattern, and the unit width of the barcode after normalization is counted. Then, the area to be identified is detected by black and white strip width, converted into a binary code, and then converted into characters by means of table lookup, for example, 100101101101 corresponds to the start character *, and finally the check is performed.
  • the test volume barcode subinterval for example, the test volume barcode A shown in FIG. 6
  • the recognition unit 206 recognizes according to the code 39 code rule. First, the training is performed according to the sample pattern, and the unit width of the barcode after normalization is counted
  • the difference between the number box and the question box (for example, the objective question area C shown in FIG. 6, the score area D of the subject question, and the screenshot E of the subject question) is that the former is right-aligned and the latter is aligned on both sides.
  • the question box for example, the objective question area C shown in FIG. 6, the score area D of the subject question, and the screenshot E of the subject question
  • school Check if the number of grid rows and columns matches the number of the student number box.
  • search for the internal grid if there is a multiple choice question (for example, the objective question area C shown in Figure 6), otherwise the subjective question (for example, the score area D of the subjective question shown in Figure 6, Screenshot of the answer area of the subjective question E).
  • the cells are cut out according to the template to identify, and the cut-out area contains only a single character, such as [1], [2], [A], [B], and the like.
  • the identification method is: the weight of the statistical cell is filled. If the weight exceeds the threshold by 30%, the cell is considered to be filled, otherwise it is unfilled.
  • the scoring service processing module 220 includes a reading unit 222, a query unit 224, a matching unit 226, and a statistics unit 228 for performing scoring logic processing to generate a scoring result.
  • the reading unit 222 is configured to read the recognition result of the image recognition module 210.
  • the query unit 224 is configured to query the standard answer of the group volume module 122.
  • Matching unit 226 is used to match the standard answer with the recognition result.
  • the statistics unit 228 is configured to perform score statistics and generate a scoring result.
  • the request determination module 230 is configured to determine whether the "question request" initiated by the user is legitimate. Since the marking terminal is used by the school, we provide an account for each school. Before they use the marking terminal, they need to log in. If you are logged in with the correct account, the request is legal.
  • FIG. 3 is a specific flowchart of a scoring method 300 based on an image pattern recognition technology according to an embodiment of the present invention. As shown in FIG. 3, the scoring method 300 based on the image pattern recognition technology includes the following steps:
  • Step 301 The test volume is defined by the test volume module 122. Specifically, the method includes generating a test paper and an answer sheet, defining basic information of the test paper, structural information, and an answer and a scoring standard, wherein the test paper and the answer sheet downloaded by the test volume module 122 according to the present invention have a specified style, and only conform to the The style test paper and the answer sheet can be recorded by the marking terminal 110.
  • Step 302 The student takes the test. Among them, the candidate needs to write and fill in the candidate information area in the answer sheet, and apply the objective question answering result to the objective body area of the answer sheet according to the test paper title, and write the result of the subjective question in the subjective answer area of the answer sheet.
  • Step 303 The teacher reviews the subjective question. After the review is completed, the scoring person fills the score of the subjective question in the subjective score of the candidate's answer sheet.
  • Step 304 The user initiates a marking request. Specifically, the user places the test paper on the scanner 112, and points The "starting the marking" on the volume device 114 is clicked, that is, the marking request is initiated.
  • Step 305 The request determination module 230 determines whether the request is legal. Since the marking terminal is used by the school, we provide an account for each school. Before they use the marking terminal, they need to log in. If the login is made using the correct account, the request is determined to be legitimate, and the flow proceeds to step 306; otherwise, the processing of the marking task is refused, and the process ends.
  • Step 306 The scanner 112 scans the test paper to generate a scan result image.
  • the scan result image may be in the .bmp format, but the invention is not limited thereto.
  • Step 307 The scan result image of the scanner 112 is received and recognized by the image recognition module 210. If the scan result image is unclear, corrupted, or the test paper is not the test paper style specified by the test volume module 122, the recognition failure may occur.
  • Step 308 Perform the marking logic processing by the marking service processing module 220 to generate a marking result.
  • Step 309 The reviewing module 126 receives and saves the marking result of the marking terminal 110, and analyzes and generates various evaluation reports.
  • the evaluation report includes information such as the student's basic grade information, minimum score warning and suggestions.
  • FIG. 4 is a specific flowchart of an image recognition method 400 according to an embodiment of the present invention. As shown in FIG. 4, the image recognition method 400 includes the following steps:
  • Step 401 Pre-processing is performed on the image by the image pre-processing unit 202. Specifically, it includes image scale normalization and binarization.
  • the scale normalization refers to scaling the image to a uniform size resolution, for example, 1600*1200, and the invention is not limited thereto.
  • Binarization refers to converting an image into a binary image according to a set threshold 200.
  • Step 402 Perform template detection on the image by the template detecting unit 204 based on the binary image obtained through the preprocessing. Specifically, first, the barcode area (such as the test volume barcode A shown in FIG. 6) is detected, and whether there is a sub-region between the black and white strips, and if so, the section is cut out and identified. Next, check the rectangle: first find the right line of the rectangle, then search for the upper line connected to the right line, and finally verify the left and bottom lines.
  • the barcode area such as the test volume barcode A shown in FIG. 6
  • Step 403 The barcode A, the student number B, and the student number B shown in FIG. 6 are recognized by the identification unit 206.
  • the objective problem area C, the scoring area D of the subjective question, and the screenshot E of the subject question are pattern recognition, and the recognition result is stored.
  • the test volume barcode subinterval (for example, the test volume barcode A shown in FIG. 6) is identified in accordance with the code 39 code rule.
  • the training is performed according to the sample pattern, and the unit width of the barcode after normalization is counted.
  • the area to be identified is detected by black and white strip width, converted into a binary code, and then converted into characters by means of table lookup, for example, 100101101101 corresponds to the start character *, and finally the check is performed.
  • the difference between the number box and the question box is that the former is right-aligned and the latter is aligned on both sides.
  • the former is right-aligned and the latter is aligned on both sides.
  • For a right-aligned box verify that its number of grid rows and columns matches the number of the study number box.
  • search for the internal grid if there is a multiple choice question (for example, the objective question area C shown in Figure 6), otherwise the subjective question (for example, the score area D of the subjective question shown in Figure 6, Screenshot of the answer area of the subjective question E).
  • the cells are cut out according to the template to identify, and the cut-out area contains only a single character, such as [1], [2], [A], [B], and the like.
  • the identification method is: the weight of the statistical cell is filled. If the weight exceeds the threshold by 30%, the cell is considered to be filled, otherwise it is unfilled.
  • FIG. 5 is a specific flowchart of a method for processing a marking service according to an embodiment of the present invention. As shown in FIG. 5, the marking service processing method 500 includes the following steps:
  • Step 501 The recognition result of the image recognition module 210 is read by the reading unit 222.
  • Step 502 The standard answer of the group volume module 122 is queried by the query unit 224.
  • Step 503 Matching the standard answer with the recognition result by the matching unit 226.
  • Step 504 Perform score calculation by the statistic unit 228 to generate a scoring result.
  • the present invention provides a new scoring method and scoring system, which greatly improves the efficiency of the teacher's scoring, stores the student test data to the back end of the website, and generates a report of the results, and the student or the teacher can view the historical test data at any time.

