WO2020156362A1 - 一种试卷批改方法、装置、电子设备及存储介质 - Google Patents

一种试卷批改方法、装置、电子设备及存储介质 Download PDF

Info

Publication number
WO2020156362A1
WO2020156362A1 PCT/CN2020/073397 CN2020073397W WO2020156362A1 WO 2020156362 A1 WO2020156362 A1 WO 2020156362A1 CN 2020073397 W CN2020073397 W CN 2020073397W WO 2020156362 A1 WO2020156362 A1 WO 2020156362A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
marking
area
answering area
test paper
Prior art date
Application number
PCT/CN2020/073397
Other languages
English (en)
French (fr)
Inventor
何涛
毛礼辉
罗欢
陈明权
Original Assignee
杭州大拿科技股份有限公司
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 杭州大拿科技股份有限公司 filed Critical 杭州大拿科技股份有限公司
Priority to US17/425,331 priority Critical patent/US11450081B2/en
Publication of WO2020156362A1 publication Critical patent/WO2020156362A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/70Determining position or orientation of objects or cameras
    • G06T7/73Determining position or orientation of objects or cameras using feature-based methods
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00Image analysis
    • G06T7/70Determining position or orientation of objects or cameras
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V30/00Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
    • G06V30/10Character recognition
    • G06V30/14Image acquisition
    • G06V30/1444Selective acquisition, locating or processing of specific regions, e.g. highlighted text, fiducial marks or predetermined fields
    • G06V30/1448Selective acquisition, locating or processing of specific regions, e.g. highlighted text, fiducial marks or predetermined fields based on markings or identifiers characterising the document or the area
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V30/00Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
    • G06V30/40Document-oriented image-based pattern recognition
    • G06V30/41Analysis of document content
    • G06V30/412Layout analysis of documents structured with printed lines or input boxes, e.g. business forms or tables
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/20Special algorithmic details
    • G06T2207/20081Training; Learning
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/20Special algorithmic details
    • G06T2207/20084Artificial neural networks [ANN]
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06TIMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00Indexing scheme for image analysis or image enhancement
    • G06T2207/30Subject of image; Context of image processing
    • G06T2207/30176Document

