CN113920318A - License plate image generation method and system and license plate image generator - Google Patents
License plate image generation method and system and license plate image generator Download PDFInfo
- Publication number
- CN113920318A CN113920318A CN202111385232.XA CN202111385232A CN113920318A CN 113920318 A CN113920318 A CN 113920318A CN 202111385232 A CN202111385232 A CN 202111385232A CN 113920318 A CN113920318 A CN 113920318A
- Authority
- CN
- China
- Prior art keywords
- license plate
- plate image
- oasis
- semantic
- image
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses a license plate image generation method, a license plate image generation system and a license plate image former, wherein the license plate image former comprises the following steps: acquiring a real license plate image, and constructing a semantic label corresponding to the real license plate image to obtain a training data set; training an OASIS generation network by using the training data set to obtain a first fixed parameter; outputting a first license plate image by an OASIS generation network under the first fixed parameter; training an OASIS semantic segmentation network by using the first license plate image and the real license plate image to obtain a second fixed parameter; taking the OASIS generation network under the second fixed parameter as a target OASIS generation network; acquiring a simulated license plate image, constructing a semantic label corresponding to the simulated license plate image to obtain a generated license plate data set, and inputting the generated license plate data set into the target OASIS generation network to obtain a target sample. The license plate sample data can be automatically generated, and the difficulty that the license plate sample is fussy to acquire and label is solved.
Description
Technical Field
The invention relates to the technical field of data processing, in particular to a method for generating a sample image.
Background
With the rapid development of deep learning technology, the end-to-end license plate recognition algorithm based on the neural network gradually replaces the license plate recognition algorithm based on the traditional machine learning. The method has the obvious advantages of good robustness, high accuracy, high processing speed and the like, but has the defect that a large number of license plate labeling samples need to be collected for training.
In order to obtain the license plate labeling data meeting the training requirements, a large amount of time, capital and manpower are consumed for collection and labeling. Firstly, the license plates are various in types, including small license plates, large double-layer license plates, new energy license plates, police license plates, military license plates, double-layer military license plates and the like, and the difficulty of sample collection is increased due to relatively rare license plates; secondly, the license plate content of each province and city is different, and the same repeated collection needs to be carried out in each region.
In order to reduce the cost, a computer image technology is generally used to artificially synthesize a license plate picture according to rules such as the color, font, character content and the like of a license plate, and the license plate picture is used as supplement of license plate labeling data. In the process of synthesizing the license plate picture, the synthesized license plate picture can simulate a real license plate picture as much as possible by adding common data enhancement technologies, such as random clipping, scaling, rotation, random brightness, contrast, chroma and color saturation, and increasing pollution, shading and the like. However, the difference between the distribution of the license plate picture synthesized by the artificial method and the distribution of the license plate picture data in a real scene is large, which often results in that the recognition model learns wrong knowledge and cannot mine the essential characteristics of the license plate picture data, so that the generalization capability of the model is reduced, and the expression effect on the real data is poor.
Disclosure of Invention
The invention provides a method and a system for automatically generating license plate sample data, which solve the difficult point of tedious license plate sample collection and labeling.
In order to solve the technical problem, the invention is solved by the following technical scheme:
a license plate image generation method comprises the following steps:
acquiring a real license plate image, and constructing a semantic label corresponding to the real license plate image to obtain a training data set;
training an OASIS generation network by using the training data set to obtain a first fixed parameter;
outputting a first license plate image by an OASIS generation network under the first fixed parameter;
training an OASIS semantic segmentation network by using the first license plate image and the real license plate image to obtain a second fixed parameter;
taking the OASIS generation network under the second fixed parameter as a target OASIS generation network;
acquiring a simulated license plate image, constructing a semantic label corresponding to the simulated license plate image to obtain a generated license plate data set, and inputting the generated license plate data set into the target OASIS generation network to obtain a target sample.
Optionally, the method for training the OASIS generation network with the training data set includes:
the training data set comprises semantic label data and license plate image data;
inputting the semantic tag data into an OASIS generation network, and outputting a first license plate image;
inputting the OASIS semantic segmentation network of the frozen network parameters into the first license plate image, and outputting a first license plate semantic segmentation graph;
and updating the OASIS generation network after loss calculation is carried out on the semantic label data and the first license plate semantic segmentation graph.