Landscapes

  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

一种基于图像模式识别技术的阅卷系统、装置和方法，其特征在于，上述装置包括：图像识别模块（210），用于通过模式识别技术识别扫描仪的扫描结果图像；阅卷业务处理模块（220），用于执行阅卷逻辑处理，生成阅卷结果；以及请求判定模块（230），用于判定用户发起的"阅卷请求"是否合法。上述基于图像模式识别技术的阅卷系统、装置和方法利用模式识别，能够极大地提升教师阅卷的效率，将学生考试数据存储到网站后端，并生成成绩报表，以便学生或教师随时查看历史考试数据。

Description

一种基于图像模式识别技术的阅卷系统、装置和方法 技术领域
本发明涉及图像处理与模式识别领域，并且特别涉及一种基于图像模式识别技术的阅卷系统、装置以及方法。
背景技术
考试是评估学生知识掌握程度的一种最常用、最普遍的手段，通过各种规模的考试有利于教师教学工作的推进、学生学习成绩的提高和整个教育水平的提升。客观题可以检测学生的判断能力，而主观题则可以反映出学生的逻辑推理、总结概括的能力。因此，一份既含有客观题又含有主观题的试卷具有相当的测试价值。
实现自动化扫描试卷的评阅是一项涉及多方面技术的问题，目前国内外在这方面都进行了一些研究和实践。应用最多的是光学标记阅读机，它基于光电转换的功能原理，以快速、准确的性能使得在国内外标准化考试的阅卷、评分等信息处理系统中得到普遍应用；但以此同时，它还存在着仅能评阅客观题、对答题卡纸张要求高、对答题卡印刷精度要求高、对填涂区灰度的浓度、颜色要求高及需要专门的光标阅读机器等问题。另外一种在大型选拔考试中广泛使用的是网上阅卷系统，它采用试卷和答卷分离的方式，主观题由不同阅卷教师通过网络在计算机上对考生答卷的电子图像分别进行评价，最终由计算机系统自动进行核分和成绩校验的一种阅卷方式。网上阅卷有以下优点：它采用一卷多评制和误差控制机制，有助于阅卷教师更好的把握评分标准、控制误差、提高阅卷质量；但同时存在对阅卷环境、扫描设别、答题纸质要求高，此外庞大的阅卷成本也使其难以在普通的考试中推广应用。
我国基础教育阶段的考试有着频繁性的特点。为了能及时检测学生对知识 点的掌握情况，促进学生知识水平的提高，除了常规的期中、期末考试外，还有增加了月考、小测试等各种规模的考试。与频繁的考试对应的是，基础教育阶段的考试迫切需要快速、准确、小成本的阅卷。教师在较短时间内评阅试卷，时间紧迫、工作量大使得各个环节不可避免的出现人工误差，而且分数登分、核分、保存等各项工作需要耗费人力、物力以及场地等资源。因此，为了满足我国基础教育阶段考试的需要，一种自动化的扫描阅卷系统的出现显得尤为重要。
发明内容
本发明的目的在于提供一种基于图像模式识别技术的阅卷装置，有效解决传统人工阅卷存在的时间紧、任务重、查找纸质试卷工作量大、登记分数和核查分数困难以及耗时耗力等问题，提高教师的阅卷质量，减轻了教师的劳动负担，有效减少人工处理及答卷周转环节中的差错，省时省力，而且试卷图像可存档，为提高阅卷质量提供保障。
本发明采用的技术方案是，设计出一种包含主观题和客观题答卷的自动识别信息阅卷装置，不仅使得教师可以方便快捷地制作答卷，而且能够高效的是吸纳答卷分数等相关信息的识别。
本发明提供一种基于图像模式识别技术的阅卷装置，其特征在于，包括：图像识别模块，通过模式识别技术识别扫描仪的扫描结果图像；阅卷业务处理模块，用于执行阅卷逻辑处理，生成阅卷结果；以及请求判定模块，用于判定用户发起的“阅卷请求”是否合法。
优选地，所述图像识别模块包括：图像预处理单元，用于对所述扫描结果图像执行预处理；模板检测单元，用于基于经过预处理得到的图像，对所述图像进行模板检测；以及识别单元，用于通过所述模式识别技术识别所述图像，并存储识别结果。
优选地，所述阅卷业务处理模块包括：读取单元，用于读取所述识别结果；查询单元，用于查询标准答案；匹配单元，用于将所述标准答案与所述识别结 果进行匹配；以及统计单元，用于进行分值统计，生成所述阅卷结果。
