CN109241869A - The recognition methods of answering card score, device and terminal device - Google Patents
The recognition methods of answering card score, device and terminal device Download PDFInfo
- Publication number
- CN109241869A CN109241869A CN201810934893.5A CN201810934893A CN109241869A CN 109241869 A CN109241869 A CN 109241869A CN 201810934893 A CN201810934893 A CN 201810934893A CN 109241869 A CN109241869 A CN 109241869A
- Authority
- CN
- China
- Prior art keywords
- answer sheet
- image information
- score
- answer
- neural network
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 60
- 238000013528 artificial neural network Methods 0.000 claims abstract description 21
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 16
- 238000012549 training Methods 0.000 claims description 55
- 230000006870 function Effects 0.000 claims description 30
- 238000012545 processing Methods 0.000 claims description 27
- 238000004590 computer program Methods 0.000 claims description 21
- 238000003062 neural network model Methods 0.000 claims description 17
- 238000007781 pre-processing Methods 0.000 claims description 13
- 230000004913 activation Effects 0.000 claims description 11
- 238000004364 calculation method Methods 0.000 claims description 9
- 210000002569 neuron Anatomy 0.000 claims description 9
- 238000013139 quantization Methods 0.000 claims description 8
- 238000012937 correction Methods 0.000 claims description 7
- 238000012360 testing method Methods 0.000 claims description 7
- 238000011176 pooling Methods 0.000 claims description 6
- 238000001514 detection method Methods 0.000 claims description 4
- 230000005477 standard model Effects 0.000 claims description 3
- 238000003709 image segmentation Methods 0.000 claims description 2
- 238000012015 optical character recognition Methods 0.000 abstract 2
- 230000008569 process Effects 0.000 description 16
- 238000010586 diagram Methods 0.000 description 5
- 230000003287 optical effect Effects 0.000 description 5
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 238000011156 evaluation Methods 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 230000000644 propagated effect Effects 0.000 description 2
- 241001672694 Citrus reticulata Species 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000008676 import Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000035772 mutation Effects 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
- 230000032258 transport Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/40—Document-oriented image-based pattern recognition
- G06V30/41—Analysis of document content
- G06V30/413—Classification of content, e.g. text, photographs or tables
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- 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
- G06V10/24—Aligning, centring, orientation detection or correction of the image
- G06V10/243—Aligning, centring, orientation detection or correction of the image by compensating for image skew or non-uniform image deformations
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Character Discrimination (AREA)
Abstract
The present invention provides a kind of answering card score recognition methods, device and terminal devices, comprising: obtains answer card image information；Pre-process the answer card image information；Identification and determining answering card score are carried out to by the pretreated answer card image information by AlexNet neural network algorithm.The recognition methods of answering card score, device and terminal device provided in an embodiment of the present invention, it is identified by using image information of the AlexNet neural network algorithm to acquisition to determine answering card score, it can reduce the requirement of answer card filling quality and the quality requirement to Image Acquisition, without purchasing and carrying out image procossing using the optical character reader (OCR) of high-end valuableness, and it can quickly obtain reading and appraising efficiency to answering card as a result, improving.
Description
Technical Field
The invention belongs to the technical field of information processing, and particularly relates to an answer sheet score identification method, an answer sheet score identification device and terminal equipment.
Background
In large-scale examinations such as college entrance examination, national english level four and six examinations, and official examinations, the answer sheet is used in large quantities for evaluating standard objective questions. In the process of evaluating, a large amount of cursor readers are used for reading cards at present, but with the improvement of teaching conditions of primary and middle schools, examination in the middle and end period of the primary and middle schools, the Mandarin level test and examination of various grades also adopt the mode of answer sheets step by step, for the small mechanisms, the use and maintenance cost of the cursor readers is too high, and the answer sheet evaluating efficiency is low.
Disclosure of Invention
In view of this, the embodiments of the present invention provide an answer sheet score identification method, an answer sheet score identification device, and a terminal device, which can solve the problems that when an optical mark reader is used for evaluating an answer sheet in the prior art, the use and maintenance costs of the optical mark reader are too high, and the answer sheet evaluation efficiency is low.
A first aspect of an embodiment of the present invention provides an answer sheet score identification method, including:
acquiring image information of an answer sheet;
preprocessing the image information of the answer sheet;
and identifying the preprocessed image information of the answer sheet through an AlexNet neural network algorithm and determining the score of the answer sheet.
A second aspect of the embodiments of the present invention provides an answer sheet score identification apparatus, including:
the acquisition module is used for acquiring image information of the answer sheet;
the preprocessing module is used for preprocessing the image information of the answer sheet;
and the identification module is used for identifying the preprocessed image information of the answer sheet through an AlexNet neural network algorithm and judging the score of the answer sheet.
