CN114581804A - Bank card identification model training method and device and bank card identification method and device - Google Patents
Bank card identification model training method and device and bank card identification method and device Download PDFInfo
- Publication number
- CN114581804A CN114581804A CN202210251569.XA CN202210251569A CN114581804A CN 114581804 A CN114581804 A CN 114581804A CN 202210251569 A CN202210251569 A CN 202210251569A CN 114581804 A CN114581804 A CN 114581804A
- Authority
- CN
- China
- Prior art keywords
- bank card
- image
- resolution
- training
- model
- 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
Images
Classifications
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformation in the plane of the image
- G06T3/40—Scaling the whole image or part thereof
- G06T3/4053—Super resolution, i.e. output image resolution higher than sensor resolution
Abstract
The invention provides a method and a device for training a bank card recognition model, which relate to the field of finance, and comprise the following steps: processing the bank card image according to a preset resolution threshold value to obtain a high-resolution image and a low-resolution image; generating a super-resolution countermeasure network model according to the bank card image, the high-resolution image and the low-resolution image; and processing the training image of the bank card identification model by using the super-resolution countermeasure network model, and training the bank card identification model by using the processed training image. According to the method, the super-resolution confrontation network model is used for processing the training images of the bank card recognition model to obtain the resolution ratio of the improved training images, the processed images with the improved resolution ratio are used for carrying out the bank card recognition model training, the resolution ratio of the training set images is improved to improve the image recognition model algorithm, and the recognition efficiency is further improved.
Description
Technical Field
The invention relates to an image processing technology, in particular to a bank card identification model training method and a bank card identification method and device.
Background
With the continuous development of science and technology, the material level of people is improved, so people seek the experience of products more and more. The bank card is an indispensable thing in life, and many life scenes all use binding bank card. However, the manual input of the bank card number is complicated in operation and prone to error, and the user experiences the bank card number more and more badly. An Optical Character Recognition technology (OCR) based on deep learning is favored by most users, and the bank card number can be recognized by photographing, so that the tedious process of inputting the bank card is greatly reduced.
However, the existing OCR technology has low recognition efficiency and cannot meet the requirements of people. At present, the OCR technology is improved, but most of the improvement methods start from the OCR algorithm, and improve the recognition rate by introducing other networks or improving the algorithm network. Although the effect is improved, a large lifting space still exists.
Disclosure of Invention
In order to overcome at least one defect existing in the identification process of the bank card in the prior art, the invention provides a training method of a bank card identification model, which comprises the following steps:
processing the bank card image according to a preset resolution threshold value to obtain a high-resolution image and a low-resolution image;
generating a super-resolution countermeasure network model according to the bank card image, the high-resolution image and the low-resolution image;
and processing the training image of the bank card identification model by using the super-resolution countermeasure network model, and training the bank card identification model by using the processed training image.
In the embodiment of the present invention, the processing the bank card image according to the preset resolution threshold to obtain the high resolution image and the low resolution image includes:
carrying out interpolation processing on the bank card image according to a first preset resolution threshold value to generate a low-resolution image;
and generating a high-resolution image according to the low-resolution image and a preset second resolution threshold by using a generating network model.
In the embodiment of the present invention, the generating a super-resolution countermeasure network model according to the bank card image, the high-resolution image and the low-resolution image includes:
establishing an initial model of a confrontation network;
and training the countermeasure network initial model by using the bank card image, the high-resolution image and the low-resolution image to generate a super-resolution countermeasure network model.
In the embodiment of the present invention, the training of the countermeasure network initial model by using the bank card image, the high resolution image and the low resolution image to generate the super-resolution countermeasure network model includes:
determining a loss function for judging network optimization according to the bank card image and the high-resolution image;
determining and generating a loss function of network optimization according to the low-resolution image;
and training the confrontation network initial model by utilizing the bank card image, the high-resolution image and the low-resolution image according to the determined loss function for judging network optimization and generating the loss function for network optimization to generate a super-resolution confrontation network model.
In the embodiment of the present invention, the determining a loss function for discriminating network optimization according to the bank card image and the high resolution image includes:
respectively determining a countermeasure loss function and a feature matching loss function according to the bank card image and the high-resolution image;
and determining a loss function for judging network optimization according to the impedance loss function and the feature matching loss function.