Definitions

  • the present invention relates to the technical field of teaching and information processing, in particular to a test paper correction method, device, electronic equipment and computer readable storage medium.
  • the purpose of the present invention is to provide a test paper correction method, device, electronic equipment and computer readable storage medium, so as to solve the problem of low efficiency of teacher correction of test papers in the prior art.
  • the present invention provides a test paper correction method, the method includes:
  • the answering area in the second image that matches the position of the first marking frame corresponding to the answering area is determined, and the determined answering area The answering area of is marked with the second marking box;
  • the determining the position information of the first marking frame corresponding to each answering area includes:
  • the determining, according to the position information of the first marking frame corresponding to each answering area of the first image, the answering area in the second image that matches the position of the first marking frame corresponding to the answering area includes:
  • the corresponding relationship between the second two-dimensional coordinate system and the test paper to be corrected is the same as the corresponding relationship between the first two-dimensional coordinate system and the standard test paper.
  • the position information of the first label frame in the first two-dimensional coordinate system includes: the coordinates of the center point of the first label frame and the height and length of the first label frame.
  • the determining the position information of the first marking frame corresponding to each answering area includes:
  • the determining, according to the position information of the first marking frame corresponding to each answering area of the first image, the answering area in the second image that matches the position of the first marking frame corresponding to the answering area includes:
  • the first marking box corresponding to each answering area of the first image determine the answering area in the second image that matches the position of the first marking box corresponding to the answering area, and compare the determined answering area Carry out the second marking box marking, where the relative position of the marked second marking box in the fourth marking box corresponding to the corresponding question of the test paper to be corrected and the first marking box corresponding to the answering area are in the standard test paper Match the relative position in the third label box corresponding to the corresponding topic.
  • the use of the first marking box to mark the answering area where each standard answer is located includes:
  • the answering area where each standard answer is located is marked with the first label box.
  • test paper correction device which includes:
  • the first obtaining module is used to obtain the first image of the standard test paper, wherein the answer area in the standard test paper is filled with standard answers;
  • the first labeling module is used to identify the answering area of each standard answer and the characters of each standard answer in the first image through a pre-trained recognition model, and use the first label box to mark the answering area of each standard answer;
  • the determining module is used to determine the position information of the first marking box corresponding to each answering area
  • the second obtaining module is used to obtain the second image of the test paper to be corrected, wherein the answer area in the test paper to be corrected is filled with the answer to be corrected;
  • the second marking module is used to determine the position of the first marking frame corresponding to the answering area in the second image according to the position information of the first marking frame corresponding to each answering area of the first image The answering area, and marking the determined answering area with a second marking frame;
  • a recognition module configured to recognize the characters of the answer to be corrected in each second annotation box of the second image through the pre-trained recognition model
  • the correction module is used to compare the characters of the standard answer in the first marking box corresponding to each answering area in the first image with the answer to be corrected in the second marking box corresponding to the corresponding answering area in the second image The characters are compared, and the correction of the test paper to be corrected is completed.
  • the determining module determining the position information of the first marking frame corresponding to each answering area includes:
  • the second labeling module determines, according to the position information of the first labeling box corresponding to each answering area of the first image, the answering that matches the position of the first labeling box corresponding to the answering area in the second image Area, including:
  • the corresponding relationship between the second two-dimensional coordinate system and the test paper to be corrected is the same as the corresponding relationship between the first two-dimensional coordinate system and the standard test paper.
  • the position information of the first label frame in the first two-dimensional coordinate system includes: the coordinates of the center point of the first label frame and the height and length of the first label frame.
  • the determining module determining the position information of the first marking frame corresponding to each answering area includes:
  • the second labeling module determines, according to the position information of the first labeling box corresponding to each answering area of the first image, the answering that matches the position of the first labeling box corresponding to the answering area in the second image Area, including:
  • the first marking box corresponding to each answering area of the first image determine the answering area in the second image that matches the position of the first marking box corresponding to the answering area, and compare the determined answering area Carry out the second marking box marking, where the relative position of the marked second marking box in the fourth marking box corresponding to the corresponding question of the test paper to be corrected and the first marking box corresponding to the answering area are in the standard test paper Match the relative position in the third label box corresponding to the corresponding topic.
  • the first marking module uses the first marking box to mark the answering area of each standard answer, including:
  • the answering area where each standard answer is located is marked with the first label box.
  • the present invention also provides an electronic device, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory complete each other through the communication bus.
  • Communication including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory complete each other through the communication bus.
  • the memory is used to store computer programs
  • the processor When the processor is used to execute the computer program stored in the memory, it realizes the test paper correction method described above.
  • the present invention also provides a computer-readable storage medium in which a computer program is stored, and when the computer program is executed, the test paper correction method as described in any one of the above is implemented.
  • the present invention recognizes the character content of the standard answer in the standard test paper and the location information of the standard answer for the standard test paper.
  • the test paper to be corrected determines the matching pending test paper according to the determined location information of the standard answer. Correct the location information of the answer, and identify the character content of the answer to be corrected, so as to compare the recognized standard answer with the matched answer to be corrected, complete the correction of the test paper to be corrected, without manual correction of the test paper, which solves the problem in the prior art
  • the teacher corrects questions with low efficiency in the test paper; in addition, for the standard test paper and the test paper to be corrected, only the character content of the answer is recognized, and the content of the rest of the test paper is ignored, which further improves the correction speed.
  • FIG. 1 is a schematic flowchart of a test paper correction method provided by an embodiment of the present invention
  • FIG. 2 is a schematic diagram of the structure of a test paper correction device provided by an embodiment of the present invention.
  • FIG. 3 is a schematic structural diagram of an electronic device provided by an embodiment of the present invention.
  • embodiments of the present invention provide a test paper correction method, device, electronic equipment, and computer-readable storage medium.
  • test paper correction method of the embodiment of the present invention can be applied to the test paper correction device of the embodiment of the present invention, and the test paper correction device can be configured on an electronic device.
  • the electronic device may be a personal computer, a mobile terminal, etc.
  • the mobile terminal may be a hardware device with various operating systems such as a mobile phone or a tablet computer.
  • FIG. 1 is a schematic flowchart of a method for correcting test papers according to an embodiment of the present invention. Please refer to Figure 1.
  • An examination paper correction method can include the following steps:
  • step S101-step S103 to process the standard test paper (such as the answer test paper with the teacher's handwritten answer).
  • the standard test paper such as the answer test paper with the teacher's handwritten answer.
  • Step S101 obtaining the first image of the standard test paper
  • the answer area in the standard test paper is filled with standard answers
  • Step S102 Recognizing the area of each standard answer and the characters of each standard answer in the first image through a pre-trained recognition model, and using a label box to mark the answering area where each standard answer is located;
  • Step S103 Determine the position information of the marking frame corresponding to each answering area.
  • each standard answer in the first image can be stored, and after determining the position information of the label box corresponding to each answer area, each The location information of the marked box is stored, so that when the number of test papers to be corrected is large, each test paper to be corrected can be corrected one by one according to the stored standard answer characters and the location information of the marked box, so that when the number of test papers to be corrected is large, further Improve the speed of correction.
  • step S104 to step S106 are executed to process the test paper to be corrected (for example, a test paper with a student's handwritten answer).
  • Step S104 obtaining a second image of the test paper to be corrected
  • the answer area in the test paper to be corrected is filled with the answer to be corrected
  • Step S105 Determine the answering area in the second image that matches the position of the labeling frame corresponding to the answering area in the first image according to the position information of the marked box corresponding to each answering area of the first image, and compare the determined answering area Mark the answer area with a box;
  • Step S106 using the pre-trained recognition model, recognize the characters of the answer to be corrected in each label box of the second image.
  • first image and second image may be obtained by scanning, or by other methods such as photographing, which is not limited in the present invention.
  • step S107 can be executed to change the standard test paper Make a one-to-one correspondence with the answers in the test paper to be corrected, and complete the correction of the test paper to be corrected.
  • Step S107 Compare the characters of the standard answer in the marked box corresponding to each answering area in the first image with the characters of the answer to be corrected in the marked box corresponding to the corresponding answering area in the second image, and complete the comparison. The correction of the test paper to be corrected.
  • step S105 there may be an unfilled answer area in the test paper to be corrected (that is, there is no character in the answer area to be corrected).
  • step S105 the character recognition result for the answer to be corrected in this answering area is empty, so in step S107, since the characters of the answer to be corrected do not match the characters of the standard answer, the correction result for this answering area is Is an error.
  • the pre-trained recognition model can be established based on the hole convolution and attention model. Specifically, the answers in the test paper training samples are labeled, and the hole convolution is used to identify the answers. Perform feature extraction on the labeled frame of, and then decode the extracted features into characters through the attention model, thereby training the recognition model.
  • the pre-trained recognition model can recognize the characters of each standard answer in the first image, and can also recognize the characters of the answer to be corrected in each label box of the second image.
  • step S102 the location of each standard answer can be recognized through the recognition model (for example, the area where the handwritten font can be recognized on the standard test paper, that is, the location of the standard answer), and then the characters of each standard answer can be recognized, and the answer area Make an annotation.
  • the recognition model for example, the area where the handwritten font can be recognized on the standard test paper, that is, the location of the standard answer
  • a label box is used to mark the answering area where each standard answer is located.
  • a pre-trained labeling model may be used to mark the answering area where each standard answer is located with a label box.
  • the determined answering area is marked with a frame.
  • the pre-trained marking model may be used to mark the determined answering area with a frame.
  • the labeling model can be a neural network-based model. Specifically, the labeling model can be obtained through the following process: labeling the answering area where the answer in the test paper training sample is located, and using the test paper training sample that has undergone the labeling process to The network is trained to obtain the annotation model.
  • the answering area is the area in brackets in the question for choice and judgment questions, so the area inside the brackets is marked with a marking box, and for fill-in-the-blank questions, the answering area is the area above the horizontal line in the question, so use the marking box Mark the area above the horizontal line.
  • the marked box corresponding to the answer area of the oral arithmetic question is the blank area behind the equal sign, and the marked box corresponding to the answer area of the calculation question is the blank area below the stem to the top of the next question.
  • the position information of the marking frame corresponding to each answering area determined in step S103 will be described in detail below.
  • the position information of the label box corresponding to each answering area can be the position of the label box corresponding to each answering area in the entire test paper, or it can be the label box corresponding to each answering area in the corresponding question (the corresponding question is the answering area Corresponding title).
  • the method of determining the position information of the label frame corresponding to each answering area in step S103 may be: establishing a first two-dimensional coordinate system for the first image, and determining the corresponding The position information of the label frame in the first two-dimensional coordinate system.