A third aspect of the embodiments of the present invention provides a terminal device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, where the processor implements the steps of the method according to any one of the first aspect when executing the program.
A fourth aspect of embodiments of the present invention provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of the method of any one of the first aspect.
The answer sheet score identification method, the answer sheet score identification device and the terminal equipment provided by the embodiment of the invention have the beneficial effects that: the AlexNet neural network algorithm is adopted to identify the collected image information to determine the score of the answer sheet, so that the requirements on the filling quality of the answer sheet and the quality of image collection can be reduced, a high-end and expensive cursor reader is not required to be purchased and used for image processing, the result can be obtained quickly, and the evaluation efficiency of the answer sheet is improved.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the embodiments or the prior art descriptions will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without inventive exercise.
Fig. 1 is a flowchart of an answer sheet score identification method according to an embodiment of the present invention;
fig. 2 is a flowchart of an answer sheet score identification method according to another embodiment of the present invention;
fig. 3 is a flowchart illustrating an answer sheet score identification method according to yet another embodiment of the present invention;
fig. 4 is a schematic diagram of an answer sheet score identification device according to an embodiment of the present invention;
fig. 5 is a schematic diagram of a terminal device provided by an embodiment of the present invention;
fig. 6 is a schematic diagram of an answer sheet template according to an embodiment of the present invention.
Detailed Description
In order to make the technical problems, technical solutions and advantageous effects to be solved by the present invention more clearly apparent, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
The specific embodiment is as follows:
fig. 1 is a flowchart of an answer sheet score identification method according to an embodiment of the present invention. As shown in fig. 1, the method in this embodiment may include:
step 101, obtaining image information of an answer sheet.
The answer sheet image information comprises guest question answer information and/or subjective question answer information in the answered answer sheet.
Optionally, the answer sheet that has answered can be shot through a camera of the terminal device to obtain answer sheet image information, and the answer sheet that has answered can be scanned through a scanner to obtain answer sheet image information, and the terminal device can be a smart phone, a tablet computer, a computer or other devices.
And 102, preprocessing the image information of the answer sheet.
The image information of the answer sheet is subjected to preliminary processing of size, pixel, format and the like, and files from various sources are processed into a standard format acceptable to a program. Optionally, operations such as inclination correction, contrast adjustment, area division and the like are performed on the answer sheet image information.
And 103, identifying the preprocessed image information of the answer sheet through an AlexNet neural network algorithm and determining the score of the answer sheet.
Specifically, an AlexNet neural network model is established by taking a standard answer of an answer sheet as a standard model, the number of layers of a hidden layer of the AlexNet neural network model and the classification number of identification are determined, the AlexNet neural network model is subjected to self-learning training, the training times are not less than 10 thousands of times, the passing rate of a test set reaches more than 99.8%, a training model is generated, and preprocessed image information of the answer sheet is input into the training model to judge the answer sheet to obtain the score of the answer sheet.
Optionally, the input and output of the training model can be retrained and modified according to the type of the answer sheet. For example, for an answer sheet only including an objective question, a training model can be obtained by training only using a training data set including printed numbers and printed characters, so that if the training model is directly used for identifying the answer sheet including the objective question and a subjective question, the identification effect is affected, so that the answer sheet for the objective question and the subjective question needs to be trained on an AlexNet neural network model by using the training data set including handwritten characters, and the scale of the hidden layer number also needs to be adjusted.
According to the answer sheet score identification method provided by the embodiment, the acquired image information is identified by adopting the AlexNet neural network algorithm to determine the answer sheet score, so that the requirements on the filling quality of the answer sheet and the quality requirement on image acquisition can be reduced, the purchase and the use of a high-end and expensive optical mark reader for image processing are not needed, the result can be quickly obtained, and the evaluation efficiency of the answer sheet is improved.
Fig. 2 is a flowchart of an answer sheet score identification method according to another embodiment of the present invention. As shown in fig. 2, the method in this embodiment may include:
step 201, obtaining image information of the answer sheet.
Step 202, preprocessing the image information of the answer sheet.
In this embodiment, steps 201 to 202 are similar to steps 101 to 102 in the above embodiment, and are not described again here.
Step 203, constructing an AlexNet neural network model, which specifically comprises the following steps:
and determining the number of layers of the hidden layer and the number of identified classifications, and establishing an activation function, a residual calculation formula and a gradient descent function to obtain the AlexNet neural network model.
Specifically, the depth of the alexnet neural network is 8 layers, wherein the alexnet neural network comprises 5 convolutional layers and 3 full-connection layers, and the number of identifiable classes is 1000. The recognized classification number refers to the kind of number or letter that can be recognized, for example: for number identification, from zero to nine, 10 classifications may be included; for the recognition of english letters, 26 classes from a to Z may be included.