In the embodiment of the present invention, the training of the countermeasure network initial model by using the bank card image, the high resolution image and the low resolution image to generate the super-resolution countermeasure network model includes:
and training the confrontation network initial model by utilizing the bank card image, the high-resolution image and the low-resolution image by adopting an RMSProp algorithm to determine a super-resolution confrontation network model.
In the embodiment of the present invention, the processing the training image of the bank card identification model by using the super-resolution countermeasure network model, and the training of the bank card identification model by using the processed training image includes:
processing the training image of the bank card identification model by using the super-resolution countermeasure network model to generate a super-resolution bank card image;
and training a bank card identification model by using the generated super-resolution bank card image.
In addition, the invention also provides a bank card identification method, which carries out model training on a bank card identification model by utilizing the bank card identification model training method;
and identifying the bank card by using the trained bank card identification model.
Meanwhile, the invention also provides a bank card recognition model training device, which comprises:
the image processing module is used for processing the bank card image according to a preset resolution threshold value to obtain a high-resolution image and a low-resolution image;
the confrontation network model determining module is used for generating a super-resolution confrontation network model according to the bank card image, the high-resolution image and the low-resolution image;
and the recognition model training module is used for processing the training images of the bank card recognition model by using the super-resolution countermeasure network model and performing bank card recognition model training by using the processed training images.
In an embodiment of the present invention, the image processing module includes:
the low-resolution image generation unit is used for carrying out interpolation processing on the bank card image according to a first preset resolution threshold value to generate a low-resolution image;
and the high-resolution image generating unit is used for generating a high-resolution image according to the low-resolution image and a preset second resolution threshold value by utilizing a generating network model.
In the embodiment of the present invention, the confrontation network model determining module includes:
the initial model establishing unit is used for establishing a confrontation network initial model;
and the training unit is used for training the confrontation network initial model by utilizing the bank card image, the high-resolution image and the low-resolution image to generate a super-resolution confrontation network model.
In the embodiment of the present invention, the training unit training the countermeasure network initial model by using the bank card image, the high resolution image and the low resolution image to generate the super-resolution countermeasure network model includes:
determining a loss function for judging network optimization according to the bank card image and the high-resolution image;
determining and generating a loss function of network optimization according to the low-resolution image;
and training the confrontation network initial model by utilizing the bank card image, the high-resolution image and the low-resolution image according to the determined loss function for judging network optimization and generating the loss function for network optimization to generate a super-resolution confrontation network model.
In the embodiment of the present invention, the recognition model training module includes:
the super-resolution image generation unit is used for processing the training image of the bank card identification model by using the super-resolution countermeasure network model to generate a super-resolution bank card image;
and the recognition model training unit is used for training the bank card recognition model by utilizing the generated super-resolution bank card image.
Meanwhile, the invention also provides computer equipment which comprises a memory, a processor and a computer program which is stored on the memory and can run on the processor, wherein the processor realizes the method when executing the computer program.
Meanwhile, the invention also provides a computer readable storage medium, and a computer program for executing the method is stored in the computer readable storage medium.
The invention provides a bank card identification model training method, which comprises the steps of generating a super-resolution confrontation network model according to a bank card image, a high-resolution image and a low-resolution image, processing a training image of the bank card identification model by using the super-resolution confrontation network model to obtain the resolution of the improved training image, and performing bank card identification model training by using the processed image with the improved resolution.
In order to make the aforementioned and other objects, features and advantages of the invention comprehensible, preferred embodiments accompanied with figures are described in detail below.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
FIG. 1 is a flow chart of a method for training a bank card recognition model according to the present invention;
FIG. 2 is a flow chart in an embodiment of the present invention;
FIG. 3 is a block diagram of a training apparatus for bank card recognition model provided in the present invention;
fig. 4 is a schematic diagram of an electronic device provided in an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without making any creative effort based on the embodiments in the present invention, belong to the protection scope of the present invention.
OCR (Optical Character Recognition) refers to a process in which an electronic device (e.g., a scanner or a digital camera) checks a Character printed on paper, determines its shape by detecting dark and light patterns, and then translates the shape into a computer text by a Character Recognition method; the method is characterized in that characters in a paper document are converted into an image file with a black-white dot matrix in an optical mode aiming at print characters, and the characters in the image are converted into a text format through recognition software for further editing and processing by word processing software.