  • the position information of the label frame in the first two-dimensional coordinate system may include: the coordinates of the center point of the label frame, and the height and length of the label frame.
  • the first two-dimensional coordinate system may take the position of any pixel in the first image as the origin, and any two mutually perpendicular directions as the horizontal axis and the vertical axis.
  • the origin of the first two-dimensional coordinate system may be the position of the first pixel in the first image (that is, the pixel corresponding to the first row and first column of the first image), and the horizontal axis and the vertical axis are respectively Be the upper and left edges of the first image (where the upper and left edges of the first image are perpendicular to each other); or, the origin of the first two-dimensional coordinate system can be the upper left vertex of the standard test paper in the first image
  • the horizontal axis and the vertical axis are respectively the upper edge and the left edge of the standard test paper in the first image (where the upper and left edges of the standard test paper are perpendicular to each other).
  • step S105 according to the position information of the label box corresponding to each answer area of the first image, determine the answer area in the second image that matches the position of the label box corresponding to the answer area.
  • the way can be:
  • a second two-dimensional coordinate system is established for the second image, and for the label box corresponding to each answering area of the first image, the second image in the second two-dimensional coordinate system is determined The answering area whose position information matches the position information of the marking frame corresponding to the answering area in the first two-dimensional coordinate system.
  • the origin of the first two-dimensional coordinate system is the pixel corresponding to the upper left vertex of the standard test paper in the first image, and the horizontal axis and vertical axis are the upper and left edges of the standard test paper in the first image, then,
  • the origin of the two-dimensional coordinate system is the pixel corresponding to the upper left vertex of the test paper to be corrected in the second image, and the horizontal axis and the vertical axis are the upper and left edges of the test paper to be corrected in the second image, respectively.
  • the origin of the first two-dimensional coordinate system is the position of the first pixel in the first image, and the horizontal axis and the vertical axis are the upper and left edges of the first image
  • the second two-dimensional The origin of the coordinate system is the first pixel in the second image, and the horizontal axis and the vertical axis are the upper and left edges of the second image, respectively. It should be noted that this situation requires that the first image and the second image have the same basis for establishing a two-dimensional coordinate system.
  • a specific scanning device can be used to scan the standard test paper and the test paper to be corrected to obtain the first image And the second image, the specific scanning device can fix the test paper to be scanned at the same specific position and scan it, so that it can ensure that the two two-dimensional coordinate systems in the first image and the second image obtained The correspondence between the test papers is the same.
  • the corresponding marking box for each answer area of the first image is based on the The position information of the label box in the first two-dimensional coordinate system can be found in the second two-dimensional coordinate system to match the location information, that is, the answering area of the same question in the second image can be found.
  • the center point coordinates of the label box corresponding to answering area 1 in the first image in the first two-dimensional coordinate system are (10, 10), and the height and length are 4 and 4, respectively.
  • the area where the center point coordinates of the two-dimensional coordinate system are (10, 10), and the height and length are 4 and 4 respectively, which is the determined answering area of the same question in the second image.
  • the center point coordinates are not strictly required to be exactly equal, and the center point coordinates may be allowed to be within a certain error range.
  • the method of determining the position information of the marking box corresponding to each answering area in step S103 may be: identifying the area where each question of the standard test paper in the first image is located, and marking the box Mark; determine the relative position of the mark box corresponding to each answering area in the mark box corresponding to the corresponding question.
  • a pre-trained item recognition model may be used to identify the area where each item of the standard test paper in the first image is located, and mark the area with a frame.
  • the topic recognition model can be a neural network-based model. Use the trained topic recognition model to extract the two-dimensional feature vector from the first image, generate anchor points of different shapes in each grid of the two-dimensional feature vector, and use the label box to mark the area of each identified topic, and The label box and the generated anchor point can be processed by regression to make the label box closer to the actual position of the title. After identifying the topic area, each topic can be cut into a single area, or not actually cut, and each topic area is distinguished during processing, processed as a single area, and sorted according to the topic location information.
  • step S105 according to the position information of the label box corresponding to each answer area of the first image, determine the answer area in the second image that matches the position of the label box corresponding to the answer area.
  • the way can be:
  • the relative position of the marked marking box in the marking box corresponding to the corresponding question of the test paper to be corrected is the same as the relative position of the marking box corresponding to the answering area in the marking box corresponding to the corresponding question of the standard test paper match.
  • the above-mentioned pre-trained item recognition model can be used to identify the area where each item of the test paper to be corrected in the second image is located, and mark the area of each item after identifying the item area. It is a single area, or not actually cut, and each topic area is distinguished during processing, processed as a single area, and sorted according to the position information of the topic.
  • the relative position of the marking box corresponding to each answering area within the marking box corresponding to the corresponding question can be determined.
  • the area where each question of the test paper to be corrected in the second image is located is also marked with a mark box, and further, for the mark box corresponding to each answer area in the first image, according to the mark box in the mark box corresponding to the corresponding question
  • For the relative position an area with a matching relative position can be found in the label box corresponding to the corresponding question in the second image, that is, the answering area of the same question in the second image can be found.
  • the relative position is not strictly required to be equal, and the relative position can be allowed to be at a certain level. Within the error range.
  • step S105 according to the position information of the label frame corresponding to each answer area of the first image, determine the answer area in the second image that matches the position of the label frame corresponding to the answer area, and compare After the determined answering area is marked with the label box, the correspondence between the matching answering areas in the two images can also be established. In this way, in step S107, the answers in the two images can be directly mapped to each other according to the correspondence. Make corrections.
  • the present invention recognizes the character content of the standard answer in the standard test paper and the location information of the standard answer for the standard test paper.
  • the test paper to be corrected determines the matching pending test paper according to the determined location information of the standard answer. Correct the location information of the answer, and identify the character content of the answer to be corrected, so as to compare the recognized standard answer with the matched answer to be corrected, complete the correction of the test paper to be corrected, without manual correction of the test paper, which solves the problem in the prior art
  • the teacher corrects questions with low efficiency in the test paper; in addition, for the standard test paper and the test paper to be corrected, only the character content of the answer is recognized, and the content of the rest of the test paper is ignored, which further improves the correction speed.
  • the present invention provides a test paper correction device.
  • the device may include:
  • the first obtaining module 201 is used to obtain a first image of a standard test paper, wherein the answering area in the standard test paper is filled with standard answers;
  • the first labeling module 202 is configured to recognize the area of each standard answer and the characters of each standard answer in the first image through a pre-trained recognition model, and use a label box to mark the answering area where each standard answer is located;
  • the determining module 203 is configured to determine the position information of the marking frame corresponding to each answering area
  • the second obtaining module 204 is used to obtain a second image of the test paper to be corrected, wherein the answer area in the test paper to be corrected is filled with the answer to be corrected;
  • the second marking module 205 is configured to determine the answering area in the second image that matches the position of the marking frame corresponding to the answering area in the second image according to the position information of the marking box corresponding to each answering area of the first image, And mark the determined answering area with a frame;
  • the recognition module 206 is configured to recognize the characters of the answer to be corrected in each label box of the second image through the pre-trained recognition model;
  • the correction module 207 is used to compare the characters of the standard answer in the marked box corresponding to each answering area in the first image with the characters of the answer to be corrected in the marked box corresponding to the corresponding answering area in the second image , Complete the correction of the test paper to be corrected.
  • the determining module 203 determines the location information of the marking frame corresponding to each answering area, specifically:
  • the second marking module 205 determines the answering area in the second image that matches the position of the marking frame corresponding to the answering area according to the position information of the marking box corresponding to each answering area of the first image, specifically for:
  • a second two-dimensional coordinate system is established for the second image, and for the label box corresponding to each answering area of the first image, the second image in the second two-dimensional coordinate system is determined The answering area whose position information matches the position information of the marking frame corresponding to the answering area in the first two-dimensional coordinate system;
  • the corresponding relationship between the second two-dimensional coordinate system and the test paper to be corrected is the same as the corresponding relationship between the first two-dimensional coordinate system and the standard test paper.
  • the position information of the label frame in the first two-dimensional coordinate system includes: the coordinates of the center point of the label frame, and the height and length of the label frame.
  • the determining module 203 determines the location information of the marking frame corresponding to each answering area, specifically:
  • the second marking module 205 determines the answering area in the second image that matches the position of the marking frame corresponding to the answering area according to the position information of the marking box corresponding to each answering area of the first image, specifically for:
  • the relative position of the marked marking box in the marking box corresponding to the corresponding question of the test paper to be corrected is the same as the relative position of the marking box corresponding to the answering area in the marking box corresponding to the corresponding question of the standard test paper match.
  • the first marking module 202 uses a marking box to mark the answering area where each standard answer is located, specifically:
  • the answering area where each standard answer is located is marked with a label box.
  • the present invention also provides an electronic device, as shown in FIG. 3, including a processor 301, a communication interface 302, a memory 303, and a communication bus 304.
  • the processor 301, the communication interface 302, and the memory 303 complete each other through the communication bus 304. Communication between,
  • the memory 303 is used to store computer programs
  • the processor 301 is configured to implement the following steps when executing the computer program stored in the memory 303:
  • the position information of the marked box corresponding to each answering area of the first image determine the answering area in the second image that matches the position of the marked box corresponding to the answering area, and perform a check on the determined answering area Label box label;
  • test paper correction method implemented by the processor 301 executing the computer program stored in the memory 303 are the same as the implementations mentioned in the foregoing method embodiment section, and will not be repeated here.
  • the communication bus mentioned in the above electronic device may be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus.
  • PCI Peripheral Component Interconnect
  • EISA Extended Industry Standard Architecture
  • the communication bus can be divided into address bus, data bus, control bus and so on. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.
  • the communication interface is used for communication between the aforementioned electronic device and other devices.
  • the memory may include random access memory (Random Access Memory, RAM), and may also include non-volatile memory (Non-Volatile Memory, NVM), such as at least one disk storage.
  • NVM non-Volatile Memory
  • the memory may also be at least one storage device located far away from the foregoing processor.
  • the foregoing processor may be a general-purpose processor, including a central processing unit (CPU), a network processor (Network Processor, NP), etc.; it may also be a digital signal processor (Digital Signal Processing, DSP), a dedicated integrated Circuit (Application Specific Integrated Circuit, ASIC), Field-Programmable Gate Array (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components.
  • CPU central processing unit
  • NP Network Processor
  • DSP Digital Signal Processing
  • ASIC Application Specific Integrated Circuit
  • FPGA Field-Programmable Gate Array
  • FPGA Field-Programmable Gate Array
  • the present invention also provides a computer-readable storage medium in which a computer program is stored, and when the computer program is executed by a processor, the method steps of the above test paper correction method are realized.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Multimedia (AREA)
  • Artificial Intelligence (AREA)
  • Image Analysis (AREA)
  • Character Input (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