Optionally, the method for training the OASIS semantic segmentation network by using the first license plate image and the real license plate image includes:
the training data set comprises semantic label data and license plate image data;
inputting the semantic label data into an OASIS generation network with frozen network parameters, and outputting a second license plate image;
inputting the second license plate image into an OASIS semantic segmentation network and outputting a second license plate semantic segmentation image;
inputting the license plate image data into an OASIS semantic segmentation network, and outputting a second license plate image segmentation graph;
and respectively calculating the license plate image data and the second license plate image segmentation map, and the loss functions of the license plate image data and the second license plate image segmentation map, and updating the OASIS semantic segmentation network.
Optionally, the method further includes an iterative training method:
acquiring a real license plate image, constructing a semantic label corresponding to the real license plate image, and acquiring a training data set and a verification data set;
the verification data set comprises a verification license plate image and a verification license plate semantic label;
executing verification data set training every time a preset iteration number is passed, and verifying the OASIS generation network;
calculating the semantic tag of the license plate to be verified, and generating a third license plate image by the OASIS generation network;
and performing FID operation on the verification license plate image and the third license plate image until the FID value is stable, and finishing OASIS model training.
Optionally, the process of obtaining a simulated license plate semantic image, constructing a semantic label corresponding to the simulated license plate semantic image to obtain a generated license plate data set, inputting the generated license plate data set into the OASIS generation network, and obtaining a target sample further includes:
a, acquiring a semantic image of a simulated license plate, and constructing a semantic label corresponding to the semantic image of the simulated license plate to obtain N semantic images;
b, normalizing the N semantic images and inputting the normalized N semantic images into a trained OASIS generation network to obtain N license plate images corresponding to the semantic images one by one;
and performing the step a-b M times to obtain target samples of the M-N array.
Optionally, the method for acquiring a real license plate image and constructing a semantic label corresponding to the real license plate image includes:
acquiring original license plate images of different types;
correcting the original license plate image to obtain a license plate image;
and marking the license plate image according to the color parameter or the gray parameter, and constructing a semantic label corresponding to the license plate image.
The invention also provides a license plate image generation system, which comprises:
the system comprises a sample acquisition unit, a license plate image acquisition unit and a license plate image acquisition unit, wherein the sample acquisition unit is used for acquiring original license plate data and preprocessing the original license plate data to obtain license plate image data;
the semantic operation unit is used for constructing a semantic label corresponding to the license plate image data to obtain a training data set;
a first training unit, configured to train an OASIS generation network with the training data set to obtain a first fixed parameter;
the second training unit is used for training an OASIS semantic segmentation network by using the first license plate image and the real license plate image to obtain a second fixed parameter; taking the OASIS generation network under the second fixed parameter as a target OASIS generation network;
and the license plate generation unit is used for inputting a simulated license plate image, constructing a semantic label corresponding to the simulated license plate image to obtain a generated license plate data set, and inputting the generated license plate data set into the target OASIS generation network to obtain a target sample.
There is also a license plate image forming device, comprising:
the license plate input unit is used for inputting a simulated license plate image;
an arithmetic unit for calculating a target sample by the method;
and the output unit is used for outputting the target sample.
And the license plate simulator is used for outputting a simulated license plate image.
The invention has the beneficial effects that:
according to the technical scheme disclosed by the invention, the simulated license plate pictures with different contents and distributed approximate to real license plate data can be generated according to different input information, the license plate labeling samples can be effectively expanded, and the cost of manual labeling is reduced.
Compared with the traditional license plate simulation sample, the generated target sample is more real; and can generate license plates with any type and any content.
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, and 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 these drawings without creative efforts.
FIG. 1 is a general flowchart of a license plate image generation method;
FIG. 2 is a flow diagram of OASIS generation network training;
FIG. 3 is a flow diagram of OASIS semantic segmentation network training;
fig. 4 is a flowchart of the operation of the generator of embodiment 3.
Detailed Description
The present invention will be described in further detail with reference to examples, which are illustrative of the present invention and are not to be construed as being limited thereto.