本发明还提供一种基于图像模式识别技术的阅卷系统，其特征在于，包括：阅卷终端，包括扫描仪和阅卷装置，其中所述扫描仪用于扫描经过主观题评阅的考生试卷，所述阅卷装置连接到所述扫描仪，用于通过模式识别技术识别所述扫描仪的扫描结果图像，执行阅卷逻辑处理，产生阅卷结果；网站后端，连接到所述阅卷终端，用于生成试卷，接收并保存所述阅卷终端产生的所述阅卷结果，生成评测报表；以及客户端，连接到所述网站后端，用于接收考试结果消息，查看所述评测报表。
有利地，所述阅卷装置包括：图像识别模块，用于通过所述模式识别技术识别所述扫描仪的所述扫描结果图像；阅卷业务处理模块，用于执行所述阅卷逻辑处理，生成所述阅卷结果；以及请求判定模块，用于判定用户发起的“阅卷请求”是否合法。
有利地，所述图像识别模块包括：图像预处理单元，用于通过所述模式识别技术对所述扫描结果图像执行预处理；模板检测单元，用于基于经过预处理得到的图像，对所述图像进行模板检测；以及识别单元，用于识别所述图像，并存储识别结果。
有利地，所述阅卷业务处理模块包括：读取单元，用于读取所述识别结果；查询单元，用于查询标准答案；匹配单元，用于将所述标准答案与所述识别结果进行匹配；以及统计单元，用于进行分值统计，生成所述阅卷结果。
有利地，所述网站后端包括：组卷模块，用于生成试卷和答题卡，定义试卷的基本信息、结构信息以及答案和评分标准；学生信息模块，用于存贮学生和教师用户信息；以及评测模块，用于接收并保存所述阅卷终端的所述阅卷结果，生成所述评测报表。
本发明又提供一种基于图像模式识别技术的阅卷方法，其特征在于，包括：用户发起阅卷请求；由请求判定模块判定所述阅卷请求是否合法；由扫描仪扫描试卷，生成扫描结果图像；由图像识别模块通过模式识别技术接收并识别所 述扫描仪的所述扫描结果图像，生成识别结果；由阅卷业务处理模块执行阅卷逻辑处理，生成阅卷结果；以及由评测模块接收并保存所述阅卷结果，生成评测报表。
有利地，由图像识别模块接收并识别所述扫描仪的所述扫描结果图像的所述步骤包括：由图像预处理单元对所述扫描结果图像执行预处理；由模板检测单元基于经过预处理得到的图像，进行模板检测；以及由识别单元通过所述模式识别技术识别所述图像，并存储识别结果。
有利地，由阅卷业务处理模块执行阅卷逻辑处理，生成阅卷结果的所述步骤包括：由读取单元读取所述识别结果；由查询单元标准答案；由匹配单元将所述标准答案与所述识别结果进行匹配；以及由统计单元进行分值统计，生成所述阅卷结果。
有利地，在用户发起阅卷请求的所述步骤之前，还包括：由组卷模块定义试卷；学生参加考试；以及教师评阅主观题。
本发明所提供的基于图像模式识别技术的阅卷系统、装置和方法利用模式识别，能够极大地提升了教师阅卷的效率，将学生考试数据存储到网站后端，并生成成绩报表，学生或教师随时查看历史考试数据。此外，答题卡纸张及印刷要求低，只需使用A4纸张即可，也可只使用黑色笔来填涂而不一定要求2B等铅笔。此外，系统可以允许答题纸部分倾斜，在识别时系统可自动纠正倾斜图像。
附图说明
为了更清楚地说明本发明实施例或现有技术中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本发明的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1所示是本发明一实施例提供的基于图像模式识别技术的阅卷系统的结构示意图。
图2所示是本发明一实施例提供的基于图像模式识别技术的阅卷装置的结构示意图。
图3所示是本发明一实施例提供的基于图像模式识别技术的阅卷方法的具体流程图。
图4所示是本发明一实施例提供的图像识别方法的具体流程图。
图5所示是本发明一实施例提供的阅卷业务处理方法的具体流程图。
图6所示是本发明一实施例提供的基于图像模式识别技术的阅卷系统采用的答卷各要素说明图。
具体实施方式
为了使本发明的目的、技术方案及优点更加清楚明白，以下结合附图及实施例，对本发明进行进一步详细说明。应当理解，此处所描述的具体实施例仅仅用以解释本发明，并不用于限定本发明。
图1是本发明一实施例提供的基于图像模式识别技术的阅卷系统100的结构示意图。如图1所示，阅卷系统100包括阅卷终端110、连接到阅卷终端110的网站后端120以及连接到网站后端120的客户端130。