Optionally, a ReLu activation function may be used to perform the linear fitting operation, where the ReLu activation function is: f (z) relu (z) max (0, z), where f (z) is the output of the neuron.
Further, rewriting the ReLu activation function into a vector form to obtain the activation function,
the activation function is:
wherein, aiIs the output of a neuron, wiIs a probability vector, x is the transpose of the input dimension vector, biIs the weight proportion of the initial value;
the residual calculation formula is as follows:
wherein, delta(k-2)In order to be the value of the residual error,is the weight proportion, f' (V), of the current layer node(l-2)) A difference value calculated for the forward direction;
the gradient descent function is:
wherein, wijAnd b is the weight proportion of the current layer node, L (w, b) is a loss function, and a is the output of the neuron.
Step 204, training the AlexNet neural network model through a training data set to generate a training model, specifically comprising:
step 2041, a training data set is constructed.
Specifically, for an objective answer sheet including only objective questions, the objective answer sheet includes printed numbers and printed English letters. The subjective answer sheet including objective questions and subjective questions includes printed numbers, printed English letters, handwritten English letters and punctuation marks. Therefore, the training data set adopted by the AlexNet neural network model established for the objective answer sheet during model training is different from the training data set adopted by the AlexNet neural network model established for the subjective answer sheet during model training in sample capacity.
For the identification of numbers in an answer sheet, in order to eliminate unrecognizable factors caused by stained or blurred answer sheets, a handwritten character dataset MNIST (Mixed National Institute of Standards and technology database, simple machine vision dataset) is used to identify the printed letters. In order to improve the identification accuracy, the MNIST can be expanded by adopting a character set with five fonts of Song, imitation Song, Arial, Times New Roman and Georgia.
For character recognition in the subjective answer sheet, an englishh data set (English alphabet handwriting data set) is used, which includes 60000 character manual writing modes.
The image data is read by an image binary reader of TensorFlow, and the data is converted into a tensor format. TensorFlow is a second generation artificial intelligence learning system developed by Google, and the naming of TensorFlow is derived from the operation principle of TensorFlow. Tensor means an N-dimensional array, Flow means computation based on a dataflow graph, and TensorFlow is a computation process in which tensors Flow from one end of the Flow graph to the other. TensorFlow is a system that transports complex data structures into artificial intelligent neural networks for analysis and processing.
Step 2042, inputting a sample of the training data set, and training the standard model according to the activation function, the residual calculation formula and the gradient descent function to generate a training model.
And the output result of each hidden layer and the quantization threshold value of the expected value are measured according to a mean square error formula.
The model training process comprises the following steps: forward propagation operations and backward propagation operations;
in the forward propagation operation, the samples of the training data set are firstly transmitted to the nodes of the hidden layer, after the neuron activation function operation, the output result is output through the output nodes, and in the propagation process, the state of the neuron of each layer only affects the neuron of the next layer. And calculating the residual between the actual output result and the expected value according to a residual calculation formula, and if the actual output result does not meet the output requirement, turning to reverse propagation.
In the back propagation operation, a feedback path of the residual value is calculated according to the gradient descent function, so that the residual value is reversely propagated along the optimal fastest path, the weight proportion of each layer of neuron is modified, and the residual value is gradually propagated to the input layer for calculation.
Through repeated circulation of two operation processes of forward propagation operation and backward propagation operation, the weight proportion of each node is continuously corrected, so that the residual value is gradually reduced, and the output requirement is met. The training model can be generated.
In addition, it is also possible to set a training time, and stop the model training when the set training time is reached, or set a classification number, and stop the model training when the model training reaches the set classification number.
In the scheme, the fastest path of residual value back propagation is set by adopting a gradient descent function, so that the utilization rate of the CPU can be improved, and the time consumption of training is reduced.
When the AlexNet neural network model is trained, the output result of each hidden layer and the quantization threshold value of an expected value are measured in a mean square error mode, if the quantization threshold value obtained according to the mean square error formula cannot meet a preset condition, model training is stopped, and the AlexNet neural network model needs to be adjusted or a training data set needs to be adjusted and then trained again. The preset condition may be whether the quantization threshold curve has a mutation.
The mean square error formula is:
where MSE is the quantization threshold, yiExpected value of the current hidden layer, f (x)i) Is the output result of the current hidden layer.
And step 205, inputting the preprocessed image information of the answer sheet into the training model for recognition, and determining the score of the answer sheet according to the recognition result.
Inputting the preprocessed image information of the answer sheet into the training model for recognition, and outputting a classification recognition result. Optionally, a standard answer profile is started and compared with the recognition result, and the weight after comparison corresponds to the score of the examinee. The standard answer configuration file may be a JSON format file, and may be configured and generated by a user through a UI (user interface) tool. Further, the configuration file of the standard answer supports the import and export of an XML (eXtensible Markup Language) format, and provides good support for cross-platform operation of the system.