一种试卷批改方法、装置、电子设备及存储介质,方法包括:获得标准试卷的第一图像(S101);识别第一图像中各个标准答案的区域以及字符,并使用标注框标注出各个标准答案所在的作答区域(S102);确定每个标注框的位置信息(S103);获得待批改试卷的第二图像(S104);根据第一图像的每个标注框的位置信息,确定第二图像中与该标注框的位置相匹配的作答区域,并对所确定的作答区域进行标注框标注(S105);识别第二图像的各个标注框内待批改答案的字符(S106);将第一图像中每一标注框内标准答案的字符与第二图像中相对应的标注框内待批改答案的字符进行比较,完成对待批改试卷的批改(S107)。应用该方法可以解决现有技术中老师批改试卷效率较低的问题。

Description

一种试卷批改方法、装置、电子设备及存储介质 技术领域
本发明涉及教学及信息处理技术领域,尤其涉及一种试卷批改方法、装置、电子设备和计算机可读存储介质。
背景技术
目前,老师大多以试卷的形式给学生布置作业,以及测验学生的学习效果,这也就导致了老师们需要批改大量的学生试卷。然而现有技术中,老师批改试卷的方式也都比较传统,通常是基于手写的方式批改,这种方式效率普遍低下而且也容易出错,例如,如果一个老师每天要批改60个同学的试卷,每个同学批改5分钟,每天需要工作5个小时。
因此,如何提高老师批改试卷的效率是一个亟待解决的问题。
发明内容
本发明的目的在于提供一种试卷批改方法、装置、电子设备和计算机可读存储介质,以解决现有技术中老师批改试卷效率较低的问题。
为达到上述目的,本发明提供了一种试卷批改方法,所述方法包括:
获得标准试卷的第一图像,其中,所述标准试卷中的作答区域填写有标准答案;
通过预先训练的识别模型识别所述第一图像中各个标准答案的作答区域以及各个标准答案的字符,并使用第一标注框标注出各个标准答案的作答区域;
确定每个作答区域对应的第一标注框的位置信息;
获得待批改试卷的第二图像,其中,所述待批改试卷中的作答区域填写有待批改答案;
根据所述第一图像的每个作答区域对应的第一标注框的位置信息,确定所述第二图像中与该作答区域对应的第一标注框的位置相匹配的作答区域,并对所确定的作答区域进行第二标注框标注;
通过所述预先训练的识别模型,识别所述第二图像的各个第二标注框内待批改答案的字符;
将所述第一图像中每一作答区域对应的第一标注框内标准答案的字符与所述第二图像中相对应的作答区域对应的第二标注框内待批改答案的字符进行比较,完成对所述待批改试卷的批改。
可选的,所述确定每个作答区域对应的第一标注框的位置信息,包括:
对所述第一图像建立第一二维坐标系,确定每个作答区域对应的第一标注框在所述第一二维坐标系中的位置信息;
所述根据所述第一图像的每个作答区域对应的第一标注框的位置信息,确定所述第二图像中与该作答区域对应的第一标注框的位置相匹配的作答区域,包括:
对所述第二图像建立第二二维坐标系,根据所述第一图像的每个作答区域对应的第一标注框,确定所述第二图像中的且在所述第二二维坐标系中的位置信息与该作答区域对应的第一标注框在所述第一二维坐标系中的位置信息相匹配的作答区域;
其中,所述第二二维坐标系与所述待批改试卷之间的对应关系与所述第一二维坐标系与所述标准试卷之间的对应关系相同。
可选的,第一标注框在所述第一二维坐标系中的位置信息包括:第一标注框的中心点坐标以及第一标注框的高度和长度。
可选的,所述确定每个作答区域对应的第一标注框的位置信息,包括:
识别所述第一图像中所述标准试卷的各个题目所在区域,并进行第三标注框标注;
确定每个作答区域对应的第一标注框在相应题目对应的第三标注框内的相对位置;
所述根据所述第一图像的每个作答区域对应的第一标注框的位置信息,确定所述第二图像中与该作答区域对应的第一标注框的位置相匹配的作答区域,包括:
识别所述第二图像中所述待批改试卷的各个题目所在区域,并进行第四 标注框标注;
根据所述第一图像的每个作答区域对应的第一标注框,确定所述第二图像中与该作答区域对应的第一标注框的位置相匹配的作答区域,并对所确定的作答区域进行第二标注框标注,其中,所标注的第二标注框在所述待批改试卷的相应题目对应的第四标注框内的相对位置与该作答区域对应的第一标注框在所述标准试卷的相应题目对应的第三标注框内的相对位置相匹配。
可选的,所述使用第一标注框标注出各个标准答案所在的作答区域,包括:
采用预先训练的标注模型,将各个标准答案所在的作答区域用第一标注框标注出来。
为达到上述目的,本发明还提供了一种试卷批改装置,所述装置包括:
第一获得模块,用于获得标准试卷的第一图像,其中,所述标准试卷中的作答区域填写有标准答案;
第一标注模块,用于通过预先训练的识别模型识别所述第一图像中各个标准答案的作答区域以及各个标准答案的字符,并使用第一标注框标注出各个标准答案的作答区域;
确定模块,用于确定每个作答区域对应的第一标注框的位置信息;
第二获得模块,用于获得待批改试卷的第二图像,其中,所述待批改试卷中的作答区域填写有待批改答案;
第二标注模块,用于根据所述第一图像的每个作答区域对应的第一标注框的位置信息,确定所述第二图像中与该作答区域对应的第一标注框的位置相匹配的作答区域,并对所确定的作答区域进行第二标注框标注;
识别模块,用于通过所述预先训练的识别模型,识别所述第二图像的各个第二标注框内的待批改答案的字符;
批改模块,用于将所述第一图像中每一作答区域对应的第一标注框内标准答案的字符与所述第二图像中相对应的作答区域对应的第二标注框内待批改答案的字符进行比较,完成对所述待批改试卷的批改。
可选的,所述确定模块确定每个作答区域对应的第一标注框的位置信息, 包括:
对所述第一图像建立第一二维坐标系,确定每个作答区域对应的第一标注框在所述第一二维坐标系中的位置信息;
所述第二标注模块根据所述第一图像的每个作答区域对应的第一标注框的位置信息,确定所述第二图像中与该作答区域对应的第一标注框的位置相匹配的作答区域,包括:
对所述第二图像建立第二二维坐标系,根据所述第一图像的每个作答区域对应的第一标注框,确定所述第二图像中的且在所述第二二维坐标系中的位置信息与该作答区域对应的第一标注框在所述第一二维坐标系中的位置信息相匹配的作答区域;
其中,所述第二二维坐标系与所述待批改试卷之间的对应关系与所述第一二维坐标系与所述标准试卷之间的对应关系相同。
可选的,第一标注框在所述第一二维坐标系中的位置信息包括:第一标注框的中心点坐标以及第一标注框的高度和长度。
可选的,所述确定模块确定每个作答区域对应的第一标注框的位置信息,包括:
识别所述第一图像中所述标准试卷的各个题目所在区域,并进行第三标注框标注;
确定每个作答区域对应的第一标注框在相应题目对应的第三标注框内的相对位置;
所述第二标注模块根据所述第一图像的每个作答区域对应的第一标注框的位置信息,确定所述第二图像中与该作答区域对应的第一标注框的位置相匹配的作答区域,包括:
识别所述第二图像中所述待批改试卷的各个题目所在区域,并进行第四标注框标注;
根据所述第一图像的每个作答区域对应的第一标注框,确定所述第二图像中与该作答区域对应的第一标注框的位置相匹配的作答区域,并对所确定的作答区域进行第二标注框标注,其中,所标注的第二标注框在所述待批改 试卷的相应题目对应的第四标注框内的相对位置与该作答区域对应的第一标注框在所述标准试卷的相应题目对应的第三标注框内的相对位置相匹配。
可选的,所述第一标注模块使用第一标注框标注出各个标准答案的作答区域,包括:
采用预先训练的标注模型,将各个标准答案所在的作答区域用第一标注框标注出来。
为达到上述目的,本发明还提供了一种电子设备,包括处理器、通信接口、存储器和通信总线,其中,所述处理器、所述通信接口和所述存储器通过所述通信总线完成相互间的通信;
所述存储器用于存放计算机程序;
所述处理器用于执行所述存储器上所存放的所述计算机程序时,实现如上任一所述的试卷批改方法。
为达到上述目的,本发明还提供了一种计算机可读存储介质,所述计算机可读存储介质内存储有计算机程序,所述计算机程序被执行时实现如上任一项所述的试卷批改方法。
与现有技术相比,本发明对于标准试卷,识别标准试卷中标准答案的字符内容以及确定标准答案的位置信息,对于待批改试卷,根据所确定的标准答案的位置信息来确定相匹配的待批改答案的位置信息,并识别待批改答案的字符内容,从而将识别的标准答案和相匹配的待批改答案进行比较,完成对待批改试卷的批改,不需要人工批改试卷,解决了现有技术中老师批改试卷效率较低的问题;另外,对于标准试卷和待批改试卷,只需识别其中的答案的字符内容,而忽略试卷中其余部分的内容,进一步提高了批改速度。
附图说明
图1是本发明一实施例提供的试卷批改方法的流程示意图;
图2是本发明一实施例提供的试卷批改装置的结构示意图;
图3是本发明一实施例提供的电子设备的结构示意图。
具体实施方式
以下结合附图和具体实施例对本发明提出的一种试卷批改方法、装置、电子设备及计算机可读存储介质作进一步详细说明。根据权利要求书和下面说明,本发明的优点和特征将更清楚。需说明的是,附图均采用非常简化的形式且均使用非精准的比例,仅用以方便、明晰地辅助说明本发明实施例的目的。
为解决现有技术的问题,本发明实施例提供了一种试卷批改方法、装置、电子设备及计算机可读存储介质。