在一实施例中，阅卷终端110包括扫描仪112和阅卷装置114。具体而言，扫描仪112用于扫描经过主观题评阅的考生试卷。阅卷装置114用于识别扫描仪112的扫描结果图像，执行阅卷逻辑处理，产生阅卷结果。
在一实施例中，网站后端120包括组卷模块122、学生信息模块124以及评测模块126。具体而言，组卷模块122用于生成试卷和答题卡，定义试卷的基本信息、结构信息以及答案和评分标准，其中，基于本发明的组卷模块122下载的试卷和答题卡有指定的样式，只有符合这种样式的试卷和答题卡才能通过阅卷终端110进行阅卷。学生信息模块124用于存贮学生和教师用户信息。评测模块126用于接收并保存阅卷终端110的阅卷结果，进行分析生成各类评测报表，其中，评测报表包括学生的基本成绩信息、最低分警告和提分建议等信息。
在一实施例中，客户端130用于用户使用帐号登陆，接收考试结果消息，查看评测报表等信息，可以是安卓或IOS客户端。
图2是本发明一实施例提供的基于图像模式识别技术的阅卷装置114的结构示意图。图6所示是本发明一实施例提供的基于图像模式识别技术的阅卷系统采用的答卷各要素说明图。如图2所示，阅卷装置200包括图像识别模块210、阅卷业务处理模块220以及请求判定模块230。
在一实施例中，图像识别模块210用于识别扫描仪112的扫描结果，包括图像预处理单元202、模板检测单元204和识别单元206。
具体而言，图像预处理单元202用于对扫描结果图像执行预处理，包括图像尺度归一化和二值化。其中尺度归一化是指，将图像放缩为统一大小的分辨率，比如，1600*1200，本发明并不以此为限。二值化是指根据设定的阈值200将图像转换为二值图像。
模板检测单元204用于基于经过预处理得到的二值图像，对图像进行模板检测。具体而言，模板检测单元204首先检测条形码区域(如图6中所示的试卷条形码A)，是否存在黑白条相间的子区域，若存在，将该区间切出进行识别。其次检测矩形框：先找到矩形框的右边线，再搜索与右边线连接的上边线，最后验证左边线和下边线。
识别单元206用于识别图6中所示的试卷条形码A、学生学号B、客观题区域C、主观题的得分区D、主观题的答题区截图E，并存储识别结果。具体而言，对于试卷条形码子区间(例如，图6中所示的试卷条形码A)，识别单元206按照code39码规则进行识别。首先根据样本图形进行训练学习，统计出归一化之后条码的单位宽度。然后将待识别的区域进行黑白条宽度检测，转换为二进制码，再通过查表的方式，将二进制码转换为字符，比如100101101101对应起始字符*，最后进行校验。对于检测到的矩形框进行分类。首先学号框和题目框(例如，图6中所示的客观题区域C、主观题的得分区D以及主观题的答题区截图E)的区别是，前者右对齐，后者两边对齐。对于右对齐的框，校 验其网格行数和列数是否符合学号框的数目。对于题目框，搜索内部网格，若存在则为选择题(例如，图6中所示的客观题区域C)，否则为主观题(例如，图6中所示的主观题的得分区D、主观题的答题区截图E)。矩形框进行分类之后，根据模板将单元格切出进行识别，切出区域只包含单个字符，如[1]、[2]、[A]、[B]等。识别方法为：统计单元格被填涂的权重，若权重超过阈值30％，则认为该单元格被填涂，否则为未填涂。
在一实施例中，阅卷业务处理模块220包括读取单元222、查询单元224、匹配单元226以及统计单元228，用于执行阅卷逻辑处理，生成阅卷结果。具体而言，读取单元222用于读取图像识别模块210的识别结果。查询单元224用于查询组卷模块122的标准答案。匹配单元226用于将标准答案与识别结果进行匹配。统计单元228用于进行分值统计，生成阅卷结果。
在一实施例中，请求判定模块230用于判定用户发起的“阅卷请求”是否合法。由于阅卷终端是学校在使用，我们为每一间学校提供一个账号，他们使用阅卷终端前，需要登录。如果是使用正确帐号登录，则判定请求合法。
图3是本发明一实施例提供的基于图像模式识别技术的阅卷方法300的具体流程图。如图3所示，基于图像模式识别技术的阅卷方法300包括以下步骤：
步骤301：由组卷模块122定义试卷。具体而言，包括生成试卷和答题卡，定义试卷的基本信息、结构信息以及答案和评分标准，其中，基于本发明的组卷模块122下载的试卷和答题卡有指定的样式，只有符合这种样式的试卷和答题卡才能通过阅卷终端110进行阅卷。
