CN108388895A - A kind of paper answering card automatic processing method based on machine learning - Google Patents
A kind of paper answering card automatic processing method based on machine learning Download PDFInfo
- Publication number
- CN108388895A CN108388895A CN201810176948.0A CN201810176948A CN108388895A CN 108388895 A CN108388895 A CN 108388895A CN 201810176948 A CN201810176948 A CN 201810176948A CN 108388895 A CN108388895 A CN 108388895A
- Authority
- CN
- China
- Prior art keywords
- paper
- answer
- answering card
- region
- score
- Prior art date
- Legal status (The legal status 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 status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/22—Image preprocessing by selection of a specific region containing or referencing a pattern; Locating or processing of specific regions to guide the detection or recognition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/50—Extraction of image or video features by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Data Mining & Analysis (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Electrically Operated Instructional Devices (AREA)
Abstract
The invention discloses a kind of paper answering card automatic processing method based on machine learning.Method is:Object detector is scribbled in training first;Then it determines paper format, generate paper and prints;Examiner and reader is commented to fill in corresponding region, examiner scribbles objective item answer region and subjective item answer region, comments reader to read and appraise subjective item and scribbles corresponding marking region;Then it is image by the scanning of the paper of completion, carries out image procossing and subregion cutting;Target discrimination is carried out using trained object detector of scribbling, score calculating is carried out then in conjunction with correct option;Statistics and analysis finally is carried out to paper:Include calculating the average mark of paper by the gross, the average mark of each topic;The distribution situation of paper by the gross and each topic is calculated;Trend analysis is carried out, and transfers answer part from database and teacher or individual is supplied to consult in person.The present invention is at low cost, precision is high, flexibility is good, and the situation that can answer to paper carries out efficient statistics and analysis.
Description
Technical field
The present invention relates to computer vision and machine learning techniques field, especially a kind of paper based on machine learning is answered
Inscribe card automatic processing method.
Background technology
Computer vision had been a hot spot of research in recent years and difficult point, and target detection is as high-rise visual task
Basis, it has also become vital in computer vision field to study a question.Computer vision is imitated by machine vision
The vision system of human eye, cognitive psychology and Neurobiology the study found that the mankind identify a specific object be assorted
There is a kind of ability of very strong perception object before.For the image of width complexity, the vision system of the mankind is opened one
Certain parts therein only can be paid close attention in the reaction time of beginning, and ignore the inapparent part of remaining in image.This is furtherly
It is bright before identifying a specific things, there are a simple vision noticing mechanism, the mechanism in the vision system of the mankind
For filtering out the most possibly region containing object.
Tradition uses limitation using the answering card of optical character reader (OCR) identification answer there are many, such as needs specific identification hard
Part, answering card be of high cost, paper format flexibility difference and need specific pencil to scribble and apply change after write error it is cumbersome etc..This
Outside, low to the statistics and analysis efficiency of testing situations there are clergy, electric marking is difficult to actually use in quiz
Problem.
Invention content
Present invention aims at provide it is a kind of it is at low cost, precision is high, the good paper answer based on machine learning of flexibility
Card automatic processing method, and efficient statistics and analysis is carried out to paper situation of answering.
Realize that the technical solution of the object of the invention is:A kind of paper answering card side of automatically processing based on machine learning
Method includes the following steps:
Object detector is scribbled in step 1, training;
Step 2 determines paper format, generates paper and prints;
Step 3, examiner and reader is commented to fill in corresponding region;
The paper completed in step 3 scanning is that image, then progress image procossing and subregion are cut by step 4;
Step 5, using step 1 it is trained scribble object detector carry out target discrimination, then in conjunction with correct option into
Row score calculates;
Step 6 carries out statistics and analysis to paper.
Further, object detector is scribbled in the training described in step 1, specific as follows:
1.1) the positive and negative sample set for scribbling target detection window is established, and the positive negative training sample in positive and negative sample set is contracted
It is put into 32*24 resolution ratio;
1.2) the HOG features of all positive negative training samples are extracted;
1.3) the HOG features of all positive negative training samples are input in Linear SVM and are trained, obtain grader.