Example 1:
a license plate image generation method is characterized by comprising the following steps:
preparing a training sample: acquiring a real license plate image, and constructing a semantic label corresponding to the real license plate image to obtain a training data set;
a simulation training stage: training an OASIS generation network by using the training data set to obtain a first fixed parameter;
outputting a first license plate image by an OASIS generation network under the first fixed parameter;
training an OASIS semantic segmentation network by using the first license plate image and the real license plate image to obtain a second fixed parameter;
taking the OASIS generation network under the second fixed parameter as a target OASIS generation network;
and (3) generating a license plate: acquiring a simulated license plate image, constructing a semantic label corresponding to the simulated license plate image to obtain a generated license plate data set, and inputting the generated license plate data set into the target OASIS generation network to obtain a target sample.
The method for acquiring the real license plate image and constructing the semantic label corresponding to the real license plate image comprises the following steps:
acquiring original license plate images of different types;
correcting the original license plate image to obtain a real license plate image;
and marking the license plate image according to the color parameter or the gray parameter, and constructing a semantic label corresponding to the license plate image.
In particular, the method comprises the following steps of,
1) collecting different types of original license plate images, including common single-layer license plates and double-layer license plates, wherein the single-layer license plates comprise new energy license plates (cars and trucks), blue license plates and yellow license plates (coaches and trailers), and the double-layer license plates comprise blue license plates, yellow license plates (trailers) and the like.
2) Correcting the collected original license plate image, specifically comprising the following steps: marking 4 corner points of the license plate for each original license plate image, and performing perspective transformation conversion according to coordinates of the 4 corner points and predefined correction corner point coordinates to obtain a corrected real license plate image.
3) And (3) performing pixel labeling on the real license plate image obtained in the step (1-1-2) according to the color, and constructing a semantic label corresponding to the real license plate image to obtain a training data set. Each main color of the license plate in the image is taken as a category of semantic labels, the semantic labels comprise Blue, Yellow, White, Black, Red and Green, and except that the Green is the gradient background color of the new energy license plate, the other labels are pure colors.
4) The labeled sample also comprises a verification data set which is divided into a training data set and a verification data set which is 4: 1.
The method for training the OASIS generation network by using the training data set comprises the following steps:
the training data set comprises semantic label data and license plate image data;
inputting the semantic tag data into an OASIS generation network, and outputting a first license plate image;
inputting the OASIS semantic segmentation network of the frozen network parameters into the first license plate image, and outputting a first license plate semantic segmentation graph;
and updating the OASIS generation network after loss calculation is carried out on the semantic label data and the first license plate semantic segmentation graph.
Specifically, as shown in fig. 2;
let, semantic tag data ITargetNumber plate image data IRealFirst license plate image IFake1The first license plate meaning segmentation drawing DFake1
Freezing network parameters of OASIS semantic segmentation network D, and labeling semantic data ITargetSending into OASIS generation network G to obtain first license plate image IFake1The first license plate image IFake1Sending the data into an OASIS semantic segmentation network D to obtain a first license semantic segmentation graph DFake1The first license plate image IFake1The first license plate meaning segmentation drawing DFake1And performing cross entropy loss calculation and back propagation to update network parameters of the OASIS semantic segmentation network.
The method for training the OASIS semantic segmentation network by using the first license plate image and the real license plate image comprises the following steps of:
the training data set comprises semantic label data and license plate image data;
inputting the semantic label data into an OASIS generation network with frozen network parameters, and outputting a second license plate image;
inputting the second license plate image into an OASIS semantic segmentation network and outputting a second license plate semantic segmentation image;
inputting the license plate image data into an OASIS semantic segmentation network, and outputting a second license plate image segmentation graph;
and respectively calculating the license plate image data and the second license plate image segmentation map, and the loss functions of the license plate image data and the second license plate image segmentation map, and updating the OASIS semantic segmentation network.