At present, the OCR technology is improved by starting from the OCR algorithm, and improving the recognition rate by introducing other networks or improving the algorithm network. Although the effect is improved, a large lifting space still exists. Most methods ignore the quality of the training image and directly influence the effect of the OCR algorithm, and can start from improving the training image, wherein one method is to improve the resolution of the image.
At present, on the basis of an image super-resolution algorithm, a convolutional neural network is more based, however, the expression capability of the neural network is limited, more detailed features of an image cannot be reserved, and the improvement effect is not ideal. Therefore, the super-resolution countermeasure network is utilized to improve the fitting capability of the network, and the intermediate features are used for feature matching, so that the details and the definition of the bank card image are improved.
In view of the above, the present invention provides a method for training a bank card recognition model, as shown in fig. 1, the method includes:
step S101, processing the bank card image according to a preset resolution threshold value to obtain a high-resolution image and a low-resolution image;
step S102, generating a super-resolution countermeasure network model according to the bank card image, the high-resolution image and the low-resolution image;
and S103, processing the training image of the bank card identification model by using the super-resolution countermeasure network model, and performing bank card identification model training by using the processed training image.
The invention provides a bank card identification model training method, which comprises the steps of generating a super-resolution confrontation network model according to a bank card image, a high-resolution image and a low-resolution image, processing a training image of the bank card identification model by using the super-resolution confrontation network model to obtain the resolution of the improved training image, and performing bank card identification model training by using the processed image with the improved resolution.
In the embodiment of the present invention, the processing the bank card image according to the preset resolution threshold to obtain the high resolution image and the low resolution image includes:
carrying out interpolation processing on the bank card image according to a first preset resolution threshold value to generate a low-resolution image;
and generating a high-resolution image according to the generated low-resolution image by utilizing a generating network model and a preset second resolution threshold.
Specifically, in the embodiment of the invention, a high-resolution camera is used for acquiring a high-definition bank card image.
Carrying out interpolation processing on the bank card image by using a bicubic interpolation method for the acquired high-definition bank card image to obtain a low-resolution image meeting the preset resolution;
and generating a high-resolution image generated by the network according to the obtained low-resolution image.
In the embodiment of the invention, the generation network learns a function from low resolution to high resolution, and the low resolution image is utilized to generate the high resolution image through the generation network.
In the embodiment of the present invention, the generating a super-resolution countermeasure network model according to the bank card image, the high-resolution image and the low-resolution image includes:
establishing an initial model of a confrontation network;
and training the countermeasure network initial model by using the bank card image, the high-resolution image and the low-resolution image to generate a super-resolution countermeasure network model.
In the embodiment of the present invention, the training of the countermeasure network initial model by using the bank card image, the high resolution image and the low resolution image to generate the super-resolution countermeasure network model includes:
determining a loss function for judging network optimization according to the bank card image and the high-resolution image;
determining and generating a loss function of network optimization according to the low-resolution image;
and training the confrontation network initial model by utilizing the bank card image, the high-resolution image and the low-resolution image according to the determined loss function for judging network optimization and generating the loss function for network optimization to generate a super-resolution confrontation network model.
In the embodiment of the present invention, the determining a loss function for discriminating network optimization according to the bank card image and the high resolution image includes:
respectively determining a countermeasure loss function and a feature matching loss function according to the bank card image and the high-resolution image;
and determining a loss function for judging network optimization according to the impedance loss function and the feature matching loss function.
In the embodiment of the present invention, the training of the countermeasure network initial model by using the bank card image, the high resolution image and the low resolution image to generate the super-resolution countermeasure network model includes:
and training the countermeasure network initial model by utilizing the bank card image, the high-resolution image and the low-resolution image by adopting an RMSProp algorithm to determine a super-resolution countermeasure network model.
In the embodiment of the present invention, the processing the training image of the bank card identification model by using the super-resolution countermeasure network model, and the training of the bank card identification model by using the processed training image includes:
processing the training image of the bank card identification model by using the super-resolution countermeasure network model to generate a super-resolution bank card image;
and training a bank card identification model by using the generated super-resolution bank card image.