And step 2051, inputting the preprocessed image information of the answer sheet into the training model.
And step 2052, performing convolution operation on the preprocessed image information of the answer sheet through a convolution layer.
Specifically, 16 convolution kernels such as 5 × 5 to 21 × 21 are adopted to sequentially perform convolution operation on the preprocessed answer card image information through a convolution layer in the order of increasing dimensionality, and effective image features are extracted. The alexnet neural network comprises 5 convolutional layers, each layer having 16 convolutional operations.
And step 2053, performing pooling on the data subjected to the convolution operation through a pooling layer.
Optionally, the convolved data is pooled in a size of dimension 8 × 8, where the pooled data includes 3 channels of data, and the data dimension of each channel is 1024.
Preferably, the pooling operation may use a maximum pooling method, which can avoid the phenomenon of using stuck due to the occupation of the mobile terminal memory and the CPU utilization.
And step 2054, performing full connection operation on the pooled data through an output layer to generate an identification result.
And step 2055, residual error detection is carried out on the identification result, and the training model is updated according to the residual error detection result.
Specifically, residual values are calculated according to a residual calculation formula, and the weight proportion of each hidden layer node is updated according to the residual values.
And step 2056, comparing the standard answers with the identification results, and obtaining the scores of the answer sheet according to the weights obtained by comparison.
And the identification result comprises identification answer information of each question obtained by identifying the preprocessed image information of the answer sheet through a training model, the identification answer information is stored in a first database, and the first database is compared with a second database in which standard answer information is stored to obtain the score of the answer sheet.
Optionally, the embodiment may employ CUDA (computer Unified Device Architecture) to accelerate the computation of the AlexNet neural network, which can reduce a large number of matrix operations during model training.
In the scheme, the fastest path of residual value back propagation is set by adopting a gradient descent function, so that the utilization rate of the CPU can be improved, and the time consumption of training is reduced. By monitoring the quantization threshold value by using a mean square error method, whether the AlexNet neural network model has a major problem or not can be found in time, the model training is terminated as soon as possible to carry out corresponding adjustment, and the time for model training is saved.
Fig. 3 is a flowchart of an answer sheet score identification method according to another embodiment of the present invention. As shown in fig. 2, the answer sheet score identification method in this embodiment may include:
step 301, modifying the answer template according to the test paper information and generating an answer sheet, wherein the answer sheet comprises a positioning area used as an image segmentation reference.
Optionally, the terminal device for implementing the answer sheet score identification method provided by the embodiment of the present invention includes an answer sheet electronic generation tool, an answer sheet template is built in the answer sheet electronic generation tool, the answer sheet template may be written in a ladachy language, and the answer sheet template may be customized and modified by a user according to test paper information to generate a new answer sheet template.
The answer sheet template can generate a PDF (Portable Document Format) answer sheet through PDF (Portable Document Format) software, a user can print or copy the PDF answer sheet to form a paper answer sheet for an examination, and the paper type and the pen for the examination are not limited.
As shown in fig. 6, the answer sheet template may include a type area 61, a location area 62, an objective area 63 and a subjective area 64. The type field may include a 12-digit number input field, the number of digits of which may be modified according to the user's needs. The objective area 63 is used for filling in answers of objective questions, and the number of the questions can be modified according to the test paper information. The subjective area 64 is used for inputting letters and English characters, and is used for questions such as English complete shape filling or translation, and the number of the questions in the input area can be modified according to the test paper information. The rightmost positioning area 40 includes a plurality of black squares, i.e., positioning references, each of the black squares corresponds to a row of numbers to be recognized, and when the image information is subjected to area division, the area division is performed according to the positions of the black squares.
The answer sheet score is judged by setting the positioning area to position the answer filling area and judging whether the option is covered by an AlexNet neural network algorithm.
And step 302, acquiring image information of the answer sheet.
In this embodiment, step 302 is similar to step 101 in the above embodiment, and is not described here again.
Step 303, performing a first process on the image information of the answer sheet, where the first process includes at least one of a color balance process, a contrast adjustment process, a tilt correction process, and an automatic effective area delineation process.
And 304, if the answer sheet image information after the first processing is a multi-channel image, performing area division on the answer sheet image information after the first processing according to the color information of the answer sheet image information after the first processing to obtain preprocessed answer sheet image information.
If the answer sheet image information after the first processing is a grayscale image, performing area division on the answer sheet image information after the first processing according to a positioning reference included in the positioning area in the answer sheet image information after the first processing to obtain preprocessed answer sheet image information, which specifically includes: acquiring position information of a positioning reference and dividing areas according to the position information; and extracting the outline of each divided region, and correcting the position of each divided region. The position correction may include inclination correction, deformation correction, overlay correction, and the like.