需要说明的是,本发明实施例的试卷批改方法可应用于本发明实施例的试卷批改装置,该试卷批改装置可被配置于电子设备上。其中,该电子设备可以是个人计算机、移动终端等,该移动终端可以是手机、平板电脑等具有各种操作系统的硬件设备。
图1是本发明一实施例提供的一种试卷批改方法的流程示意图。请参考图1,一种试卷批改方法可以包括如下步骤:
首先,执行步骤S101-步骤S103,对标准试卷(如教师手写答案的作答试卷)进行处理。
步骤S101,获得标准试卷的第一图像;
其中,所述标准试卷中的作答区域填写有标准答案;
步骤S102,通过预先训练的识别模型识别所述第一图像中各个标准答案的区域以及各个标准答案的字符,并使用标注框标注出各个标准答案所在的作答区域;
步骤S103,确定每个作答区域对应的标注框的位置信息。
可以理解的是,在识别出所述第一图像中各个标准答案的字符后,可以对各个标准答案的字符进行存储,以及,在确定每个作答区域对应的标注框的位置信息后,对各个标注框的位置信息进行存储,以便于当待批改试卷的数量较多时,可以根据存储的标准答案字符以及标注框的位置信息逐一对各个待批改试卷进行批改,从而在待批改试卷数量较多时进一步提高批改速度。
然后,执行步骤S104-步骤S106,对待批改试卷(如学生手写答案的作 答试卷)进行处理。
步骤S104,获得待批改试卷的第二图像;
其中,所述待批改试卷中的作答区域填写有待批改答案;
步骤S105,根据所述第一图像的每个作答区域对应的标注框的位置信息,确定所述第二图像中与该作答区域对应的标注框的位置相匹配的作答区域,并对所确定的作答区域进行标注框标注;
步骤S106,通过所述预先训练的识别模型,识别所述第二图像的各个标注框内待批改答案的字符。
上述第一图像和第二图像可以通过扫描的方式获得,也可以通过拍照等其它方式获得,本发明对此不做限定。
最后,在对标准试卷和待批改试卷分别进行以上处理后,可以得到标准试卷和待批改试卷中各个答案的字符内容以及两个试卷中答案的对应关系,因此,可以执行步骤S107,将标准试卷和待批改试卷中的答案进行一一对应,完成对待批改试卷的批改。
步骤S107,将所述第一图像中每一作答区域对应的标注框内标准答案的字符与所述第二图像中相对应的作答区域对应的标注框内待批改答案的字符进行比较,完成对所述待批改试卷的批改。
需要说明的是,待批改试卷中可能存在未填写答案的作答区域(即这种作答区域内没有待批改答案的字符),对于这种作答区域,在步骤S105中同样会被标注出来,而在步骤S106中针对这种作答区域内待批改答案的字符识别结果即为空,从而在步骤S107中,由于待批改答案的字符与标准答案的字符不匹配,因此针对这种作答区域的批改结果即为错误。
在实际应用中,在步骤S102和S106中,预先训练的识别模型可以是基于空洞卷积和注意力模型建立的,具体的,对试卷训练样本中的答案进行标注,采用空洞卷积对答案所在的标注框进行特征提取,再通过注意力模型将提取到的特征解码成字符,从而训练得到识别模型。则预先训练的识别模型可以识别出所述第一图像中各个标准答案的字符,也可以识别出第二图像的各个标注框内待批改答案的字符。
步骤S102中通过识别模型可以识别出各个标准答案所在的位置(例如可以在标准试卷上识别出手写字体的区域,即为标准答案所在的位置),然后识别各个标准答案的字符,并对作答区域进行标注。
在步骤S102中使用标注框标注出各个标准答案所在的作答区域,具体可以采用预先训练的标注模型,将各个标准答案所在的作答区域用标注框标注出来。同样的,在步骤S105中对所确定的作答区域进行标注框标注,具体可以采用所述预先训练的标注模型,对所确定的作答区域进行标注框标注。
标注模型可以是基于神经网络的模型,具体的,标注模型可以通过如下过程得到:对试卷训练样本中答案所在的作答区域进行标注框标注处理,利用经过所述标注处理的试卷训练样本,对神经网络进行训练,以得到所述标注模型。
在标注试卷训练样本时,对于选择判断题,作答区域为题目中的括号内区域,因此用标注框标注括号内区域,对于填空题,作答区域为题目中的横线上方区域,因此用标注框标注横线上方区域,类似的,口算题的作答区域对应的标注框为等号后边的空白区域,计算题的作答区域对应的标注框为题干下方直至下一道题目上方的空白区域。
下面对步骤S103中确定每个作答区域对应的标注框的位置信息进行详细说明。每个作答区域对应的标注框的位置信息可以为每个作答区域对应的标注框在整个试卷中的位置,也可以为每个作答区域对应的标注框在相应题目(相应题目即为该作答区域所对应的题目)中的位置。
对于第一种情形,步骤S103中所述确定每个作答区域对应的标注框的位置信息的方式,可以为:对所述第一图像建立第一二维坐标系,确定每个作答区域对应的标注框在所述第一二维坐标系中的位置信息。
其中,标注框在所述第一二维坐标系中的位置信息可以包括:标注框的中心点坐标、标注框的高度和长度。
可以理解的是,第一二维坐标系可以是以第一图像中任一像素点所在的位置为原点,以任意两个互相垂直的方向为横轴和纵轴。例如,第一二维坐标系的原点可以为第一图像中的第一个像素点(即第一图像中第一行第一列 所对应的像素点)所在的位置,横轴和纵轴分别为第一图像的上边缘和左边缘(其中,第一图像的上边缘和左边缘是互相垂直的);或者,第一二维坐标系的原点可以为第一图像中标准试卷的左上顶点所对应的像素点,横轴和纵轴分别为第一图像中标准试卷的上边缘和左边缘(其中,标准试卷的上边缘和左边缘是互相垂直的)。建立第一二维坐标系后,即可确定每个作答区域对应的标注框的中心点坐标,以及标注框的高度和长度。
相应的,步骤S105中所述根据所述第一图像的每个作答区域对应的标注框的位置信息,确定所述第二图像中与该作答区域对应的标注框的位置相匹配的作答区域的方式,可以为:
对所述第二图像建立第二二维坐标系,针对所述第一图像的每个作答区域对应的标注框,确定所述第二图像中的、在所述第二二维坐标系中的位置信息与该作答区域对应的标注框在所述第一二维坐标系中的位置信息相匹配的作答区域。
需要强调的是,所述第二二维坐标系与所述待批改试卷之间的对应关系与所述第一二维坐标系与所述标准试卷之间的对应关系相同。
例如,若第一二维坐标系的原点为第一图像中标准试卷的左上顶点所对应的像素点,横轴和纵轴分别为第一图像中标准试卷的上边缘和左边缘,则,第二二维坐标系的原点为第二图像中待批改试卷的左上顶点所对应的像素点,横轴和纵轴分别为第二图像中待批改试卷的上边缘和左边缘。
再如,若第一二维坐标系的原点为第一图像中的第一个像素点所在的位置,横轴和纵轴分别为第一图像的上边缘和左边缘,则,第二二维坐标系的原点为第二图像中的第一个像素点,横轴和纵轴分别为第二图像的上边缘和左边缘。需要注意的是,这种情况需要第一图像和第二图像具有相同的建立二维坐标系的基础,在实际应用中可以使用特定的扫描设备对标准试卷和待批改试卷进行扫描得到第一图像和第二图像,该特定的扫描设备能够将待扫描的试卷固定在相同的特定位置上再进行扫描,这样就可以保证所得到的第一图像和第二图像中,两个二维坐标系与试卷之间的对应关系相同。
由于第二二维坐标系与待批改试卷之间的对应关系与第一二维坐标系与 标准试卷之间的对应关系相同,则针对第一图像的每个作答区域对应的标注框,根据该标注框在第一二维坐标系中的位置信息,可以从第二二维坐标系中找到位置信息相匹配的区域,即找到同一题目在第二图像中的作答区域。
举例而言,第一图像中作答区域1对应的标注框在第一二维坐标系的中心点坐标为(10,10)、高度和长度分别为4和4,则在第二图像中寻找第二二维坐标系的中心点坐标为(10,10)、高度和长度分别为4和4的区域,即为所确定的同一题目在第二图像中的作答区域。需要注意的是,为提高位置信息匹配的容错性,在第二图像中寻找位置信息相匹配的作答区域时,不严格要求中心点坐标完全相等,可以允许中心点坐标在一定的误差范围内。
对于第二种情形,步骤S103中所述确定每个作答区域对应的标注框的位置信息的方式,可以为:识别所述第一图像中所述标准试卷的各个题目所在区域,并进行标注框标注;确定每个作答区域对应的标注框在相应题目对应的标注框内的相对位置。
其中,可以通过预先训练的题目识别模型,来识别所述第一图像中所述标准试卷的各个题目所在区域,并进行标注框标注。题目识别模型可以是基于神经网络的模型。利用训练好的题目识别模型从第一图像中提取二维特征向量,在二维特征向量的每个网格生成不同形状的锚点,使用标注框将识别出的各个题目的区域进行标注,还可以将标注框与生成的锚点作回归(regression)处理,以使标注框更贴近题目的实际位置。识别完题目区域后可以将每道题目进行切割为单个区域,或者不实际切割,而在处理时将每个题目区域区分开,作为单个区域进行处理,并根据题目位置信息进行排序。
相应的,步骤S105中所述根据所述第一图像的每个作答区域对应的标注框的位置信息,确定所述第二图像中与该作答区域对应的标注框的位置相匹配的作答区域的方式,可以为:
识别所述第二图像中所述待批改试卷的各个题目所在区域,并进行标注框标注;
针对所述第一图像的每个作答区域对应的标注框,确定所述第二图像中与该作答区域对应的标注框的位置相匹配的作答区域,并对所确定的作答区 域进行标注框标注,其中,所标注的标注框在所述待批改试卷的相应题目对应的标注框内的相对位置与该作答区域对应的标注框在所述标准试卷的相应题目对应的标注框内的相对位置相匹配。