Further, the positive and negative sample set of target detection window is scribbled in the foundation described in step 1.1, and by positive and negative sample set
In positive negative training sample zoom to 32*24 resolution ratio, it is specific as follows:
1.1.1) training positive sample is full of and scribbles 60% or more rectangular area on answering card and do not make apparent error painting
Write the sample of label;
1.1.2) training negative sample, i.e., scribble the sample of 60% or more rectangular area, have answer person on underfill answering card
Make the apparent sample for scribbling error flag and blank does not scribble the sample of answer.
Further, the determination paper format described in step 2 generates paper and prints, specific as follows:
2.1) prototype tool is used, the format of answering card is designed;
2.2) module location information of designed answering card is stored into positioning xml document;
2.3) it generates paper and prints.
Further, the examiner described in step 3 and reader is commented to fill in corresponding region, it is specific as follows:
3.1) examiner scribbles objective item answer region and subjective item answer region;
3.2) it comments reader to read and appraise subjective item, and scribbles corresponding marking region.
Further, described in step 4 by the paper completed in step 3 scanning be image, then carry out image procossing and
Subregion is cut, specific as follows:
4.1) scanning paper is image file, resolution ratio 3500*2500;
4.2) by image gray processing, using the rectangular detector based on opencv respectively in four apex angles of paper image
The rectangle anchor point of paper is detected in the region of 200*200 resolution ratio;
4.3) according to the angle information of anchor point, Slant Rectify is carried out to image;
4.4) read step 2.2) generate xml document in length and width, according to the location information of anchor point, to answering card
Content zooms in and out fine tuning;
4.5) read step 2.2) generate xml document in each module location information, each regions module is cut
It cuts, subjective item answer partial shrinkage is stored into database, and step 5 is sent into the answer part of objective item and the marking part of subjective item
It is handled.
Further, the trained object detector of scribbling of use step 1 described in step 5 carries out target discrimination, then
Score calculating is carried out in conjunction with correct option, it is specific as follows:
5.1) the answer part of the objective item obtained according to step 4, the grader pair trained using sliding window and step 1
Answer part carries out scribbling target detection;
It is defined as follows rule:
Continue be whether the mark subsequently judged, if 0, then the topic is scored at 0;If 1, then the topic needs
Continue subsequently to judge;
Count () is counting function,Indicate that the i-th problem confidence level is the testing result set of c;
Type () is the function of topic types, and single choice test items are indicated when being 0, and multiple-choice question is indicated when being 1;If inspection
It surveys result of the detection confidence level more than 90% in result and is more than 1, then for single choice test items, be then directly judged as mistake;
Multinomial choosing is then inscribed, then testing result is all as candidate answers;
5.2) the answer region region corresponding with the location information in xml of scribbling that step 5.1) obtains is subjected to IOU meters
It calculates, if scribbling answer region with the IOU of the band of position of some option more than 0.6, can get the specific of the painting writing position
Answer choice;
5.3) answer choice that step 5.2) obtains is compared with correct option, calculates and remember according to score rule
Record the topic score;
5.4) the marking part of the subjective item obtained according to step 4, the grader pair trained using sliding window and step 1
Marking part carries out scribbling target detection;If detecting result of the confidence level more than 90% in testing result is more than 1, confidence is taken
It spends highest as testing result;If testing result number is 0, this entitled 0 point;
5.5) the marking testing result for obtaining step 5.4) carries out IOU calculating with the location information in xml, when marking is examined
It surveys result and the IOU of the band of position of some score in xml is more than 0.6, then can get the specific score of the topic, record is somebody's turn to do
Point;
5.6) score of step 5.3) is added with the score of step 5.5), as the total score of this paper.
Further, statistics and analysis is carried out to paper described in step 6, it is specific as follows:
6.1) to the average mark of paper by the gross, the average mark of each topic is calculated;
6.2) distribution situation of paper by the gross and each topic is calculated;
6.3) to carry out different time according to the scoring event of previous paper intersegmental with interindividual trend analysis, from number
It is supplied to teacher or individual to consult in person according to answer part is transferred in library.