Specifically, in the embodiment of the present invention, a flow of training a super-resolution countermeasure network is shown in fig. 2.
Acquiring a real bank card image;
carrying out bicubic interpolation processing on the obtained real bank card image to generate a low-resolution image meeting a preset resolution threshold;
generating a high-resolution image, and generating the high-resolution image according to the low-resolution image obtained after the double interpolation processing;
training a super-resolution countermeasure network;
and (4) training an OCR network model, wherein a trained super-resolution countermeasure network is utilized to process a training set of the OCR model, and the image resolution is improved.
In the embodiment of the invention, in the process of training the super-resolution antagonistic network, the loss function for judging network optimization is as follows:
LD＝LA+LF
wherein L isARepresents the antagonistic loss function, LFA feature matching loss function, a countermeasure loss and a feature matching loss are represented.
Like most deep learning algorithms, the training process for training the super-resolution countermeasure network in the embodiment of the invention needs to minimize one loss, and the loss function needs to be minimized in the network training process. The minimization process is a process of enabling the model to reach the optimum through network learning and continuous iteration. The smaller the loss value, the better the model and the higher the recognition rate.
Specifically, the penalty-fighting function in this embodiment is as follows:
wherein, IHRRepresenting real images acquired using a high resolution camera, ISRRepresenting a high resolution image of the output of the generating network, E representing a mathematical expectation, IHRP represents IHRSubject to the distribution P, D denotes the variance,
is shown as IHRObeying the p distribution, D (I)HR) A mathematical expectation of (d);
in the same way, it can be known that,
wherein p denotes the distribution of the real image, q denotes the distribution of the generated image, and λ is a hyper-parameter, which is a parameter adjusted according to the specific application scenario.
A e U (0,1), the parameter a is a random real number obeying a uniform distribution between 0 and 1,
has the meaning ofHRAnd ISRRandom interpolation in between (i.e. values lying between them),
to represent
D denotes the discrimination network.
The feature matching loss function is as follows:
wherein φ (-) represents the feature graph of the last ResNet output of the discrimination network. The feature matching loss function acts as a function of the loss function, countering the loss and the feature matching loss,
the loss function for generating network optimization of the invention is as follows:
wherein, ILRAnd G represents the generation network. The generation of the countermeasure network G is to alternately train through two networks, and finally find a balance point which is a Nash balance point. After the network training is finished, the test can be carried out only by generating the network.
In the embodiment of the invention, the function of the loss function is to train an optimal model, and the smaller the loss is, the better the model is. The training process principle is the process of optimizing (minimizing) the loss function, and the loss is minimized by continuously iteratively updating and adjusting parameters.
The training process adopts a RMSProp optimization algorithm, and convergence is achieved through ten thousand times of iterative networks. The RMSProp algorithm is a full-name Root Mean Square Prop, and is an optimization algorithm, the deep learning network learning process can be regarded as learning or fitting a function, and the G generation network in the embodiment learns a function from low resolution to high resolution. In the learning process, a learning objective is needed to guide the network learning, and the objective is a loss function, and the learning process of the network is influenced by different designs of the loss function. A good loss function may help the network converge faster and learn better. The mathematical representation in the embodiment is a mathematical representation which is often used in the deep learning field, and no representation ambiguity exists.
In addition, the invention also provides a bank card identification method, which utilizes the bank card identification model training method provided in the embodiment to carry out model training on the bank card identification model;
and identifying the bank card by using the trained bank card identification model.
In the training process of the bank card recognition model, the resolution of the training set image is improved from the training set of the OCR, and the efficiency of the bank card recognition model recognition is further improved
The bank card recognition model training method provided by the embodiment of the invention is characterized in that a ResNet (Deep residual network) based bank card super-resolution method is built by using a ResNet structure with stronger expression capability and better fitting capability, and is trained by using a countermeasure form, so that a generated image is clearer; meanwhile, the characteristic matching loss is added to the countermeasure loss, so that the network can keep more detailed characteristics of the image when generating the super-resolution bank card image. The method starts from the training set of the OCR, prompts the efficiency of recognizing the model by improving the resolution of the images of the training set, thereby improving the OCR algorithm, and further improves the recognition efficiency by combining with a new OCR improved algorithm.