步骤302：学生参加考试。其中，考生需要书写并填涂答卷中的考生信息区域，并根据试卷题目将客观题答题结果涂在答卷的客观体区域，将主观题的答题结果书写在答卷的主观题答题区域。
步骤303：教师评阅主观题，评阅完成后，阅卷人员将主观题的得分填涂在考生答卷的主观题分数涂卡区中。
步骤304：用户发起阅卷请求。具体而言，用户将试卷放在扫描仪112，点 击阅卷装置114上的“开始阅卷”，即发起了阅卷请求。
步骤305：请求判定模块230判定请求是否合法。由于阅卷终端是学校在使用，我们为每一间学校提供一个账号，他们使用阅卷终端前，需要登录。如果是使用正确帐号登录，则判定请求合法，流程前进到步骤306；否则，拒绝处理阅卷任务，结束流程。
步骤306：扫描仪112扫描试卷，生成扫描结果图像。其中，扫描结果图像可以是.bmp格式，但本发明并不以此为限。
步骤307：由图像识别模块210接收并识别扫描仪112的扫描结果图像。如果扫描结果图像不清晰、有毁坏、或试卷不是使用组卷模块122指定的试卷样式都会引起识别失败。
步骤308：由阅卷业务处理模块220执行阅卷逻辑处理，生成阅卷结果。
步骤309：由评测模块126接收并保存阅卷终端110的阅卷结果，进行分析生成各类评测报表。其中，评测报表包括学生的基本成绩信息、最低分警告和提分建议等信息。
图4是本发明一实施例提供的图像识别方法400的具体流程图。如图4所示，图像识别方法400包括以下步骤：
步骤401：由图像预处理单元202对图像执行预处理。具体而言，包括图像尺度归一化和二值化。其中尺度归一化是指，将图像放缩为统一大小的分辨率，比如，1600*1200，本发明并不以此为限。二值化是指根据设定的阈值200将图像转换为二值图像。
步骤402：由模板检测单元204基于经过预处理得到的二值图像，对图像进行模板检测。具体而言，首先检测条形码区域(如图6中所示的试卷条形码A)，是否存在黑白条相间的子区域，若存在，将该区间切出进行识别。其次检测矩形框：先找到矩形框的右边线，再搜索与右边线连接的上边线，最后验证左边线和下边线。
步骤403：由识别单元206对图6中所示的试卷条形码A、学生学号B、 客观题区域C、主观题的得分区D、主观题的答题区截图E进行模式识别，并存储识别结果。具体而言，对于试卷条形码子区间(例如，图6中所示的试卷条形码A)，按照code39码规则进行识别。首先根据样本图形进行训练学习，统计出归一化之后条码的单位宽度。然后将待识别的区域进行黑白条宽度检测，转换为二进制码，再通过查表的方式，将二进制码转换为字符，比如100101101101对应起始字符*，最后进行校验。对于检测到的矩形框进行分类。首先学号框和题目框(例如，图6中所示的客观题区域C、主观题的得分区D以及主观题的答题区截图E)的区别是，前者右对齐，后者两边对齐。对于右对齐的框，校验其网格行数和列数是否符合学号框的数目。对于题目框，搜索内部网格，若存在则为选择题(例如，图6中所示的客观题区域C)，否则为主观题(例如，图6中所示的主观题的得分区D、主观题的答题区截图E)。矩形框进行分类之后，根据模板将单元格切出进行识别，切出区域只包含单个字符，如[1]、[2]、[A]、[B]等。识别方法为：统计单元格被填涂的权重，若权重超过阈值30％，则认为该单元格被填涂，否则为未填涂。
图5是本发明一实施例提供的阅卷业务处理方法500的具体流程图。如图5所示，阅卷业务处理方法500包括以下步骤：
步骤501：由读取单元222读取图像识别模块210的识别结果。
步骤502：由查询单元224查询组卷模块122的标准答案。
步骤503：由匹配单元226将标准答案与识别结果进行匹配。
步骤504：由统计单元228进行分值统计，生成阅卷结果。
有利地，本发明提供一种新的阅卷方法和阅卷系统，极大地提升了教师阅卷的效率，将学生考试数据存储到网站后端，并生成成绩报表，学生或教师随时查看历史考试数据。
以上所述仅为本发明的较佳实施例而已，并不用以限制本发明，凡在本发明的精神和原则之内所作的任何修改、等同替换和改进等，均应包含在本发明的保护范围之内。