Compared with prior art, the present invention its remarkable advantage is:(1) specific identification hardware, at low cost, paper version are not needed
Formula flexibility is good, does not need specific pencil and scribbles, and is changed simply after applying write error;(2) paper can be completed to answer the statistics of situation
And analysis.
Description of the drawings
Fig. 1 is the flow diagram of the paper answering card automatic processing method the present invention is based on machine learning.
Fig. 2 is that object detector training flow chart is scribbled in the present invention.
Fig. 3 is to scribble object detector in the present invention to train positive negative sample schematic diagram.
Specific implementation mode
Paper answering card automatic processing method based on machine learning proposed by the invention, makes full use of computer skill
Art introduces image procossing and the target detection based on machine learning, realizes that use is realized with a low cost high-precision high flexibility
Paper processing, while paper can be completed and answered the statistics and simple analysis of situation.
The present invention will be further described below with reference to the drawings.
In conjunction with Fig. 1, realization of the invention includes following 6 steps:
Object detector is scribbled in step 1, training;
Step 2 determines paper format, generates paper and prints;
Step 3, examiner and reader is commented to fill in corresponding region;
The paper completed in step 3 scanning is that image, then progress image procossing and subregion are cut by step 4;
Step 5, using step 1 it is trained scribble object detector carry out target discrimination, then in conjunction with correct option into
Row score calculates;
Step 6 carries out statistics and analysis to paper.
Further, in conjunction with the flow chart of Fig. 2, object detector is scribbled in the training described in step 1, specific as follows:
1.1) the positive and negative sample set for scribbling target detection window is established, and the positive negative training sample in positive and negative sample set is contracted
It is put into 32*24 resolution ratio;
1.2) the HOG features of all positive negative training samples are extracted;
1.3) the HOG features of all positive negative training samples are input in Linear SVM and are trained, obtain grader.
The step is not limited to use HOG+SVM as feature and grader, if the requirement higher to accuracy of detection, can make
Use deep learning method as the basic methods of this step.
In conjunction with the sample schematic diagram of Fig. 3, the positive and negative sample set of target detection window is scribbled in the foundation described in step 1.1, and
Positive negative training sample in positive and negative sample set is zoomed into 32*24 resolution ratio, it is specific as follows:
1.1.1) training positive sample is full of and scribbles 60% or more rectangular area on answering card and do not make apparent error painting
Write the sample of label;If here as apparent error scribble label be intended to refer to a people filled be more than 60% area
Domain, but he feel to apply it is wrong, it can draw above a fork represent it is invalid, it is believed that this is not counting positive sample, such as Fig. 3
4th figure of negative sample.
1.1.2) training negative sample, i.e., scribble the sample of 60% or more rectangular area, have answer person on underfill answering card
Make the apparent sample for scribbling error flag and blank does not scribble the sample of answer.
Further, the determination paper format described in step 2 generates paper and prints, specific as follows:
2.1) prototype tool is used, the format of answering card is designed;
2.2) module location information of designed answering card is stored into positioning xml document;
2.3) it generates paper and prints.
Further, the examiner described in step 3 and reader is commented to fill in corresponding region, it is specific as follows:
3.1) examiner scribbles objective item answer region and subjective item answer region;
3.2) it comments reader to read and appraise subjective item, and scribbles corresponding marking region.
Further, described in step 4 by the paper completed in step 3 scanning be image, then carry out image procossing and
Subregion is cut, specific as follows:
4.1) scanning paper is image file, resolution ratio 3500*2500;
4.2) by image gray processing, using the rectangular detector based on opencv respectively in four apex angles of paper image
The rectangle anchor point of paper is detected in the region of 200*200 resolution ratio;
4.3) according to the angle information of anchor point, Slant Rectify is carried out to image;
4.4) read step 2.2) generate xml document in length and width, according to the location information of anchor point, to answering card
Content zooms in and out fine tuning;
4.5) read step 2.2) generate xml document in each module location information, each regions module is cut
It cuts, subjective item answer partial shrinkage is stored into database, and step 5 is sent into the answer part of objective item and the marking part of subjective item
It is handled.