Meanwhile, the present invention also provides a device for training a bank card recognition model, as shown in fig. 3, including:
the image processing module 301 is configured to process the bank card image according to a preset resolution threshold to obtain a high-resolution image and a low-resolution image;
the confrontation network model determining module 302 is used for generating a super-resolution confrontation network model according to the bank card image, the high-resolution image and the low-resolution image;
and the recognition model training module 303 is configured to process a training image of the bank card recognition model by using the super-resolution countermeasure network model, and perform bank card recognition model training by using the processed training image.
For those skilled in the art, the implementation of the device for training a bank card recognition model provided by the present invention can be clearly understood through the description of the foregoing embodiments, and details are not repeated herein.
It should be noted that the method and the device for training the bank card identification model disclosed by the invention can be used for training the bank card identification model in the financial field, and can also be used in any field except the financial field.
The present embodiment also provides an electronic device, which may be a desktop computer, a tablet computer, a mobile terminal, and the like, but is not limited thereto. In this embodiment, the electronic device may refer to the embodiments of the method and the apparatus, and the contents thereof are incorporated herein, and repeated descriptions are omitted.
Fig. 4 is a schematic block diagram of a system configuration of an electronic apparatus 600 according to an embodiment of the present invention. As shown in fig. 4, the electronic device 600 may include a central processor 100 and a memory 140; the memory 140 is coupled to the central processor 100. Notably, this diagram is exemplary; other types of structures may also be used in addition to or in place of the structure to implement telecommunications or other functions.
In one embodiment, the bank card recognition model training function may be integrated into the central processor 100. The central processor 100 may be configured to control as follows:
processing the bank card image according to a preset resolution threshold value to obtain a high-resolution image and a low-resolution image;
generating a super-resolution countermeasure network model according to the bank card image, the high-resolution image and the low-resolution image;
and processing the training image of the bank card identification model by using the super-resolution countermeasure network model, and training the bank card identification model by using the processed training image.
In another embodiment, the bank card recognition model training device may be configured separately from the central processor 100, for example, the bank card recognition model training device may be configured as a chip connected to the central processor 100, and the bank card recognition model training function is realized by the control of the central processor.
As shown in fig. 4, the electronic device 600 may further include: communication module 110, input unit 120, audio processing unit 130, display 160, power supply 170. It is noted that the electronic device 600 does not necessarily include all of the components shown in FIG. 4; furthermore, the electronic device 600 may also comprise components not shown in fig. 4, which may be referred to in the prior art.
As shown in fig. 4, the central processor 100, sometimes referred to as a controller or operational control, may include a microprocessor or other processor device and/or logic device, the central processor 100 receiving input and controlling the operation of the various components of the electronic device 600.
The memory 140 may be, for example, one or more of a buffer, a flash memory, a hard drive, a removable media, a volatile memory, a non-volatile memory, or other suitable devices. The information relating to the failure may be stored, and a program for executing the information may be stored. And the central processing unit 100 may execute the program stored in the memory 140 to realize information storage or processing, etc.
The input unit 120 provides input to the cpu 100. The input unit 120 is, for example, a key or a touch input device. The power supply 170 is used to provide power to the electronic device 600. The display 160 is used to display an object to be displayed, such as an image or a character. The display may be, for example, an LCD display, but is not limited thereto.
The memory 140 may be a solid state memory such as Read Only Memory (ROM), Random Access Memory (RAM), a SIM card, or the like. There may also be a memory that holds information even when power is off, can be selectively erased, and is provided with more data, an example of which is sometimes called an EPROM or the like. The memory 140 may also be some other type of device. Memory 140 includes buffer memory 141 (sometimes referred to as a buffer). The memory 140 may include an application/function storage section 142, and the application/function storage section 142 is used to store application programs and function programs or a flow for executing the operation of the electronic device 600 by the central processing unit 100.
The memory 140 may also include a data store 143, the data store 143 for storing data, such as contacts, digital data, pictures, sounds, and/or any other data used by the electronic device. The driver storage portion 144 of the memory 140 may include various drivers of the electronic device for a communication function and/or for performing other functions of the electronic device (e.g., a messaging application, a directory application, etc.).