Optionally, for the subjective answer sheet, after the first processed image information of the answer sheet is subjected to region division, it is further determined whether the content of the subjective question part written on the answer sheet by the examinee is in the effective region, that is, whether the content exceeds the region corresponding to the empty ruled line and whether the content is filled with multiple lines, and if the content exceeds the empty ruled line or is filled with multiple lines, it is determined that the processed subjective answer sheet is an abnormal answer sheet, and the abnormal answer sheet is detected without performing an identification operation.
And 305, identifying the preprocessed answer sheet image information through an AlexNet neural network algorithm and determining the score of the answer sheet.
In this embodiment, step 305 is similar to steps 203 to 205 in the above embodiments, and is not described herein again.
Step 306, storing an original answer sheet image and an identified answer sheet image, and manually checking and comparing the original answer sheet image and the identified answer sheet image to verify the correctness of the answer score; the method comprises the steps of storing the scores of the answer sheets, conducting index sorting on the scores of the answer sheets to obtain the ranks of respondents corresponding to the scores of the answer sheets and the historical records of each examination, and drawing and analyzing the score curves of the respondents according to the historical records.
Optionally, the original image and the identification analysis image may be saved by SQLite to facilitate manual spot check and comparison. And storing the identified result and score, and performing index sequencing to realize functions of ranking, history recording and the like of the user. SQLite is an in-process library that enables a self-sufficient, serverless, zero-configuration, transactional SQL database engine. It is a zero-configuration database and does not need to be configured in the system.
In the embodiment, the grading judgment of the answer sheet is carried out by adopting the full-mobile terminal, the AlexNet neural network artificial intelligence technology is introduced into the image identification field of the answer sheet, the cost can be saved, the use convenience is improved, and the problems that special paper and special pens are required to be adopted when an optical mark reader is used for identifying the answer sheet and slight fouling and creasing of the answer sheet cause the problem that the answer sheet cannot be identified or is mistakenly identified are solved.
Fig. 4 is a schematic diagram of an answer sheet score identification device according to another embodiment of the present invention. As shown in fig. 4, the answer sheet score identification device in this embodiment may include:
an obtaining module 401, configured to obtain image information of an answer sheet;
a preprocessing module 402, configured to preprocess the image information of the answer sheet;
and the identification module 403 is configured to identify the preprocessed image information of the answer sheet through an AlexNet neural network algorithm and determine the score of the answer sheet.
The answer sheet score identification device in this embodiment may be configured to execute the answer sheet score identification method in any one of the first to third embodiments, and specific implementation principles may be found in the first to third embodiments, which are not described herein again.
EXAMPLE five
Fig. 5 is a schematic diagram of a terminal device according to a fifth embodiment of the present invention. As shown in fig. 5, the terminal device 5 of this embodiment includes: a processor 50, a memory 51 and a computer program 52 stored in said memory 51 and executable on said processor 50. The processor 50 executes the computer program 52 to implement the steps in the embodiment with the terminal device as the execution main body, such as the steps 101 to 104 shown in fig. 1. Alternatively, the processor 50, when executing the computer program 52, implements the functions of each module/unit in the fourth embodiment, for example, the functions of the modules 401 to 402 shown in fig. 4.
Illustratively, the computer program 52 may be partitioned into one or more modules/units that are stored in the memory 51 and executed by the processor 50 to implement the present invention. The one or more modules/units may be a series of computer program instruction segments capable of performing specific functions, which are used to describe the execution process of the computer program 52 in the terminal device 5. For example, the computer program 52 may be divided into an acquisition module, a generation module, a superposition module, and a processing module, and each module specifically functions as follows:
the acquisition module is used for acquiring image information of the answer sheet; the preprocessing module is used for preprocessing the image information of the answer sheet; and the identification module is used for identifying the preprocessed image information of the answer sheet through an AlexNet neural network algorithm and judging the score of the answer sheet.
The terminal device 5 may be a desktop computer, a notebook, a palm computer, a cloud server, or other computing devices. The server may include, but is not limited to, a processor 50, a memory 51. Those skilled in the art will appreciate that fig. 5 is merely an example of a terminal device 5 and does not constitute a limitation of the terminal device 5 and may include more or less components than those shown, or some components may be combined, or different components, e.g., the server may also include input output devices, network access devices, buses, etc.
The Processor 50 may be a Central Processing Unit (CPU), other general purpose Processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), an off-the-shelf Programmable Gate Array (FPGA) or other Programmable logic device, discrete Gate or transistor logic, discrete hardware components, etc. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
The memory 51 may be an internal storage unit of the terminal device 5, such as a hard disk or a memory of the terminal device 5. The memory 51 may also be an external storage device of the terminal device 5, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card), and the like, which are provided on the terminal device 5. Further, the memory 51 may also include both an internal storage unit and an external storage device of the terminal device 5. The memory 51 is used for storing the computer program and other programs and data required by the terminal device 5. The memory 51 may also be used to temporarily store data that has been output or is to be output.