Further, the trained object detector of scribbling of use step 1 described in step 5 carries out target discrimination, then
Score calculating is carried out in conjunction with correct option, it is specific as follows:
5.1) the answer part of the objective item obtained according to step 4, the grader pair trained using sliding window and step 1
Answer part carries out scribbling target detection;
It is defined as follows rule:
Continue be whether the mark subsequently judged, if 0, then the topic is scored at 0;If 1, then the topic needs
Continue subsequently to judge;Count () is counting function,Indicate that the i-th problem confidence level is the testing result set of c;
Type () is the function of topic types, and single choice test items are indicated when being 0, and multiple-choice question is indicated when being 1;If in testing result
It detects result of the confidence level more than 90% and is more than 1, then for single choice test items, be then directly judged as mistake;For multinomial
Choosing is then inscribed, then testing result is all as candidate answers;
5.2) the answer region region corresponding with the location information in xml of scribbling that step 5.1 obtains is subjected to IOU meters
It calculates, if scribbling answer region with the IOU of the band of position of some option more than 0.6, can get the specific of the painting writing position
Answer choice;
5.3) answer choice that step 5.2 obtains is compared with correct option, calculates and remember according to score rule
Record the topic score;
5.4) the marking part of the subjective item obtained according to step 4, the grader pair trained using sliding window and step 1
Marking part carries out scribbling target detection;If detecting result of the confidence level more than 90% in testing result is more than 1, confidence is taken
It spends highest as testing result;If testing result number is 0, this entitled 0 point;
5.5) the marking testing result for obtaining step 5.4 carries out IOU calculating with the location information in xml, when marking is examined
It surveys result and the IOU of the band of position of some score in xml is more than 0.6, then can get the specific score of the topic, record is somebody's turn to do
Point;
5.6) score of step 5.3 is added with the score of step 5.5, as the total score of this paper.
Further, statistics and analysis is carried out to paper described in step 6, it is specific as follows:
6.1) to the average mark of paper by the gross, the average mark of each topic is calculated;
6.2) distribution situation of paper by the gross and each topic is calculated;
6.3) to carry out different time according to the scoring event of previous paper intersegmental with interindividual trend analysis, for
Special score transfers its answer part from database and teacher or individual is supplied to consult in person.
In conclusion the paper answering card automatic processing method provided by the invention based on machine learning, makes full use of meter
Calculation machine technology introduces image procossing and the target detection based on machine learning, using common paper paper as answering card, uses
Scanner converts paper to image, and then paper is corrected and cut using image processing techniques, further to use
Target detection technique carries out the score judgement of pinpoint accuracy, final output paper answer situation and further statistics and point
Analysis.Using method proposed by the present invention, the paper for being realized with a low cost high-precision high flexibility can be used to handle, while can complete
Paper is answered the statistics and simple analysis of situation.
Claims (8)
1. a kind of paper answering card automatic processing method based on machine learning, which is characterized in that include the following steps:
Object detector is scribbled in step 1, training;
Step 2 determines paper format, generates paper and prints;
Step 3, examiner and reader is commented to fill in corresponding region;
The paper completed in step 3 scanning is that image, then progress image procossing and subregion are cut by step 4;
Step 5 carries out target discrimination using the trained object detector of scribbling of step 1, is divided then in conjunction with correct option
Number calculates;
Step 6 carries out statistics and analysis to paper.
2. the paper answering card automatic processing method according to claim 1 based on machine learning, which is characterized in that step
Object detector is scribbled in training described in 1, specific as follows:
1.1) the positive and negative sample set for scribbling target detection window is established, and the positive negative training sample in positive and negative sample set is zoomed to
32*24 resolution ratio;
1.2) the HOG features of all positive negative training samples are extracted;
1.3) the HOG features of all positive negative training samples are input in Linear SVM and are trained, obtain grader.