The communication module 110 is a transmitter/receiver 110 that transmits and receives signals via an antenna 111. The communication module (transmitter/receiver) 110 is coupled to the central processor 100 to provide an input signal and receive an output signal, which may be the same as in the case of a conventional mobile communication terminal.
Based on different communication technologies, a plurality of communication modules 110, such as a cellular network module, a bluetooth module, and/or a wireless local area network module, may be provided in the same electronic device. The communication module (transmitter/receiver) 110 is also coupled to a speaker 131 and a microphone 132 via an audio processor 130 to provide audio output via the speaker 131 and receive audio input from the microphone 132 to implement general telecommunications functions. Audio processor 130 may include any suitable buffers, decoders, amplifiers and so forth. In addition, an audio processor 130 is also coupled to the central processor 100, so that recording on the local can be enabled through a microphone 132, and so that sound stored on the local can be played through a speaker 131.
Embodiments of the present invention further provide a computer-readable program, where when the program is executed in an electronic device, the program causes a computer to execute the method for training a bank card identification model in the electronic device according to the above embodiments.
The embodiment of the present invention further provides a storage medium storing a computer-readable program, where the computer-readable program enables a computer to execute the training of the bank card identification model in the electronic device according to the above embodiment.
The preferred embodiments of the present invention have been described above with reference to the accompanying drawings. The many features and advantages of the embodiments are apparent from the detailed specification, and thus, it is intended by the appended claims to cover all such features and advantages of the embodiments that fall within the true spirit and scope thereof. Further, since numerous modifications and changes will readily occur to those skilled in the art, it is not desired to limit the embodiments of the invention to the exact construction and operation illustrated and described, and accordingly, all suitable modifications and equivalents may be resorted to, falling within the scope thereof.
As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The principle and the implementation mode of the invention are explained by applying specific embodiments in the invention, and the description of the embodiments is only used for helping to understand the method and the core idea of the invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present invention.
Claims (11)
1. A bank card recognition model training method is characterized by comprising the following steps:
processing the bank card image according to a preset resolution threshold value to obtain a high-resolution image and a low-resolution image;
generating a super-resolution countermeasure network model according to the bank card image, the high-resolution image and the low-resolution image;
and processing the training image of the bank card identification model by using the super-resolution countermeasure network model, and training the bank card identification model by using the processed training image.
2. The training method of the bank card recognition model according to claim 1, wherein the processing the bank card image according to the preset resolution threshold to obtain the high resolution image and the low resolution image comprises:
carrying out interpolation processing on the bank card image according to a first preset resolution threshold value to generate a low-resolution image;
and generating a high-resolution image according to the low-resolution image and a preset second resolution threshold by using a generating network model.
3. The training method of the bank card recognition model according to claim 1, wherein the generating the super-resolution countermeasure network model according to the bank card image, the high-resolution image and the low-resolution image comprises:
establishing an initial model of a confrontation network;
and training the countermeasure network initial model by using the bank card image, the high-resolution image and the low-resolution image to generate a super-resolution countermeasure network model.
4. The method for training the bank card recognition model according to claim 3, wherein the training of the countermeasure network initial model by using the bank card image, the high resolution image and the low resolution image to generate the super-resolution countermeasure network model comprises:
determining a loss function for judging network optimization according to the bank card image and the high-resolution image;
determining and generating a loss function of network optimization according to the low-resolution image;
and training the confrontation network initial model by utilizing the bank card image, the high-resolution image and the low-resolution image according to the determined loss function for judging network optimization and generating the loss function for network optimization to generate a super-resolution confrontation network model.
5. The training method of the bank card recognition model according to claim 4, wherein the determining the loss function for discriminating network optimization according to the bank card image and the high resolution image comprises:
respectively determining a countermeasure loss function and a feature matching loss function according to the bank card image and the high-resolution image;
and determining a loss function for judging network optimization according to the impedance loss function and the feature matching loss function.
6. The method for training the bank card recognition model according to claim 4, wherein the training of the countermeasure network initial model by using the bank card image, the high resolution image and the low resolution image to generate the super-resolution countermeasure network model comprises:
and training the countermeasure network initial model by utilizing the bank card image, the high-resolution image and the low-resolution image by adopting an RMSProp algorithm to determine a super-resolution countermeasure network model.