An embodiment of the present invention further provides a computer-readable storage medium, where a computer program is stored, and when the computer program is executed by a processor, the computer program implements the steps of any one of the above methods.
It should be understood that, the sequence numbers of the steps in the foregoing embodiments do not imply an execution sequence, and the execution sequence of each process should be determined by its function and inherent logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.
It will be apparent to those skilled in the art that, for convenience and brevity of description, only the above-mentioned division of the functional units and modules is illustrated, and in practical applications, the above-mentioned function distribution may be performed by different functional units and modules according to needs, that is, the internal structure of the apparatus is divided into different functional units or modules to perform all or part of the above-mentioned functions. Each functional unit and module in the embodiments may be integrated in one processing unit, or each unit may exist alone physically, or two or more units are integrated in one unit, and the integrated unit may be implemented in a form of hardware, or in a form of software functional unit. In addition, specific names of the functional units and modules are only for convenience of distinguishing from each other, and are not used for limiting the protection scope of the present application. The specific working processes of the units and modules in the system may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.
In the above embodiments, the descriptions of the respective embodiments have respective emphasis, and reference may be made to the related descriptions of other embodiments for parts that are not described or illustrated in a certain embodiment.
Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.
In the embodiments provided in the present invention, it should be understood that the disclosed apparatus/terminal device and method may be implemented in other ways. For example, the above-described embodiments of the apparatus/terminal device are merely illustrative, and for example, the division of the modules or units is only one logical division, and there may be other divisions when actually implemented, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
In addition, functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.
The integrated modules/units, if implemented in the form of software functional units and sold or used as separate products, may be stored in a computer readable storage medium. Based on such understanding, all or part of the flow of the method according to the embodiments of the present invention may also be implemented by a computer program, which may be stored in a computer-readable storage medium, and when the computer program is executed by a processor, the steps of the method embodiments may be implemented. . Wherein the computer program comprises computer program code, which may be in the form of source code, object code, an executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, usb disk, removable hard disk, magnetic disk, optical disk, computer Memory, Read-Only Memory (ROM), Random Access Memory (RAM), electrical carrier wave signals, telecommunications signals, software distribution medium, and the like. It should be noted that the computer readable medium may contain content that is subject to appropriate increase or decrease as required by legislation and patent practice in jurisdictions, for example, in some jurisdictions, computer readable media does not include electrical carrier signals and telecommunications signals as is required by legislation and patent practice.
The above-mentioned embodiments are only used for illustrating the technical solutions of the present invention, and not for limiting the same; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; such modifications and substitutions do not substantially depart from the spirit and scope of the embodiments of the present invention, and are intended to be included within the scope of the present invention.
Claims (10)
1. An answer sheet score identification method is characterized by comprising the following steps:
acquiring image information of an answer sheet;
preprocessing the image information of the answer sheet;
and identifying the preprocessed image information of the answer sheet through an AlexNet neural network algorithm and determining the score of the answer sheet.
2. The method for identifying the score of the answer sheet according to claim 1, wherein the step of identifying the preprocessed image information of the answer sheet by using an AlexNet neural network algorithm and determining the score of the answer sheet comprises the following steps:
constructing an AlexNet neural network model;
training the AlexNet neural network model through a training data set to generate a training model;
inputting the preprocessed image information of the answer sheet into the training model for recognition, and determining the score of the answer sheet according to the recognition result.
3. The answer sheet score identification method of claim 2, wherein constructing an AlexNet neural network model comprises:
determining the number of layers of the hidden layer and the number of identified classifications, and establishing an activation function, a residual calculation formula and a gradient descent function to obtain an AlexNet neural network model;
training the AlexNet neural network model through a training data set to generate a training model, comprising:
constructing a training data set;
inputting a sample of the training data set, and training the standard model according to the activation function, the residual calculation formula and the gradient descent function to generate a training model;
and the output result of each hidden layer and the quantization threshold value of the expected value are measured according to a mean square error formula.
4. The answer sheet score recognition method of claim 3,
the activation function is:
wherein, aiIs the output of the neuron or the like,is a probability vector, x is the transpose of the input dimension vector, biIs the weight proportion of the initial value;
the residual calculation formula is as follows:
wherein, delta(k-2)In order to be the value of the residual error,is the weight proportion of the current layer node,a difference value calculated for the forward direction;
the gradient descent function is:
wherein,is the weight proportion of the current layer node, b is the branch vector,as a loss function, a is the output of the neuron;
the mean square error formula is:
where MSE is the quantization threshold, yiExpected value of the current hidden layer, f (x)i) Is the output result of the current hidden layer.