3. the paper answering card automatic processing method according to claim 2 based on machine learning, which is characterized in that step
The positive and negative sample set of target detection window is scribbled in foundation described in 1.1, and the positive negative training sample in positive and negative sample set is scaled
It is specific as follows to 32*24 resolution ratio:
1.1.1) training positive sample is full of and scribbles 60% or more rectangular area on answering card and do not scribble mark as apparent error
The sample of note;
1.1.2) training negative sample, i.e., scribble that the sample of 60% or more rectangular area, to have answer person to make bright on underfill answering card
The aobvious sample for scribbling error flag and blank do not scribble the sample of answer.
4. the paper answering card automatic processing method according to claim 1 based on machine learning, which is characterized in that step
Determination paper format described in 2 generates paper and prints, specific as follows:
2.1) prototype tool is used, the format of answering card is designed;
2.2) module location information of designed answering card is stored into positioning xml document;
2.3) it generates paper and prints.
5. the paper answering card automatic processing method according to claim 1 based on machine learning, which is characterized in that step
Examiner described in 3 and reader is commented to fill in corresponding region, it is specific as follows:
3.1) examiner scribbles objective item answer region and subjective item answer region;
3.2) it comments reader to read and appraise subjective item, and scribbles corresponding marking region.
6. the paper answering card automatic processing method according to claim 4 based on machine learning, which is characterized in that step
Described in 4 is image by the paper completed in step 3 scanning, then carries out image procossing and subregion cutting, specific as follows:
4.1) scanning paper is image file, resolution ratio 3500*2500;
4.2) by image gray processing, using the rectangular detector based on opencv respectively in the 200* of four apex angles of paper image
The rectangle anchor point of paper is detected in the region of 200 resolution ratio;
4.3) according to the angle information of anchor point, Slant Rectify is carried out to image;
4.4) read step 2.2) generate xml document in length and width, according to the location information of anchor point, to answering card content
Zoom in and out fine tuning;
4.5) read step 2.2) each module in the xml document that generates location information, each regions module is cut, it is main
Sight topic answer partial shrinkage is stored into database, and the answer part of objective item and the marking part of subjective item are sent into step 5 and are carried out
Processing.
7. the paper answering card automatic processing method according to claim 6 based on machine learning, which is characterized in that step
The trained object detector of scribbling of use step 1 described in 5 carries out target discrimination, and score meter is carried out then in conjunction with correct option
It calculates, it is specific as follows:
5.1) the answer part of the objective item obtained according to step 4, the grader trained using sliding window and step 1 is to answer
Part carries out scribbling target detection;
It is defined as follows rule:
Continue be whether the mark subsequently judged, if 0, then the topic is scored at 0;If 1, then the topic needs to continue
Subsequently judged;
Count () is counting function,Indicate that the i-th problem confidence level is the testing result set of c;
Type () is the function of topic types, and single choice test items are indicated when being 0, and multiple-choice question is indicated when being 1;If detection knot
Result of the confidence level more than 90% is detected in fruit and is more than 1, then for single choice test items, is then directly judged as mistake;For
Multinomial choosing is then inscribed, then testing result is all as candidate answers;
5.2) the answer region region corresponding with the location information in xml of scribbling that step 5.1) obtains is subjected to IOU calculating, such as
Fruit scribbles answer region and the IOU of the band of position of some option is more than 0.6, then can get the specific answer choosing of the painting writing position
;
5.3) answer choice that step 5.2) obtains is compared with correct option, is calculated according to score rule and records this
Inscribe score;
5.4) the marking part of the subjective item obtained according to step 4, the grader trained using sliding window and step 1 is to marking
Part carries out scribbling target detection;If detecting result of the confidence level more than 90% in testing result is more than 1, confidence level is taken most
High conduct testing result;If testing result number is 0, this entitled 0 point;
5.5) the marking testing result for obtaining step 5.4) carries out IOU calculating with the location information in xml, when marking detection knot
Fruit and the IOU of the band of position of some score in xml are more than 0.6, then can get the specific score of the topic, record the score;
5.6) score of step 5.3) is added with the score of step 5.5), as the total score of this paper.