同样的,可以通过上述的预先训练的题目识别模型,来识别所述第二图像中所述待批改试卷的各个题目所在区域,并进行标注框标注识别完题目区域后可以将每道题目进行切割为单个区域,或者不实际切割,而在处理时将每个题目区域区分开,作为单个区域进行处理,并根据题目位置信息进行排序。
可以理解的是,对第一图像中标准试卷的各个作答区域以及各个题目所在区域均进行标注框标注,则可以确定每个作答区域对应的标注框在相应题目对应的标注框内的相对位置。并且,对第二图像中待批改试卷的各个题目所在区域也进行标注框标注,进而,针对第一图像中每个作答区域对应的标注框,根据该标注框在相应题目对应的标注框内的相对位置,可以从第二图像中相应题目对应的标注框内找到相对位置相匹配的区域,即找到同一题目在第二图像中的作答区域。
需要注意的是,为提高位置信息匹配的容错性,在待批改试卷的各个题目对应的批注框内寻找相对位置相匹配的作答区域时,不严格要求相对位置完全相等,可以允许相对位置在一定的误差范围内。
另外,步骤S105在根据所述第一图像的每个作答区域对应的标注框的位置信息,确定所述第二图像中与该作答区域对应的标注框的位置相匹配的作答区域,并对所确定的作答区域进行标注框标注之后,还可以建立两个图像中相匹配的作答区域之间的对应关系,这样,在步骤S107中可以直接根据对应关系,将两个图像中的答案一一对应进行批改。
与现有技术相比,本发明对于标准试卷,识别标准试卷中标准答案的字符内容以及确定标准答案的位置信息,对于待批改试卷,根据所确定的标准答案的位置信息来确定相匹配的待批改答案的位置信息,并识别待批改答案的字符内容,从而将识别的标准答案和相匹配的待批改答案进行比较,完成对待批改试卷的批改,不需要人工批改试卷,解决了现有技术中老师批改试 卷效率较低的问题;另外,对于标准试卷和待批改试卷,只需识别其中的答案的字符内容,而忽略试卷中其余部分的内容,进一步提高了批改速度。
相应于上述试卷批改方法实施例,本发明提供了一种试卷批改装置,参见图2,该装置可以包括:
第一获得模块201,用于获得标准试卷的第一图像,其中,所述标准试卷中的作答区域填写有标准答案;
第一标注模块202,用于通过预先训练的识别模型识别所述第一图像中各个标准答案的区域以及各个标准答案的字符,并使用标注框标注出各个标准答案所在的作答区域;
确定模块203,用于确定每个作答区域对应的标注框的位置信息;
第二获得模块204,用于获得待批改试卷的第二图像,其中,所述待批改试卷中的作答区域填写有待批改答案;
第二标注模块205,用于根据所述第一图像的每个作答区域对应的标注框的位置信息,确定所述第二图像中与该作答区域对应的标注框的位置相匹配的作答区域,并对所确定的作答区域进行标注框标注;
识别模块206,用于通过所述预先训练的识别模型,识别所述第二图像的各个标注框内的待批改答案的字符;
批改模块207,用于将所述第一图像中每一作答区域对应的标注框内标准答案的字符与所述第二图像中相对应的作答区域对应的标注框内待批改答案的字符进行比较,完成对所述待批改试卷的批改。
可选的,所述确定模块203确定每个作答区域对应的标注框的位置信息,具体为:
对所述第一图像建立第一二维坐标系,确定每个作答区域对应的标注框在所述第一二维坐标系中的位置信息;
所述第二标注模块205根据所述第一图像的每个作答区域对应的标注框的位置信息,确定所述第二图像中与该作答区域对应的标注框的位置相匹配的作答区域,具体为:
对所述第二图像建立第二二维坐标系,针对所述第一图像的每个作答区域对应的标注框,确定所述第二图像中的、在所述第二二维坐标系中的位置信息与该作答区域对应的标注框在所述第一二维坐标系中的位置信息相匹配的作答区域;
其中,所述第二二维坐标系与所述待批改试卷之间的对应关系与所述第一二维坐标系与所述标准试卷之间的对应关系相同。
可选的,标注框在所述第一二维坐标系中的位置信息包括:标注框的中心点坐标、标注框的高度和长度。
可选的,所述确定模块203确定每个作答区域对应的标注框的位置信息,具体为:
识别所述第一图像中所述标准试卷的各个题目所在区域,并进行标注框标注;
确定每个作答区域对应的标注框在相应题目对应的标注框内的相对位置;
所述第二标注模块205根据所述第一图像的每个作答区域对应的标注框的位置信息,确定所述第二图像中与该作答区域对应的标注框的位置相匹配的作答区域,具体为:
识别所述第二图像中所述待批改试卷的各个题目所在区域,并进行标注框标注;
针对所述第一图像的每个作答区域对应的标注框,确定所述第二图像中与该作答区域对应的标注框的位置相匹配的作答区域,并对所确定的作答区域进行标注框标注,其中,所标注的标注框在所述待批改试卷的相应题目对应的标注框内的相对位置与该作答区域对应的标注框在所述标准试卷的相应题目对应的标注框内的相对位置相匹配。
可选的,所述第一标注模块202使用标注框标注出各个标准答案所在的作答区域,具体为:
采用预先训练的标注模型,将各个标准答案所在的作答区域用标注框标注出来。
本发明还提供了一种电子设备,如图3所示,包括处理器301、通信接口302、存储器303和通信总线304,其中,处理器301,通信接口302,存储器303通过通信总线304完成相互间的通信,
存储器303,用于存放计算机程序;
处理器301,用于执行存储器303上所存放的计算机程序时,实现如下步骤:
获得标准试卷的第一图像,其中,所述标准试卷中的作答区域填写有标准答案;
通过预先训练的识别模型识别所述第一图像中各个标准答案的区域以及各个标准答案的字符,并使用标注框标注出各个标准答案所在的作答区域;
对所述第一图像中各个标准答案所在的作答区域进行标注框标注,并确定每个作答区域对应的标注框的位置信息;
获得待批改试卷的第二图像,其中,所述待批改试卷中的作答区域填写有待批改答案;
根据所述第一图像的每个作答区域对应的标注框的位置信息,确定所述第二图像中与该作答区域对应的标注框的位置相匹配的作答区域,并对所确定的作答区域进行标注框标注;
通过所述预先训练的识别模型,识别所述第二图像的各个标注框内待批改答案的字符;
将所述第一图像中每一作答区域对应的标注框内标准答案的字符与所述第二图像中相对应的作答区域对应的标注框内待批改答案的字符进行比较,完成对所述待批改试卷的批改。
关于该方法各个步骤的具体实现以及相关解释内容可以参见上述图1所示的方法实施例,在此不做赘述。
另外,处理器301执行存储器303上所存放的计算机程序而实现的试卷批改方法的其他实现方式,与前述方法实施例部分所提及的实现方式相同,这里也不再赘述。
上述电子设备提到的通信总线可以是外设部件互连标准(Peripheral Component Interconnect,PCI)总线或扩展工业标准结构(Extended Industry Standard Architecture,EISA)总线等。该通信总线可以分为地址总线、数据总线、控制总线等。为便于表示,图中仅用一条粗线表示,但并不表示仅有一根总线或一种类型的总线。
通信接口用于上述电子设备与其他设备之间的通信。
存储器可以包括随机存取存储器(Random Access Memory,RAM),也可以包括非易失性存储器(Non-Volatile Memory,NVM),例如至少一个磁盘存储器。可选的,存储器还可以是至少一个位于远离前述处理器的存储装置。
上述的处理器可以是通用处理器,包括中央处理器(Central Processing Unit,CPU)、网络处理器(Network Processor,NP)等;还可以是数字信号处理器(Digital Signal Processing,DSP)、专用集成电路(Application Specific Integrated Circuit,ASIC)、现场可编程门阵列(Field-Programmable Gate Array,FPGA)或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件。
本发明还提供了一种计算机可读存储介质,该计算机可读存储介质内存储有计算机程序,该计算机程序被处理器执行时实现上述的试卷批改方法的方法步骤。
需要说明的是,本说明书中的各个实施例均采用相关的方式描述,各个实施例之间相同相似的部分互相参见即可,每个实施例重点说明的都是与其他实施例的不同之处。尤其,对于装置、电子设备、计算机可读存储介质实施例而言,由于其基本相似于方法实施例,所以描述的比较简单,相关之处参见方法实施例的部分说明即可。
在本文中,诸如第一和第二等之类的关系术语仅仅用来将一个实体或者操作与另一个实体或操作区分开来,而不一定要求或者暗示这些实体或操作之间存在任何这种实际的关系或者顺序。而且,术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含,从而使得包括一系列要素的过程、 方法、物品或者设备不仅包括那些要素,而且还包括没有明确列出的其他要素,或者是还包括为这种过程、方法、物品或者设备所固有的要素。在没有更多限制的情况下,由语句“包括一个……”限定的要素,并不排除在包括所述要素的过程、方法、物品或者设备中还存在另外的相同要素。
上述描述仅是对本发明较佳实施例的描述,并非对本发明范围的任何限定,本发明领域的普通技术人员根据上述揭示内容做的任何变更、修饰,均属于权利要求书的保护范围。