Specifically, as shown in fig. 3:
freezing the network parameters of OASIS generation network G, and using the semantic label data ITargetInputting the image into OASIS generation network G again to obtain a second license plate image IFake2And a second license plate image IFake2Inputting the semantic segmentation network D of OASIS to obtain a second license semantic segmentation graph DFake2；
Simultaneously converts the license plate image data IRealSimultaneously inputting OASIS semantic segmentation network D to obtain a second license plate image segmentation graph DReal. Respectively calculating a second license plate image IFake2And a second license meaning segmentation chart DFake2Cross entropy loss of (A), and license plate image data IRealSegmentation drawing D of second license plate imageRealAnd (4) performing cross entropy loss and back propagation to update the network parameters of the OASIS semantic segmentation network D.
The method comprises the following steps:
acquiring a real license plate image, constructing a semantic label corresponding to the real license plate image, and acquiring a training data set and a verification data set;
the verification data set comprises a verification license plate image and a verification license plate semantic label;
executing verification data set training every time a preset iteration number is passed, and verifying the OASIS generation network;
calculating the semantic tag of the license plate to be verified, and generating a third license plate image by the OASIS generation network;
and performing FID operation on the verification license plate image and the third generation license plate image until the FID value is stable, and finishing OASIS model training.
In particular, the method comprises the following steps of,
if, verify the license plate image IReal_valAnd verifying license plate semantic tag ITarget_val
In the iterative process of OASIS network training, the OASIS generation network G is verified in a verification data set every N iterative times, and the verification process is as follows: will verify the license plate semantic label ITarget_valSending the third license plate image into an OASIS generation network G to obtain a third license plate image IFake_valWill verify the license plate image IReal_valAnd a third license plate image IFake_valAnd performing FID calculation, calculating the FID of each pair of samples of the verification data set, taking an average value as the performance of the current model, and stopping training to obtain the trained OASIS model when the FID value of the model is not reduced in the iteration process.
The method comprises the following steps of obtaining a simulated license plate semantic image, constructing a semantic label corresponding to the simulated license plate semantic image to obtain a generated license plate data set, inputting the generated license plate data set into the OASIS generation network, and obtaining a target sample, wherein the process of obtaining the target sample further comprises the following steps:
a, acquiring a semantic image of a simulated license plate, and constructing a semantic label corresponding to the semantic image of the simulated license plate to obtain N semantic images;
b, normalizing the N semantic images and inputting the normalized N semantic images into a trained OASIS generation network to obtain N license plate images corresponding to the semantic images one by one;
and performing the step a-b M times to obtain target samples of the M-N array.
In particular, the method comprises the following steps of,
1) generating a semantic image which accords with a standard license plate structure through program simulation, wherein the semantic labels are consistent with the semantic labels in the step 1-1-3, and obtaining N semantic images;
2) normalizing the N generated semantic images into 384 multiplied by 128 and sending the normalized semantic images into an OASIS generation network to obtain N license plate images which correspond to the input semantic images one by one;
3) and circularly executing the step 1) and the step 2) M times to obtain M x N generated license plate images, namely outputting a target sample, wherein the target sample is a license plate sample required in the neural network training required by the scheme. The problem of license plate sample collection and marking difficulty is solved.
Example 2
A license plate image generation system comprising:
the system comprises a sample acquisition unit, a license plate image acquisition unit and a license plate image acquisition unit, wherein the sample acquisition unit is used for acquiring original license plate data and preprocessing the original license plate data to obtain license plate image data;
the semantic operation unit is used for constructing a semantic label corresponding to the license plate image data to obtain a training data set;
a first training unit, configured to train an OASIS generation network with the training data set to obtain a first fixed parameter;
the second training unit is used for training an OASIS semantic segmentation network by using the first license plate image and the real license plate image to obtain a second fixed parameter; taking the OASIS generation network under the second fixed parameter as a target OASIS generation network;
and the license plate generation unit is used for inputting a simulated license plate image, constructing a semantic label corresponding to the simulated license plate image to obtain a generated license plate data set, and inputting the generated license plate data set into the target OASIS generation network to obtain a target sample.
Example 3:
a license plate image generator comprising:
the license plate input unit is used for inputting a simulated license plate;
an arithmetic unit for calculating a target sample by the method disclosed in embodiment 1;
and the output unit is used for outputting the target sample.
The license plate simulator is connected with the license plate input unit.
Based on the license plate image generator, license plate samples can be produced in batch, and a large amount of time is not required to be spent for collecting and labeling license plate images.