5. The method for identifying the score of the answer sheet according to claim 2, wherein the step of inputting the preprocessed image information of the answer sheet into the training model for identification and determining the score of the answer sheet according to the identification result comprises:
inputting the preprocessed image information of the answer sheet into the training model;
carrying out convolution operation on the preprocessed image information of the answer sheet through a convolution layer;
performing pooling processing on the data subjected to the convolution operation through a pooling layer;
performing full connection operation on the pooled data through an output layer to generate an identification result;
residual error detection is carried out on the recognition result, and a training model is updated according to the residual error detection result;
and comparing the standard answers with the identification results, and obtaining the scores of the answer sheets according to the weights obtained by comparison.
6. The score recognition method for an answer sheet according to any one of claims 1 to 5,
before the obtaining of the image information of the answer sheet, the method further comprises:
modifying the answer template according to the test paper information and generating an answer sheet, wherein the answer sheet comprises a positioning area used as an image segmentation reference;
the preprocessing of the answer sheet image information comprises the following steps:
performing first processing on the image information of the answer sheet, wherein the first processing comprises at least one of color balance processing, contrast adjustment processing, inclination correction processing and automatic effective area delineation processing;
if the answer sheet image information after the first processing is a multi-channel image, performing region division on the answer sheet image information after the first processing according to the color information of the answer sheet image information after the first processing to obtain preprocessed answer sheet image information;
and if the image information after the first processing is a gray image, performing area division on the image information of the answer sheet after the first processing according to a positioning reference included in the positioning area in the image information of the answer sheet after the first processing to obtain the image information of the answer sheet after the pre-processing.
7. The answer sheet score identification method according to any one of claims 1 to 5, wherein after the step of identifying the preprocessed answer sheet image information by AlexNet neural network algorithm and determining the answer sheet score, the method further comprises the following steps:
storing an original answer sheet image and an identified answer sheet image, and manually performing spot check and comparison on the original answer sheet image and the identified answer sheet image to verify the correctness of the answer score;
the method comprises the steps of storing the scores of the answer sheets, conducting index sorting on the scores of the answer sheets to obtain the ranks of respondents corresponding to the scores of the answer sheets and the historical records of each examination, and drawing and analyzing the score curves of the respondents according to the historical records.
8. An answer sheet score recognition device, comprising:
the acquisition module is used for acquiring image information of the answer sheet;
the preprocessing module is used for preprocessing the image information of the answer sheet;
and the identification module is used for identifying the preprocessed image information of the answer sheet through an AlexNet neural network algorithm and judging the score of the answer sheet.
9. A terminal device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor, when executing the computer program, implements the method according to any of claims 1-7.
10. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the method according to any one of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810934893.5A CN109241869A (en) | 2018-08-16 | 2018-08-16 | The recognition methods of answering card score, device and terminal device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810934893.5A CN109241869A (en) | 2018-08-16 | 2018-08-16 | The recognition methods of answering card score, device and terminal device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109241869A true CN109241869A (en) | 2019-01-18 |
Family
ID=65069714
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810934893.5A Pending CN109241869A (en) | 2018-08-16 | 2018-08-16 | The recognition methods of answering card score, device and terminal device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109241869A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110322379A (en) * | 2019-07-08 | 2019-10-11 | 深圳中兴网信科技有限公司 | Paper processing method and paper processing system |
| CN111191766A (en) * | 2020-01-02 | 2020-05-22 | 深圳芯英科技有限公司 | Data processing method, device, processor and computer readable storage medium |
| 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 |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050255439A1 (en) * | 2004-05-14 | 2005-11-17 | Preston Cody | Method and system for generating and processing an assessment examination |
| CN104463101A (en) * | 2014-11-06 | 2015-03-25 | 科大讯飞股份有限公司 | Answer recognition method and system for textual test question |
| CN107025456A (en) * | 2017-03-20 | 2017-08-08 | 安徽七天教育科技有限公司 | A kind of method of teacher comment vestige automatic identification |