Claims (12)

  1. 一种基于图像模式识别技术的阅卷装置，其特征在于，包括：
    图像识别模块，通过模式识别技术识别扫描仪的扫描结果图像；
    阅卷业务处理模块，用于执行阅卷逻辑处理，生成阅卷结果；以及
    请求判定模块，用于判定用户发起的“阅卷请求”是否合法。
  2. 如权利要求1所述的阅卷装置，其特征在于，所述图像识别模块包括：
    图像预处理单元，用于对所述扫描结果图像执行预处理；
    模板检测单元，用于基于经过预处理得到的图像，对所述图像进行模板检测；以及
    识别单元，用于通过所述模式识别技术识别所述图像，并存储识别结果。
  3. 如权利要求1所述的阅卷装置，其特征在于，所述阅卷业务处理模块包括：
    读取单元，用于读取所述识别结果；
    查询单元，用于查询标准答案；
    匹配单元，用于将所述标准答案与所述识别结果进行匹配；以及
    统计单元，用于进行分值统计，生成所述阅卷结果。
  4. 一种基于图像模式识别技术的阅卷系统，其特征在于，包括：
    阅卷终端，包括扫描仪和阅卷装置，其中所述扫描仪用于扫描经过主观题评阅的考生试卷，所述阅卷装置连接到所述扫描仪，用于通过模式识别技术识别所述扫描仪的扫描结果图像，执行阅卷逻辑处理，产生阅卷结果；
    网站后端，连接到所述阅卷终端，用于生成试卷，接收并保存所述阅卷终端产生的所述阅卷结果，生成评测报表；以及
    客户端，连接到所述网站后端，用于接收考试结果消息，查看所述评测报表。
  5. 如权利要求4所述的阅卷系统，其特征在于，所述阅卷装置包括：
    图像识别模块，用于通过所述模式识别技术识别所述扫描仪的所述扫描结 果图像；
    阅卷业务处理模块，用于执行所述阅卷逻辑处理，生成所述阅卷结果；以及
    请求判定模块，用于判定用户发起的“阅卷请求”是否合法。
  6. 如权利要求5所述的阅卷系统，其特征在于，所述图像识别模块包括：
    图像预处理单元，用于通过所述模式识别技术对所述扫描结果图像执行预处理；
    模板检测单元，用于基于经过预处理得到的图像，对所述图像进行模板检测；以及
    识别单元，用于识别所述图像，并存储识别结果。
  7. 如权利要求5所述的阅卷系统，其特征在于，所述阅卷业务处理模块包括：
    读取单元，用于读取所述识别结果；
    查询单元，用于查询标准答案；
    匹配单元，用于将所述标准答案与所述识别结果进行匹配；以及
    统计单元，用于进行分值统计，生成所述阅卷结果。
  8. 如权利要求6所述的阅卷系统，其特征在于，所述网站后端包括：
    组卷模块，用于生成试卷和答题卡，定义试卷的基本信息、结构信息以及答案和评分标准；
    学生信息模块，用于存贮学生和教师用户信息；以及
    评测模块，用于接收并保存所述阅卷终端的所述阅卷结果，生成所述评测报表。
  9. 一种基于图像模式识别技术的阅卷方法，其特征在于，包括：
    用户发起阅卷请求；
    由请求判定模块判定所述阅卷请求是否合法；
    由扫描仪扫描试卷，生成扫描结果图像；
    由图像识别模块通过模式识别技术接收并识别所述扫描仪的所述扫描结果图像，生成识别结果；
    由阅卷业务处理模块执行阅卷逻辑处理，生成阅卷结果；以及
    由评测模块接收并保存所述阅卷结果，生成评测报表。
  10. 如权利要求9所述的阅卷方法，其特征在于，由图像识别模块接收并识别所述扫描仪的所述扫描结果图像的所述步骤包括：
    由图像预处理单元对所述扫描结果图像执行预处理；
    由模板检测单元基于经过预处理得到的图像，进行模板检测；以及
    由识别单元通过所述模式识别技术识别所述图像，并存储识别结果。
  11. 如权利要求9所述的阅卷方法，其特征在于，由阅卷业务处理模块执行阅卷逻辑处理，生成阅卷结果的所述步骤包括：
    由读取单元读取所述识别结果；
    由查询单元标准答案；
    由匹配单元将所述标准答案与所述识别结果进行匹配；以及
    由统计单元进行分值统计，生成所述阅卷结果。
  12. 如权利要求9所述的阅卷方法，其特征在于，在用户发起阅卷请求的所述步骤之前，还包括：
    由组卷模块定义试卷；学生参加考试；以及教师评阅主观题。
PCT/CN2016/088814 2016-07-06 2016-07-06 一种基于图像模式识别技术的阅卷系统、装置和方法 WO2018006294A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PCT/CN2016/088814 WO2018006294A1 (zh) 2016-07-06 2016-07-06 一种基于图像模式识别技术的阅卷系统、装置和方法

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2016/088814 WO2018006294A1 (zh) 2016-07-06 2016-07-06 一种基于图像模式识别技术的阅卷系统、装置和方法

Publications (1)

Publication Number Publication Date
WO2018006294A1 true WO2018006294A1 (zh) 2018-01-11

Family

ID=60901325

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2016/088814 WO2018006294A1 (zh) 2016-07-06 2016-07-06 一种基于图像模式识别技术的阅卷系统、装置和方法

Country Status (1)

Country Link
WO (1) WO2018006294A1 (zh)