In the several embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other ways. For example, the above-described device embodiments are merely illustrative, and for example, the division of the modules or units is only one logical functional division, and there may be other divisions when actually implemented, for example, a plurality of units or components may be combined or may be integrated into another device, or some features may be omitted, or not executed.
The units may or may not be physically separate, and components displayed as units may be one physical unit or a plurality of physical units, that is, may be located in one place, or may be distributed in a plurality of different places. 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 unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a readable storage medium. Based on such understanding, the technical solution of the embodiments of the present invention may be essentially or partially contributed to by the prior art, or all or part of the technical solution may be embodied in the form of a software product, where the software product is stored in a storage medium and includes several instructions to enable a device (which may be a single chip, a chip, or the like) or a processor (processor) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.
The above description is only an embodiment of the present invention, but the scope of the present invention is not limited thereto, and any changes or substitutions within the technical scope of the present invention are intended to be covered by the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the appended claims.
Claims (10)
1. A license plate image generation method is characterized by comprising the following steps:
acquiring a real license plate image, and constructing a semantic label corresponding to the real license plate image to obtain a training data set;
training an OASIS generation network by using the training data set to obtain a first fixed parameter;
outputting a first license plate image by an OASIS generation network under the first fixed parameter;
training an OASIS semantic segmentation network by using the first license plate image and the real license plate image to obtain a second fixed parameter;
taking the OASIS generation network under the second fixed parameter as a target OASIS generation network;
acquiring a simulated license plate image, constructing a semantic label corresponding to the simulated license plate image to obtain a generated license plate data set, and inputting the generated license plate data set into the target OASIS generation network to obtain a target sample.
2. The license plate image generation method according to claim 1, wherein the method of training an OASIS generation network with the training data set includes:
the training data set comprises semantic label data and license plate image data;
inputting the semantic tag data into an OASIS generation network, and outputting a first license plate image;
inputting the OASIS semantic segmentation network of the frozen network parameters into the first license plate image, and outputting a first license plate semantic segmentation graph;
and updating the OASIS generation network after loss calculation is carried out on the semantic label data and the first license plate semantic segmentation graph.
3. The license plate image generation method of claim 1, wherein the method for training the OASIS semantic segmentation network with the first license plate image and the real license plate image comprises:
the training data set comprises semantic label data and license plate image data;
inputting the semantic label data into an OASIS generation network with frozen network parameters, and outputting a second license plate image;
inputting the second license plate image into an OASIS semantic segmentation network and outputting a second license plate semantic segmentation image;
inputting the license plate image data into an OASIS semantic segmentation network, and outputting a second license plate image segmentation graph;
and respectively calculating the license plate image data and the second license plate image segmentation map, and the loss functions of the license plate image data and the second license plate image segmentation map, and updating the OASIS semantic segmentation network.
4. The license plate image generation method of claim 1, further comprising an iterative training method:
acquiring a real license plate image, constructing a semantic label corresponding to the real license plate image, and acquiring a training data set and a verification data set;
the verification data set comprises a verification license plate image and a verification license plate semantic label;
executing verification data set training every time a preset iteration number is passed, and verifying the OASIS generation network;
calculating the semantic tag of the license plate to be verified, and generating a third license plate image by the OASIS generation network;
and performing FID operation on the verification license plate image and the third license plate image until the FID value is stable, and finishing OASIS model training.
5. The license plate image generation method of claim 1, wherein obtaining a simulated license plate semantic image, constructing a semantic label corresponding to the simulated license plate semantic image to obtain a generated license plate dataset, inputting the generated license plate dataset into the OASIS generation network, and obtaining a target sample further comprises:
a, acquiring a semantic image of a simulated license plate, and constructing a semantic label corresponding to the semantic image of the simulated license plate to obtain N semantic images;
b, normalizing the N semantic images and inputting the normalized N semantic images into a trained OASIS generation network to obtain N license plate images corresponding to the semantic images one by one;
and performing the step a-b M times to obtain target samples of the M-N array.