Claims (10)

  1. 一种试卷批改方法,其特征在于,所述方法包括:
    获得标准试卷的第一图像,其中,所述标准试卷中的作答区域填写有标准答案;
    通过预先训练的识别模型识别所述第一图像中各个标准答案的作答区域以及各个标准答案的字符,并使用第一标注框标注出各个标准答案的作答区域;
    确定每个作答区域对应的第一标注框的位置信息;
    获得待批改试卷的第二图像,其中,所述待批改试卷中的作答区域填写有待批改答案;
    根据所述第一图像的每个作答区域对应的第一标注框的位置信息,确定所述第二图像中与该作答区域对应的第一标注框的位置相匹配的作答区域,并对所确定的作答区域进行第二标注框标注;
    通过所述预先训练的识别模型,识别所述第二图像的各个第二标注框内待批改答案的字符;
    将所述第一图像中每一作答区域对应的第一标注框内标准答案的字符与所述第二图像中相对应的作答区域对应的第二标注框内待批改答案的字符进行比较,完成对所述待批改试卷的批改。
  2. 如权利要求1所述的试卷批改方法,其特征在于,所述确定每个作答区域对应的第一标注框的位置信息,包括:
    对所述第一图像建立第一二维坐标系,确定每个作答区域对应的第一标注框在所述第一二维坐标系中的位置信息;
    所述根据所述第一图像的每个作答区域对应的第一标注框的位置信息,确定所述第二图像中与该作答区域对应的第一标注框的位置相匹配的作答区域,包括:
    对所述第二图像建立第二二维坐标系,根据所述第一图像的每个作答区域对应的第一标注框,确定所述第二图像中的且在所述第二二维坐标系中的 位置信息与该作答区域对应的第一标注框在所述第一二维坐标系中的位置信息相匹配的作答区域;
    其中,所述第二二维坐标系与所述待批改试卷之间的对应关系与所述第一二维坐标系与所述标准试卷之间的对应关系相同。
  3. 如权利要求2所述的试卷批改方法,其特征在于,第一标注框在所述第一二维坐标系中的位置信息包括:第一标注框的中心点坐标以及第一标注框的高度和长度。
  4. 如权利要求1所述的试卷批改方法,其特征在于,所述确定每个作答区域对应的第一标注框的位置信息,包括:
    识别所述第一图像中所述标准试卷的各个题目所在区域,并进行第三标注框标注;
    确定每个作答区域对应的第一标注框在相应题目对应的第三标注框内的相对位置;
    所述根据所述第一图像的每个作答区域对应的第一标注框的位置信息,确定所述第二图像中与该作答区域对应的第一标注框的位置相匹配的作答区域,包括:
    识别所述第二图像中所述待批改试卷的各个题目所在区域,并进行第四标注框标注;
    根据所述第一图像的每个作答区域对应的第一标注框,确定所述第二图像中与该作答区域对应的第一标注框的位置相匹配的作答区域,并对所确定的作答区域进行第二标注框标注,其中,所标注的第二标注框在所述待批改试卷的相应题目对应的第四标注框内的相对位置与该作答区域对应的第一标注框在所述标准试卷的相应题目对应的第三标注框内的相对位置相匹配。
  5. 如权利要求1所述的试卷批改方法,其特征在于,所述使用第一标注框标注出各个标准答案的作答区域,包括:
    采用预先训练的标注模型,将各个标准答案所在的作答区域用第一标注框标注出来。
  6. 一种试卷批改装置,其特征在于,所述装置包括:
    第一获得模块,用于获得标准试卷的第一图像,其中,所述标准试卷中的作答区域填写有标准答案;
    第一标注模块,用于通过预先训练的识别模型识别所述第一图像中各个标准答案的作答区域以及各个标准答案的字符,并使用第一标注框标注出各个标准答案的作答区域;
    确定模块,用于确定每个作答区域对应的第一标注框的位置信息;
    第二获得模块,用于获得待批改试卷的第二图像,其中,所述待批改试卷中的作答区域填写有待批改答案;
    第二标注模块,用于根据所述第一图像的每个作答区域对应的第一标注框的位置信息,确定所述第二图像中与该作答区域对应的第一标注框的位置相匹配的作答区域,并对所确定的作答区域进行第二标注框标注;
    识别模块,用于通过所述预先训练的识别模型,识别所述第二图像的各个第二标注框内的待批改答案的字符;
    批改模块,用于将所述第一图像中每一作答区域对应的第一标注框内标准答案的字符与所述第二图像中相对应的作答区域对应的第二标注框内待批改答案的字符进行比较,完成对所述待批改试卷的批改。
  7. 如权利要求6所述的试卷批改装置,其特征在于,所述确定模块确定每个作答区域对应的第一标注框的位置信息,包括:
    对所述第一图像建立第一二维坐标系,确定每个作答区域对应的第一标注框在所述第一二维坐标系中的位置信息;
    所述第二标注模块根据所述第一图像的每个作答区域对应的第一标注框的位置信息,确定所述第二图像中与该作答区域对应的第一标注框的位置相匹配的作答区域,包括:
    对所述第二图像建立第二二维坐标系,根据所述第一图像的每个作答区域对应的第一标注框,确定所述第二图像中的且在所述第二二维坐标系中的位置信息与该作答区域对应的第一标注框在所述第一二维坐标系中的位置信息相匹配的作答区域;
    其中,所述第二二维坐标系与所述待批改试卷之间的对应关系与所述第 一二维坐标系与所述标准试卷之间的对应关系相同。
  8. 如权利要求6所述的试卷批改装置,其特征在于,所述确定模块确定每个作答区域对应的第一标注框的位置信息,包括:
    识别所述第一图像中所述标准试卷的各个题目所在区域,并进行第三标注框标注;
    确定每个作答区域对应的第一标注框在相应题目对应的第三标注框内的相对位置;
    所述第二标注模块根据所述第一图像的每个作答区域对应的第一标注框的位置信息,确定所述第二图像中与该作答区域对应的第一标注框的位置相匹配的作答区域,包括:
    识别所述第二图像中所述待批改试卷的各个题目所在区域,并进行第四标注框标注;
    根据所述第一图像的每个作答区域对应的第一标注框,确定所述第二图像中与该作答区域对应的第一标注框的位置相匹配的作答区域,并对所确定的作答区域进行第二标注框标注,其中,所标注的第二标注框在所述待批改试卷的相应题目对应的第四标注框内的相对位置与该作答区域对应的第一标注框在所述标准试卷的相应题目对应的第三标注框内的相对位置相匹配。
  9. 一种电子设备,其特征在于,包括处理器、通信接口、存储器和通信总线,其中,所述处理器、所述通信接口和所述存储器通过所述通信总线完成相互间的通信;
    所述存储器用于存放计算机程序;
    所述处理器用于执行所述存储器上所存放的所述计算机程序时,实现如权利要求1-5中任一所述的方法。
  10. 一种计算机可读存储介质,其特征在于,所述计算机可读存储介质内存储有计算机程序,所述计算机程序被执行时实现如权利要求1-5中任一项所述的方法。
PCT/CN2020/073397 2019-02-02 2020-01-21 一种试卷批改方法、装置、电子设备及存储介质 WO2020156362A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US17/425,331 US11450081B2 (en) 2019-02-02 2020-01-21 Examination paper correction method and apparatus, electronic device, and storage medium

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910107570.3A CN109815932B (zh) 2019-02-02 2019-02-02 一种试卷批改方法、装置、电子设备及存储介质
CN201910107570.3 2019-02-02

Publications (1)

Publication Number Publication Date
WO2020156362A1 true WO2020156362A1 (zh) 2020-08-06

Family

ID=66605279

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/073397 WO2020156362A1 (zh) 2019-02-02 2020-01-21 一种试卷批改方法、装置、电子设备及存储介质

Country Status (3)

Country Link
US (1) US11450081B2 (zh)
CN (1) CN109815932B (zh)
WO (1) WO2020156362A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112989768A (zh) * 2021-04-26 2021-06-18 北京世纪好未来教育科技有限公司 连线题批改方法、装置、电子设备及存储介质
CN113971805A (zh) * 2021-12-22 2022-01-25 深圳市迪博企业风险管理技术有限公司 一种结合机器视觉和语义分析的智能阅卷评分方法