Cited By (23)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109740436A (zh) * 2018-12-03 2019-05-10 李卫强 一种智能化阅卷系统
CN110110068A (zh) * 2018-02-01 2019-08-09 北大方正集团有限公司 基于专业题库的在线学习方法、系统及计算机设备
CN110110585A (zh) * 2019-03-15 2019-08-09 西安电子科技大学 基于深度学习的智能阅卷实现方法及系统、计算机程序
CN110517316A (zh) * 2018-05-21 2019-11-29 福建鹿鸣教育科技有限公司 一种基于两点定位图像的方法及系统
CN110619326A (zh) * 2019-07-02 2019-12-27 安徽七天教育科技有限公司 一种基于扫描的英语试卷作文检测识别系统及方法
CN110633767A (zh) * 2019-08-06 2019-12-31 安徽学云教育科技有限公司 一种网上阅卷异常筛查方法及系统
CN110705669A (zh) * 2019-11-14 2020-01-17 徐州市奥丁软件信息技术服务有限责任公司 一种校园课堂试卷信息处理系统
CN111274778A (zh) * 2020-02-18 2020-06-12 深圳中兴网信科技有限公司 答题卡生成系统、方法、装置和计算机可读存储介质
CN111597931A (zh) * 2020-04-30 2020-08-28 杭州学同科技有限公司 一种基于边缘计算的机器阅卷方法
CN111666864A (zh) * 2020-06-01 2020-09-15 合肥工业大学 一种基于计算机视觉的自动阅卷方法
CN111666423A (zh) * 2020-06-08 2020-09-15 上海乂学教育科技有限公司 知识图谱构建方法及系统
CN112347946A (zh) * 2020-11-10 2021-02-09 成都兴唐信息技术有限公司 一种多类型答题卡的识别方法及系统
CN112733644A (zh) * 2020-12-30 2021-04-30 广东德诚大数据科技有限公司 一种基于扫描识别的填涂点识别系统及方法
CN112734705A (zh) * 2020-12-30 2021-04-30 广东德诚大数据科技有限公司 一种答题卡的异常检测方法、装置、电子设备及存储介质
CN112765646A (zh) * 2020-12-22 2021-05-07 南京新知艺测科技有限公司 一种艺术测评中书画录入及匿名阅卷方法及系统
CN112766213A (zh) * 2021-01-29 2021-05-07 育林汇软件有限公司 一种电子阅卷式答题卡彩色呈现方法
CN113486786A (zh) * 2021-07-05 2021-10-08 江苏鼎峰信息技术有限公司 一种自动作业批改系统
CN113657354A (zh) * 2021-10-19 2021-11-16 深圳市菁优智慧教育股份有限公司 基于深度学习的答题卡识别方法及系统
CN113705438A (zh) * 2021-08-27 2021-11-26 河北风口飞网络科技有限公司 一种基于多种数据采集形式的考评、阅卷方法及系统
CN114399623A (zh) * 2021-11-30 2022-04-26 西安交通大学 一种通用答题识别方法、系统、存储介质及计算设备
CN116798036A (zh) * 2023-06-27 2023-09-22 广州市南方人力资源评价中心有限公司 用于识别和校验答题卡客观题识别结果的方法及装置
US11967253B2 (en) 2021-05-27 2024-04-23 International Business Machines Corporation Semi-automated evaluation of long answer exams
CN112734705B (zh) * 2020-12-30 2024-06-07 广东德诚科教有限公司 一种答题卡的异常检测方法、装置、电子设备及存储介质

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20070178432A1 (en) * 2006-02-02 2007-08-02 Les Davis Test management and assessment system and method
CN101334814A (zh) * 2008-04-28 2008-12-31 华北电力大学 一种自动化的扫描阅卷系统及阅卷方法
CN101866415A (zh) * 2009-08-24 2010-10-20 深圳市海云天科技股份有限公司 一种计算机阅卷系统的答题卡识别装置及方法
CN102339352A (zh) * 2010-07-20 2012-02-01 上海海鞠电子科技有限公司 电子阅卷方法
CN102799767A (zh) * 2012-06-27 2012-11-28 苏州奇可思信息科技有限公司 一种阅卷软件系统

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20070178432A1 (en) * 2006-02-02 2007-08-02 Les Davis Test management and assessment system and method
CN101334814A (zh) * 2008-04-28 2008-12-31 华北电力大学 一种自动化的扫描阅卷系统及阅卷方法
CN101866415A (zh) * 2009-08-24 2010-10-20 深圳市海云天科技股份有限公司 一种计算机阅卷系统的答题卡识别装置及方法
CN102339352A (zh) * 2010-07-20 2012-02-01 上海海鞠电子科技有限公司 电子阅卷方法
CN102799767A (zh) * 2012-06-27 2012-11-28 苏州奇可思信息科技有限公司 一种阅卷软件系统