6. The license plate image generation method of claim 1, wherein the method of obtaining a real license plate image and constructing a semantic label corresponding to the real license plate image comprises:
acquiring original license plate images of different types;
correcting the original license plate image to obtain a license plate image;
and marking the license plate image according to the color parameter or the gray parameter, and constructing a semantic label corresponding to the license plate image.
7. A license plate image generation system, comprising:
the system comprises a sample acquisition unit, a license plate image acquisition unit and a license plate image acquisition unit, wherein the sample acquisition unit is used for acquiring original license plate data and preprocessing the original license plate data to obtain license plate image data;
the semantic operation unit is used for constructing a semantic label corresponding to the license plate image data to obtain a training data set;
a first training unit, configured to train an OASIS generation network with the training data set to obtain a first fixed parameter;
the second training unit is used for training an OASIS semantic segmentation network by using the first license plate image and the real license plate image to obtain a second fixed parameter; taking the OASIS generation network under the second fixed parameter as a target OASIS generation network;
and the license plate generation unit is used for inputting a simulated license plate image, constructing a semantic label corresponding to the simulated license plate image to obtain a generated license plate data set, and inputting the generated license plate data set into the target OASIS generation network to obtain a target sample.
8. A license plate image creator, comprising:
the license plate input unit is used for inputting a simulated license plate image;
an arithmetic unit for calculating a target sample by the method of any one of claims 1 to 6;
and the output unit is used for outputting the target sample.
9. The license plate image generation system of claim 8, further comprising a license plate simulator for outputting a simulated license plate image.
10. A computer storage medium storing a computer program that is called by a processor to implement the license plate image generation method according to any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111385232.XA CN113920318A (en) | 2021-11-22 | 2021-11-22 | License plate image generation method and system and license plate image generator |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111385232.XA CN113920318A (en) | 2021-11-22 | 2021-11-22 | License plate image generation method and system and license plate image generator |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113920318A true CN113920318A (en) | 2022-01-11 |
Family
ID=79247800
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111385232.XA Pending CN113920318A (en) | 2021-11-22 | 2021-11-22 | License plate image generation method and system and license plate image generator |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113920318A (en) |
-
2021
- 2021-11-22 CN CN202111385232.XA patent/CN113920318A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111292264B (en) | Image high dynamic range reconstruction method based on deep learning | |
| CN110111236B (en) | Multi-target sketch image generation method based on progressive confrontation generation network | |
| CN110414519A (en) | A kind of recognition methods of picture character and its identification device | |
| US20150310305A1 (en) | Learning painting styles for painterly rendering | |
| CN113505772B (en) | License plate image generation method and system based on generation countermeasure network | |
| CN112070137B (en) | Training data set generation method, target object detection method and related equipment | |
| CN106855996B (en) | Gray-scale image coloring method and device based on convolutional neural network | |
| CN112884758B (en) | Defect insulator sample generation method and system based on style migration method | |
| CN112287941B (en) | License plate recognition method based on automatic character region perception | |
| CN111027538A (en) | Container detection method based on instance segmentation model | |
| US10922852B2 (en) | Oil painting stroke simulation using neural network | |
| CN110689495A (en) | Image restoration method for deep learning | |
| CN110298898B (en) | Method for changing color of automobile image body and algorithm structure thereof | |
| CN106326895A (en) | Image processing device and image processing method | |
| CN113392702A (en) | Target identification method based on self-adaptive image enhancement under low-light environment | |
| CN110569379A (en) | Method for manufacturing picture data set of automobile parts | |
| CN114581646A (en) | Text recognition method and device, electronic equipment and storage medium | |
| CN115270184A (en) | Video desensitization method, vehicle video desensitization method and vehicle-mounted processing system | |
| CN116030453A (en) | Digital ammeter identification method, device and equipment | |
| CN117689550A (en) | Low-light image enhancement method and device based on progressive generation countermeasure network | |
| CN104915641A (en) | Method for obtaining face image light source orientation based on android platform | |
| CN115713585B (en) | Texture image reconstruction method, apparatus, computer device and storage medium | |
| CN113920318A (en) | License plate image generation method and system and license plate image generator | |
| Wu et al. | DHGAN: Generative adversarial network with dark channel prior for single‐image dehazing | |
| CN115187977A (en) | License plate data set dynamic creation method and system and electronic equipment |
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 |