| US20170262738A1 (en) * | 2014-09-16 | 2017-09-14 | Iflytek Co., Ltd. | Intelligent scoring method and system for text objective question |
| CN107590494A (en) * | 2017-09-18 | 2018-01-16 | 哈尔滨成长科技有限公司 | Answer sheet picture localization method, device, readable storage medium storing program for executing and electronic equipment |
| CN107590495A (en) * | 2017-09-18 | 2018-01-16 | 哈尔滨成长科技有限公司 | Answer sheet picture method for correcting error, device, readable storage medium storing program for executing and electronic equipment |
| CN107798321A (en) * | 2017-12-04 | 2018-03-13 | 海南云江科技有限公司 | A kind of examination paper analysis method and computing device |
| CN108172050A (en) * | 2017-12-26 | 2018-06-15 | 科大讯飞股份有限公司 | Mathematics subjective item answer result corrects method and system |
| CN108171297A (en) * | 2018-01-24 | 2018-06-15 | 谢德刚 | A kind of answer card identification method and device |
-
2018
- 2018-08-16 CN CN201810934893.5A patent/CN109241869A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050255439A1 (en) * | 2004-05-14 | 2005-11-17 | Preston Cody | Method and system for generating and processing an assessment examination |
| US20170262738A1 (en) * | 2014-09-16 | 2017-09-14 | Iflytek Co., Ltd. | Intelligent scoring method and system for text objective question |
| CN104463101A (en) * | 2014-11-06 | 2015-03-25 | 科大讯飞股份有限公司 | Answer recognition method and system for textual test question |
| CN107025456A (en) * | 2017-03-20 | 2017-08-08 | 安徽七天教育科技有限公司 | A kind of method of teacher comment vestige automatic identification |
| CN107590494A (en) * | 2017-09-18 | 2018-01-16 | 哈尔滨成长科技有限公司 | Answer sheet picture localization method, device, readable storage medium storing program for executing and electronic equipment |
| CN107590495A (en) * | 2017-09-18 | 2018-01-16 | 哈尔滨成长科技有限公司 | Answer sheet picture method for correcting error, device, readable storage medium storing program for executing and electronic equipment |
| CN107798321A (en) * | 2017-12-04 | 2018-03-13 | 海南云江科技有限公司 | A kind of examination paper analysis method and computing device |
| CN108172050A (en) * | 2017-12-26 | 2018-06-15 | 科大讯飞股份有限公司 | Mathematics subjective item answer result corrects method and system |
| CN108171297A (en) * | 2018-01-24 | 2018-06-15 | 谢德刚 | A kind of answer card identification method and device |
Non-Patent Citations (2)
| Title |
|---|
| 李冰莹: ""网络阅卷及其关键技术研究"", 《中国优秀硕士学位论文全文数据库·社会科学Ⅱ辑》 * |
| 马飒飒等编著: "《物联网基础技术及应用》", 31 January 2018, 西安：西安电子科技大学出版社 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110322379A (en) * | 2019-07-08 | 2019-10-11 | 深圳中兴网信科技有限公司 | Paper processing method and paper processing system |
| CN111191766A (en) * | 2020-01-02 | 2020-05-22 | 深圳芯英科技有限公司 | Data processing method, device, processor and computer readable storage medium |
| CN111191766B (en) * | 2020-01-02 | 2023-05-16 | 中昊芯英(杭州)科技有限公司 | Data processing method, device, processor and computer readable storage medium |
| 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 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20190180154A1 (en) | Text recognition using artificial intelligence | |
| US11455525B2 (en) | Method and apparatus of open set recognition and a computer readable storage medium | |
| RU2661750C1 (en) | Symbols recognition with the use of artificial intelligence | |
| US20190294921A1 (en) | Field identification in an image using artificial intelligence | |
| CN104463101B (en) | Answer recognition methods and system for character property examination question | |
| CN109086654B (en) | Handwriting model training method, text recognition method, device, equipment and medium | |
| CN109993201B (en) | Image processing method, device and readable storage medium | |
| US20200134382A1 (en) | Neural network training utilizing specialized loss functions | |
| WO2021042505A1 (en) | Note generation method and apparatus based on character recognition technology, and computer device | |
| CN111652332A (en) | Deep learning handwritten Chinese character recognition method and system based on two classifications | |
| Asthana et al. | Handwritten multiscript numeral recognition using artificial neural networks | |
| CN112862024B (en) | Text recognition method and system | |
| AU2020100413A4 (en) | Docket analysis methods and systems | |
| CN109241869A (en) | The recognition methods of answering card score, device and terminal device | |
| US11715288B2 (en) | Optical character recognition using specialized confidence functions | |
| CN112036295A (en) | Bill image processing method, bill image processing device, storage medium and electronic device | |
| Karunarathne et al. | Recognizing ancient sinhala inscription characters using neural network technologies | |
| CN108520261B (en) | Method and device for identifying peanut kernel number | |
| CN115620312A (en) | Cross-modal character handwriting verification method, system, equipment and storage medium | |
| CN106529490B (en) | Based on the sparse system and method for realizing writer verification from coding code book | |
| CN111414930B (en) | Deep learning model training method and device, electronic equipment and storage medium | |
| CN111612045B (en) | Universal method for acquiring target detection data set | |
| CN111881880A (en) | Bill text recognition method based on novel network | |
| CN113177602B (en) | Image classification method, device, electronic equipment and storage medium | |
| CN114332898A (en) | Automatic correcting method and device for connection test questions and storage medium |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190118 |