Families Citing this family (17)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109815932B (zh) * 2019-02-02 2021-05-28 杭州大拿科技股份有限公司 一种试卷批改方法、装置、电子设备及存储介质
CN110399838A (zh) * 2019-07-25 2019-11-01 联想(北京)有限公司 处理方法和电子设备
CN110555427A (zh) * 2019-09-09 2019-12-10 北京一起教育信息咨询有限责任公司 一种书写规范度的自动评分方法、系统、设备及存储介质
CN111104883B (zh) * 2019-12-09 2023-06-06 平安国际智慧城市科技股份有限公司 作业答案提取方法、装置、设备及计算机可读存储介质
CN112149561B (zh) * 2020-09-23 2024-04-16 杭州睿琪软件有限公司 图像处理方法和装置、电子设备和存储介质
CN112287001B (zh) * 2020-10-30 2023-05-30 北京有竹居网络技术有限公司 一种信息展示方法、装置、计算机设备及存储介质
CN112287154A (zh) * 2020-10-30 2021-01-29 北京有竹居网络技术有限公司 一种信息统计方法、装置、计算机设备及存储介质
CN112837398A (zh) * 2021-02-23 2021-05-25 百度在线网络技术(北京)有限公司 文本批注的方法、装置、电子设备及存储介质
CN112949621A (zh) * 2021-03-16 2021-06-11 新东方教育科技集团有限公司 试卷作答区域的标记方法、装置、存储介质及电子设备
CN113076872B (zh) * 2021-04-01 2022-07-15 杭州云梯科技有限公司 一种智能试卷批改方法
CN113065554B (zh) * 2021-04-08 2023-05-02 读书郎教育科技有限公司 一种用于智慧课堂智能批阅的批改区域自动识别方法
CN113111154B (zh) * 2021-06-11 2021-10-29 北京世纪好未来教育科技有限公司 相似度评估方法、答案搜索方法、装置、设备及介质
CN113255613B (zh) * 2021-07-06 2021-09-24 北京世纪好未来教育科技有限公司 判题方法、装置及计算机存储介质
CN113326815B (zh) * 2021-07-12 2022-05-06 北京世纪好未来教育科技有限公司 一种文档处理方法、装置、电子设备及可读存储介质
CN113569845A (zh) * 2021-07-13 2021-10-29 读书郎教育科技有限公司 一种智能辅助教师自动批改的方法及智能台灯
CN113255629B (zh) * 2021-07-15 2022-08-02 北京世纪好未来教育科技有限公司 文档处理方法、装置、电子设备及计算机可读存储介质
CN113705736A (zh) * 2021-10-27 2021-11-26 北京世纪好未来教育科技有限公司 一种答案确定方法、判题方法及装置和电子设备

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102855495A (zh) * 2012-08-22 2013-01-02 苏州多捷电子科技有限公司 电子版标准答案的实现方法及其应用系统
CN105046200A (zh) * 2015-06-19 2015-11-11 成都理想境界科技有限公司 基于直线检测的电子阅卷方法
CN108932508A (zh) * 2018-08-13 2018-12-04 杭州大拿科技股份有限公司 一种题目智能识别、批改的方法和系统
CN109815932A (zh) * 2019-02-02 2019-05-28 杭州大拿科技股份有限公司 一种试卷批改方法、装置、电子设备及存储介质

Family Cites Families (21)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6927872B2 (en) * 2001-07-25 2005-08-09 Hewlett-Packard Development Company, L.P. Data acquisition system and method using answer forms
JP4553241B2 (ja) * 2004-07-20 2010-09-29 株式会社リコー 文字方向識別装置、文書処理装置及びプログラム並びに記憶媒体
JP4894189B2 (ja) * 2005-08-02 2012-03-14 富士ゼロックス株式会社 教材処理装置、教材処理方法、教材処理プログラム及び記録媒体
CN101685482A (zh) * 2009-08-04 2010-03-31 上海心意答电子科技有限公司 一种自动识别处理批阅结果的电子阅卷系统及方法
JP4935891B2 (ja) * 2009-12-21 2012-05-23 ブラザー工業株式会社 画像合成装置及び画像合成プログラム
US9734132B1 (en) * 2011-12-20 2017-08-15 Amazon Technologies, Inc. Alignment and reflow of displayed character images
US9536141B2 (en) * 2012-06-29 2017-01-03 Palo Alto Research Center Incorporated System and method for forms recognition by synthesizing corrected localization of data fields
US9552516B2 (en) * 2012-08-29 2017-01-24 Palo Alto Research Center Incorporated Document information extraction using geometric models
US20150187219A1 (en) * 2013-12-27 2015-07-02 Cloud Ta Llc Systems and methods for computer-assisted grading of printed tests
US9235758B1 (en) * 2014-06-30 2016-01-12 Adobe Systems Incorporated Robust method to find layout similarity between two documents
CN104376315B (zh) * 2014-07-30 2018-01-02 天津科技大学 一种基于计算机图像处理和模式识别的检测方法及应用
CN104268603B (zh) * 2014-09-16 2017-04-12 科大讯飞股份有限公司 用于文字性客观题的智能阅卷方法及系统
US10043231B2 (en) * 2015-06-30 2018-08-07 Oath Inc. Methods and systems for detecting and recognizing text from images
US10013643B2 (en) * 2016-07-26 2018-07-03 Intuit Inc. Performing optical character recognition using spatial information of regions within a structured document
CN107784264A (zh) * 2017-05-05 2018-03-09 合肥林翔智能科技有限公司 基于图像处理的答卷分析方法及系统
US10706530B2 (en) * 2017-09-11 2020-07-07 International Business Machines Corporation Object detection
CN107590494B (zh) * 2017-09-18 2020-07-14 北京成长教育科技有限公司 答题卡图片定位方法、装置、可读存储介质以及电子设备
CN108171297B (zh) * 2018-01-24 2022-01-14 谢德刚 一种答题卡识别方法
CN108764074B (zh) * 2018-05-14 2019-03-19 山东师范大学 基于深度学习的主观题智能阅卷方法、系统及存储介质
US10755039B2 (en) * 2018-11-15 2020-08-25 International Business Machines Corporation Extracting structured information from a document containing filled form images
EP3891656A4 (en) * 2018-12-04 2022-08-24 Leverton Holding LLC METHODS AND SYSTEMS FOR AUTOMATIC TABLE RECOGNITION IN DOCUMENTS

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102855495A (zh) * 2012-08-22 2013-01-02 苏州多捷电子科技有限公司 电子版标准答案的实现方法及其应用系统
CN105046200A (zh) * 2015-06-19 2015-11-11 成都理想境界科技有限公司 基于直线检测的电子阅卷方法
CN108932508A (zh) * 2018-08-13 2018-12-04 杭州大拿科技股份有限公司 一种题目智能识别、批改的方法和系统
CN109815932A (zh) * 2019-02-02 2019-05-28 杭州大拿科技股份有限公司 一种试卷批改方法、装置、电子设备及存储介质

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112989768A (zh) * 2021-04-26 2021-06-18 北京世纪好未来教育科技有限公司 连线题批改方法、装置、电子设备及存储介质
CN113971805A (zh) * 2021-12-22 2022-01-25 深圳市迪博企业风险管理技术有限公司 一种结合机器视觉和语义分析的智能阅卷评分方法

Also Published As

Publication number Publication date
US20220101030A1 (en) 2022-03-31
US11450081B2 (en) 2022-09-20
CN109815932B (zh) 2021-05-28
CN109815932A (zh) 2019-05-28

Similar Documents

Publication Publication Date Title
WO2020156362A1 (zh) 一种试卷批改方法、装置、电子设备及存储介质
US11410407B2 (en) Method and device for generating collection of incorrectly-answered questions
CN109993112B (zh) 一种图片中表格的识别方法及装置
CN109670504B (zh) 一种手写答案识别批改方法及装置
US11790641B2 (en) Answer evaluation method, answer evaluation system, electronic device, and medium
CN109634961B (zh) 一种试卷样本生成方法、装置、电子设备和存储介质
US10902283B2 (en) Method and device for determining handwriting similarity
CN109712043B (zh) 一种答案批改方法及装置
CN110956138B (zh) 一种基于家教设备的辅助学习方法及家教设备
CN109918523B (zh) 一种基于yolo9000算法的电路板元器件检测方法
CN110647885B (zh) 基于图片识别的试卷拆分方法、装置、设备以及介质
CN113537801B (zh) 板书处理方法、装置、终端和存储介质
CN114463770A (zh) 一种用于普遍试卷题目的智能切题方法
US11749128B2 (en) Answer correction method and device
CN113569677A (zh) 一种基于扫描件的纸质试验报告生成方法
CN111666882A (zh) 一种手写体试题答案提取方法
WO2023024898A1 (zh) 题目辅助方法、题目辅助装置和题目辅助系统
CN111832550B (zh) 数据集制作方法、装置、电子设备及存储介质
CN114255465A (zh) 不规则医疗报告单图像中敏感信息脱敏方法及系统
CN112686253A (zh) 一种用于电子白板的屏幕文字提取系统及方法
CN110751140A (zh) 字符批量识别方法、装置和计算机设备
CN111507292B (zh) 手写板校正方法、装置、计算机设备以及存储介质
CN113128486B (zh) 手写数学公式样本库的构建方法、装置及终端设备
CN113065477B (zh) 一种智慧课堂课本内容识别方法
CN118053144A (zh) 批改系统的文本检测方法、装置、设备及可读存储介质

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: 20748752

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20748752

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 20748752

Country of ref document: EP

Kind code of ref document: A1

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 24.02.2022)

122 Ep: pct application non-entry in european phase

Ref document number: 20748752

Country of ref document: EP

Kind code of ref document: A1