8. being based on the paper answering card automatic processing method of machine learning according to claim 7, which is characterized in that step 6 institute
That states carries out statistics and analysis to paper, specific as follows:
6.1) to the average mark of paper by the gross, the average mark of each topic is calculated;
6.2) distribution situation of paper by the gross and each topic is calculated;
6.3) to carry out different time according to the scoring event of previous paper intersegmental with interindividual trend analysis, from database
In transfer answer part be supplied to teacher or individual I consult.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810176948.0A CN108388895B (en) | 2018-03-04 | 2018-03-04 | Machine learning-based automatic processing method for test paper answer sheet |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810176948.0A CN108388895B (en) | 2018-03-04 | 2018-03-04 | Machine learning-based automatic processing method for test paper answer sheet |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108388895A true CN108388895A (en) | 2018-08-10 |
CN108388895B CN108388895B (en) | 2022-02-18 |
Family
ID=63069770
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810176948.0A Active CN108388895B (en) | 2018-03-04 | 2018-03-04 | Machine learning-based automatic processing method for test paper answer sheet |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108388895B (en) |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109446483A (en) * | 2018-09-30 | 2019-03-08 | 大连海事大学 | A kind of machine of the objective item to contain subjective information sentences volume method |
CN109684980A (en) * | 2018-09-19 | 2019-04-26 | 腾讯科技(深圳)有限公司 | Automatic marking method and device |
CN109886923A (en) * | 2019-01-17 | 2019-06-14 | 柳州康云互联科技有限公司 | It is a kind of for internet detection in measurement detection system and method based on machine learning |
CN110163211A (en) * | 2018-09-06 | 2019-08-23 | 腾讯科技(深圳)有限公司 | A kind of image-recognizing method, device and storage medium |
CN110348400A (en) * | 2019-07-15 | 2019-10-18 | 京东方科技集团股份有限公司 | A kind of scoring acquisition methods, device and electronic equipment |
CN110413789A (en) * | 2019-07-31 | 2019-11-05 | 广西师范大学 | A kind of exercise automatic classification method based on SVM |
CN113657354A (en) * | 2021-10-19 | 2021-11-16 | 深圳市菁优智慧教育股份有限公司 | Answer sheet identification method and system based on deep learning |
CN116798036A (en) * | 2023-06-27 | 2023-09-22 | 广州市南方人力资源评价中心有限公司 | Method and device for identifying and checking answer sheet objective question identification result |
CN117237585A (en) * | 2023-11-10 | 2023-12-15 | 山东科技大学 | Optical mark positioning and identifying method, system, equipment and storage medium for answer sheet |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104376315A (en) * | 2014-07-30 | 2015-02-25 | 天津科技大学 | Detection method based on computer image processing and mode recognition and application of detection method |
CN107729936A (en) * | 2017-10-12 | 2018-02-23 | 科大讯飞股份有限公司 | One kind corrects mistakes to inscribe reads and appraises method and system automatically |
-
2018
- 2018-03-04 CN CN201810176948.0A patent/CN108388895B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104376315A (en) * | 2014-07-30 | 2015-02-25 | 天津科技大学 | Detection method based on computer image processing and mode recognition and application of detection method |
CN107729936A (en) * | 2017-10-12 | 2018-02-23 | 科大讯飞股份有限公司 | One kind corrects mistakes to inscribe reads and appraises method and system automatically |
Non-Patent Citations (1)
Title |
---|
张站等: "基于图像识别的阅卷系统的设计与实现", 《微型机与应用》 * |
Cited By (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110163211B (en) * | 2018-09-06 | 2023-02-28 | 腾讯科技(深圳)有限公司 | Image recognition method, device and storage medium |
CN110163211A (en) * | 2018-09-06 | 2019-08-23 | 腾讯科技(深圳)有限公司 | A kind of image-recognizing method, device and storage medium |
CN109684980B (en) * | 2018-09-19 | 2022-12-13 | 腾讯科技(深圳)有限公司 | Automatic scoring method and device |
CN109684980A (en) * | 2018-09-19 | 2019-04-26 | 腾讯科技(深圳)有限公司 | Automatic marking method and device |