Cited By (33)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110110068B (zh) * 2018-02-01 2021-12-03 北大方正集团有限公司 基于专业题库的在线学习方法、系统及计算机设备
CN110110068A (zh) * 2018-02-01 2019-08-09 北大方正集团有限公司 基于专业题库的在线学习方法、系统及计算机设备
CN110517316A (zh) * 2018-05-21 2019-11-29 福建鹿鸣教育科技有限公司 一种基于两点定位图像的方法及系统
CN109740436A (zh) * 2018-12-03 2019-05-10 李卫强 一种智能化阅卷系统
CN110110585A (zh) * 2019-03-15 2019-08-09 西安电子科技大学 基于深度学习的智能阅卷实现方法及系统、计算机程序
CN110110585B (zh) * 2019-03-15 2023-05-30 西安电子科技大学 基于深度学习的智能阅卷实现方法及系统、计算机程序
CN110619326A (zh) * 2019-07-02 2019-12-27 安徽七天教育科技有限公司 一种基于扫描的英语试卷作文检测识别系统及方法
CN110619326B (zh) * 2019-07-02 2023-04-18 安徽七天网络科技有限公司 一种基于扫描的英语试卷作文检测识别系统及方法
CN110633767A (zh) * 2019-08-06 2019-12-31 安徽学云教育科技有限公司 一种网上阅卷异常筛查方法及系统
CN110705669A (zh) * 2019-11-14 2020-01-17 徐州市奥丁软件信息技术服务有限责任公司 一种校园课堂试卷信息处理系统
CN110705669B (zh) * 2019-11-14 2023-04-18 江西三端科技有限公司 一种校园课堂试卷信息处理系统
CN111274778A (zh) * 2020-02-18 2020-06-12 深圳中兴网信科技有限公司 答题卡生成系统、方法、装置和计算机可读存储介质
CN111597931A (zh) * 2020-04-30 2020-08-28 杭州学同科技有限公司 一种基于边缘计算的机器阅卷方法
CN111666864B (zh) * 2020-06-01 2023-03-24 合肥工业大学 一种基于计算机视觉的自动阅卷方法
CN111666864A (zh) * 2020-06-01 2020-09-15 合肥工业大学 一种基于计算机视觉的自动阅卷方法
CN111666423B (zh) * 2020-06-08 2024-04-05 上海松鼠课堂人工智能科技有限公司 知识图谱构建方法及系统
CN111666423A (zh) * 2020-06-08 2020-09-15 上海乂学教育科技有限公司 知识图谱构建方法及系统
CN112347946A (zh) * 2020-11-10 2021-02-09 成都兴唐信息技术有限公司 一种多类型答题卡的识别方法及系统
CN112765646B (zh) * 2020-12-22 2023-05-30 南京新知艺测科技有限公司 一种艺术测评中书画录入及匿名阅卷方法及系统
CN112765646A (zh) * 2020-12-22 2021-05-07 南京新知艺测科技有限公司 一种艺术测评中书画录入及匿名阅卷方法及系统
CN112734705A (zh) * 2020-12-30 2021-04-30 广东德诚大数据科技有限公司 一种答题卡的异常检测方法、装置、电子设备及存储介质
CN112733644A (zh) * 2020-12-30 2021-04-30 广东德诚大数据科技有限公司 一种基于扫描识别的填涂点识别系统及方法
CN112734705B (zh) * 2020-12-30 2024-06-07 广东德诚科教有限公司 一种答题卡的异常检测方法、装置、电子设备及存储介质
CN112766213A (zh) * 2021-01-29 2021-05-07 育林汇软件有限公司 一种电子阅卷式答题卡彩色呈现方法
US11967253B2 (en) 2021-05-27 2024-04-23 International Business Machines Corporation Semi-automated evaluation of long answer exams
CN113486786A (zh) * 2021-07-05 2021-10-08 江苏鼎峰信息技术有限公司 一种自动作业批改系统
CN113486786B (zh) * 2021-07-05 2024-04-09 江苏鼎峰信息技术有限公司 一种自动作业批改系统
CN113705438A (zh) * 2021-08-27 2021-11-26 河北风口飞网络科技有限公司 一种基于多种数据采集形式的考评、阅卷方法及系统
CN113657354A (zh) * 2021-10-19 2021-11-16 深圳市菁优智慧教育股份有限公司 基于深度学习的答题卡识别方法及系统
CN114399623A (zh) * 2021-11-30 2022-04-26 西安交通大学 一种通用答题识别方法、系统、存储介质及计算设备
CN114399623B (zh) * 2021-11-30 2024-04-16 西安交通大学 一种通用答题识别方法、系统、存储介质及计算设备
CN116798036A (zh) * 2023-06-27 2023-09-22 广州市南方人力资源评价中心有限公司 用于识别和校验答题卡客观题识别结果的方法及装置
CN116798036B (zh) * 2023-06-27 2024-04-02 网才科技(广州)集团股份有限公司 用于识别和校验答题卡客观题识别结果的方法及装置

Similar Documents

Publication Publication Date Title
WO2018006294A1 (zh) 一种基于图像模式识别技术的阅卷系统、装置和方法
US11508251B2 (en) Method and system for intelligent identification and correction of questions
US11790641B2 (en) Answer evaluation method, answer evaluation system, electronic device, and medium
CN109271401B (zh) 一种题目搜索、批改方法、装置、电子设备和存储介质
CN109710590B (zh) 一种错题本生成方法及装置
CN111144191B (zh) 字体识别方法、装置、电子设备及存储介质
WO2019075820A1 (zh) 一种试卷批阅系统
CN110110581B (zh) 一种基于人工智能的试卷批改方法和系统
CN110689013A (zh) 一种基于特征识别的自动阅卷方法及系统
CN106846961B (zh) 电子试卷的处理方法和装置
CN109284355B (zh) 一种批改试卷中口算题的方法及装置
US8768241B2 (en) System and method for representing digital assessments
CN109189895B (zh) 一种针对口算题的题目批改方法及装置
CN109389109B (zh) 一种ocr全文本识别正确率的自动化测试方法及设备
CN110135225B (zh) 样本标注方法及计算机存储介质
CN110909035A (zh) 个性化复习题集生成方法、装置、电子设备及存储介质
CN108876195A (zh) 一种智能化的教师教学质量评测系统
CN110826872A (zh) 考试成绩分析报告生成方法、系统、装置
CN106778717A (zh) 一种基于图像识别和k近邻的测评表识别方法
CN114663904A (zh) 一种pdf文档布局检测方法、装置、设备及介质
CN111079641A (zh) 作答内容识别方法、相关设备及可读存储介质
CN113610068B (zh) 基于试卷图像的试题拆解方法、系统、存储介质及设备
CN111008594A (zh) 改错题评阅方法、相关设备及可读存储介质
CN110689447A (zh) 一种基于深度学习的社交软件用户发布内容的实时检测方法
CN113569677A (zh) 一种基于扫描件的纸质试验报告生成方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 16907799

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 13/05//2019)

122 Ep: pct application non-entry in european phase

Ref document number: 16907799

Country of ref document: EP

Kind code of ref document: A1