7. The method for training the bank card recognition model according to claim 1, wherein the processing of the training image of the bank card recognition model by using the super-resolution countermeasure network model, and the training of the bank card recognition model by using the processed training image comprises:
processing the training image of the bank card identification model by using the super-resolution countermeasure network model to generate a super-resolution bank card image;
and training a bank card identification model by using the generated super-resolution bank card image.
8. A bank card identification method is characterized by comprising the following steps: performing model training on a bank card recognition model by using the bank card recognition model training method of claims 1-7;
and identifying the bank card by using the trained bank card identification model.
9. A bank card recognition model training device is characterized by comprising:
the image processing module is used for processing the bank card image according to a preset resolution threshold value to obtain a high-resolution image and a low-resolution image;
the confrontation network model determining module is used for generating a super-resolution confrontation network model according to the bank card image, the high-resolution image and the low-resolution image;
and the recognition model training module is used for processing the training images of the bank card recognition model by using the super-resolution countermeasure network model and performing bank card recognition model training by using the processed training images.
10. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the method of any of claims 1 to 7 when executing the computer program.
11. A computer-readable storage medium, characterized in that it stores a computer program for executing the method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210251569.XA CN114581804A (en) | 2022-03-15 | 2022-03-15 | Bank card identification model training method and device and bank card identification method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210251569.XA CN114581804A (en) | 2022-03-15 | 2022-03-15 | Bank card identification model training method and device and bank card identification method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114581804A true CN114581804A (en) | 2022-06-03 |
Family
ID=81780832
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210251569.XA Pending CN114581804A (en) | 2022-03-15 | 2022-03-15 | Bank card identification model training method and device and bank card identification method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114581804A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116757231A (en) * | 2023-08-22 | 2023-09-15 | 北京紫光青藤微系统有限公司 | Method and device for generating super-resolution training atlas for bar code image |
-
2022
- 2022-03-15 CN CN202210251569.XA patent/CN114581804A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116757231A (en) * | 2023-08-22 | 2023-09-15 | 北京紫光青藤微系统有限公司 | Method and device for generating super-resolution training atlas for bar code image |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111652093B (en) | Text image processing method and device | |
| US11227182B2 (en) | Method, apparatus, and storage medium for recognizing image object | |
| US11507766B2 (en) | Method and apparatus for processing encoded pattern, storage medium, and electronic apparatus | |
| CN111524526B (en) | Voiceprint recognition method and voiceprint recognition device | |
| US20230245451A1 (en) | Background Audio Construction | |
| CN111105375A (en) | Image generation method, model training method and device thereof, and electronic equipment | |
| JP2008234479A (en) | Image quality improvement device, method, and program | |
| EP4300411A1 (en) | Training method and apparatus for image processing network, computer device, and storage medium | |
| CN114581804A (en) | Bank card identification model training method and device and bank card identification method and device | |
| US20220319215A1 (en) | Image processing method, device, electronic apparatus, and storage medium | |
| CN113781356A (en) | Training method of image denoising model, image denoising method, device and equipment | |
| CN116452706A (en) | Image generation method and device for presentation file | |
| CN102782705B (en) | Comprise the resolution adjustment of the image of the text of experience OCR process | |
| CN113469148B (en) | Text erasing method, model training method, device and storage medium | |
| CN111723653A (en) | Drawing book reading method and device based on artificial intelligence | |
| CN111353493B (en) | Text image direction correction method and device | |
| CN113850238B (en) | Document detection method and device, electronic equipment and storage medium | |
| CN112465092B (en) | Two-dimensional code sample generation method and device, server and storage medium | |
| CN111492408A (en) | Cartoon data display system, method, and program | |
| CN113610064B (en) | Handwriting recognition method and device | |
| KR102477177B1 (en) | Electronic device for providing editable document template corresponding to captured document and method for operating thereof | |
| CN113850805B (en) | Multi-document detection method and device, electronic equipment and storage medium | |
| JP7304772B2 (en) | Image analysis device, image analysis method, and program | |
| JP5246195B2 (en) | Portable information terminal | |
| CN111383165B (en) | Image processing method, system 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 |