CN109446483A (en) * | 2018-09-30 | 2019-03-08 | 大连海事大学 | A kind of machine of the objective item to contain subjective information sentences volume method |
CN109446483B (en) * | 2018-09-30 | 2022-09-30 | 大连海事大学 | Machine appraisal method for objective questions containing subjective information |
CN109886923A (en) * | 2019-01-17 | 2019-06-14 | 柳州康云互联科技有限公司 | It is a kind of for internet detection in measurement detection system and method based on machine learning |
CN110348400A (en) * | 2019-07-15 | 2019-10-18 | 京东方科技集团股份有限公司 | A kind of scoring acquisition methods, device and electronic equipment |
CN110413789A (en) * | 2019-07-31 | 2019-11-05 | 广西师范大学 | A kind of exercise automatic classification method based on SVM |
CN113657354A (en) * | 2021-10-19 | 2021-11-16 | 深圳市菁优智慧教育股份有限公司 | Answer sheet identification method and system based on deep learning |
CN116798036A (en) * | 2023-06-27 | 2023-09-22 | 广州市南方人力资源评价中心有限公司 | Method and device for identifying and checking answer sheet objective question identification result |
CN116798036B (en) * | 2023-06-27 | 2024-04-02 | 网才科技(广州)集团股份有限公司 | Method and device for identifying and checking answer sheet objective question identification result |
CN117237585A (en) * | 2023-11-10 | 2023-12-15 | 山东科技大学 | Optical mark positioning and identifying method, system, equipment and storage medium for answer sheet |
CN117237585B (en) * | 2023-11-10 | 2024-01-30 | 山东科技大学 | Optical mark positioning and identifying method, system, equipment and storage medium for answer sheet |
Also Published As
Publication number | Publication date |
---|---|
CN108388895B (en) | 2022-02-18 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108388895A (en) | A kind of paper answering card automatic processing method based on machine learning | |
CN110008933B (en) | Universal intelligent marking system and method | |
CN110348400B (en) | Score obtaining method and device and electronic equipment | |
CN202939608U (en) | Test paper inspecting system | |
CN109697905A (en) | A kind of exam paper marking system | |
CN110322379A (en) | Paper processing method and paper processing system | |
CN103886302A (en) | Test paper identification analysis achievement method and device | |
CN106599887B (en) | Paper answer region recognition and examination question read and appraise method, system | |
CN107784264A (en) | Test paper analysis method and system based on image procossing | |
CN111091538A (en) | Method and device for automatically identifying and detecting pipeline welding seam and defect | |
CN108537218A (en) | A kind of identifying processing method and device of answering card | |
CN109636089A (en) | A kind of method for student synthetic quality system | |
CN109242856A (en) | A kind of papery operation electronization reads and makes comments method and device | |
CN112506360A (en) | Intelligent identification and correction method for conventional paper writing operation or test paper | |
CN108876195A (en) | A kind of intelligentized teachers ' teaching quality evaluating system | |
CN109034590A (en) | A kind of intelligentized teaching quality evaluation for teachers management system | |
CN111444389A (en) | Conference video analysis method and system based on target detection | |
CN108428073A (en) | A kind of intelligent evaluation system for teachers ' teaching quality | |
CN113888757A (en) | Examination paper intelligent analysis method, examination paper intelligent analysis system and storage medium based on benchmarking evaluation | |
CN108681713A (en) | A kind of system for teaching quality evaluation for teachers | |
JP2004029107A (en) | System for processing marking of answer | |
CN111666882A (en) | Method for extracting answers of handwritten test questions | |
CN110956173A (en) | Topic content identification method and device, readable storage medium and computer equipment | |
KR100795951B1 (en) | System for grading examination paper and control method | |
CN116128681A (en) | Intelligent scoring system for oxygen preparation experiment of middle school |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |