CN115424250A - License plate recognition method and device - Google Patents
License plate recognition method and device Download PDFInfo
- Publication number
- CN115424250A CN115424250A CN202110518506.1A CN202110518506A CN115424250A CN 115424250 A CN115424250 A CN 115424250A CN 202110518506 A CN202110518506 A CN 202110518506A CN 115424250 A CN115424250 A CN 115424250A
- Authority
- CN
- China
- Prior art keywords
- license plate
- image
- recognized
- determining
- recognition
- 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
-
- 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
Abstract
The invention provides a license plate recognition method and device. The method comprises the following steps: determining the confidence of each license plate type corresponding to the license plate image to be recognized; determining the license plate type corresponding to the license plate image to be recognized according to the confidence coefficient; and determining license plate characters corresponding to the license plate image to be recognized according to the license plate character prediction branch corresponding to the license plate type. The license plate recognition method and the license plate recognition device can improve the accuracy of license plate recognition and can recognize different types of license plate recognition; and the recognition of characters of multiple types of license plates can be realized, so that the applicability of a license plate recognition network is improved.
Description
Technical Field
The invention relates to the technical field of image recognition, in particular to a license plate recognition method and device.
Background
The license plate recognition means that the position of the license plate of a vehicle is accurately found in an image containing the vehicle, and all Chinese characters, characters and numbers on the license plate are recognized. The main methods of the existing license plate recognition technology can be roughly divided into two types, namely a traditional method based on image processing and a method based on a deep learning technology.
The traditional license plate recognition method is generally divided into three parts: license plate positioning, character segmentation and character recognition. The traditional method mainly depends on manual feature extraction, and the license plate recognition is completed through an image processing technology. When the license plate in the image has the bad factors which are not beneficial to characteristic identification, such as unclear brightness, uneven illumination, inclination, large scale change and the like, the false detection probability of the traditional method is greatly increased.
The license plate recognition technology based on the deep learning technology utilizes the feature extraction capability and the self-learning capability of a deep convolutional neural network to enable a network model to have anti-interference capability and adaptive capability, but the two parts of the existing license plate positioning and license plate character recognition are usually realized separately, and the feature extraction processes of the two parts are carried out separately, so that the integral license plate recognition effect is influenced; and only aiming at the characteristics of a certain type of license plate, a corresponding network model is constructed, and the method has no wide adaptability.
Disclosure of Invention
The invention provides a license plate recognition method and a license plate recognition device, which are used for solving the technical problems of complexity, low recognition accuracy and single recognition type of a license plate recognition method in the prior art.
The invention provides a license plate recognition method, which comprises the following steps:
determining the confidence of each license plate type corresponding to the license plate image to be recognized;
determining the license plate type corresponding to the license plate image to be recognized according to the confidence coefficient;
and determining license plate characters corresponding to the license plate image to be recognized according to the license plate character prediction branch corresponding to the license plate type.
In one embodiment, the determining the confidence of each license plate type corresponding to the license plate image to be recognized includes:
inputting the license plate image to be recognized into a trained license plate recognition network, and acquiring a corresponding feature map and a prediction result of a license plate position;
mapping the prediction result to the corresponding position of the characteristic diagram to obtain an interested area;
and determining the confidence of each license plate type corresponding to the features in the region of interest.
In one embodiment, the determining the license plate characters corresponding to the license plate image to be recognized according to the license plate character prediction branch corresponding to the license plate type includes:
obtaining a character recognition result of the region of interest through a classifier in the license plate character prediction branch;
and combining the character recognition results to determine the license plate characters.
In one embodiment, the license plate recognition network is trained by:
determining a license plate data set and license plate marking information as a first training set;
acquiring a pre-training model according to the first training set;
generating the license plate recognition network and corresponding network model parameters according to the license plate data set and the license plate sequence information based on the pre-training model;
wherein the license plate data set is constructed based on a vehicle sampling image; the license plate labeling information comprises: and license plate position information and license plate type information corresponding to the vehicle sampling image.
In one embodiment, the license plate recognition network is trained by:
and under the condition that the number of iterations of the license plate recognition network is smaller than a preset number of iterations, updating the weight corresponding to the network model parameter until the loss function value of the license plate recognition network is smaller than a preset threshold value.
In one embodiment, the inputting the license plate image to be recognized into a trained license plate recognition network to obtain a corresponding feature map and a prediction result of a license plate position includes:
inputting the license plate image to be recognized into the trained license plate recognition network, and extracting the corresponding features of the license plate image to be recognized through convolution operation and concat stacking;
and fusing the characteristic layer of the characteristic and the characteristic layer of the residual volume block to obtain the fused characteristic graph and the prediction result.
The invention provides a license plate recognition device, comprising:
the first determining module is used for determining the confidence coefficient of each license plate type corresponding to the license plate image to be recognized;
the second determining module is used for determining the license plate type corresponding to the license plate image to be recognized according to the confidence coefficient;
and the third determining module is used for determining the license plate characters corresponding to the license plate image to be recognized according to the license plate character prediction branch corresponding to the license plate type.
In one embodiment, the first determining module is configured to input the license plate image to be recognized into a trained license plate recognition network, and obtain a prediction result of a corresponding feature map and a license plate position;
mapping the prediction result to the corresponding position of the characteristic diagram to obtain an interested area;
and determining the confidence of each license plate type corresponding to the features in the region of interest.
The invention provides electronic equipment, which comprises a memory and a memory stored with a computer program, wherein when the processor executes the program, the steps of any one of the license plate recognition methods are realized.
The present invention provides a processor-readable storage medium having stored thereon a computer program for causing a processor to execute the steps of any of the above-described license plate recognition methods.
According to the license plate recognition method and device, the license plate type corresponding to the image to be recognized is confirmed by determining the confidence, so that the license plate recognition accuracy can be improved, and different types of license plate recognition can be recognized; and the corresponding license plate characters are determined through the license plate character prediction branch corresponding to the license plate type, so that the recognition of the multi-type license plate characters is realized, and the applicability of a license plate recognition network is improved.
Drawings
In order to more clearly illustrate the technical solutions of the present invention or the prior art, the drawings needed for 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 some embodiments of the present invention, and those skilled in the art can also obtain other drawings according to the drawings without creative efforts.
FIG. 1 is a schematic flow chart of a license plate recognition method according to the present invention;
FIG. 2 is a schematic diagram of a license plate character prediction branch of the license plate recognition method provided by the present invention;
FIG. 3 is a schematic diagram of an overall network of a license plate recognition method provided by the present invention;
FIG. 4 is a flowchart illustrating an embodiment of a license plate recognition method according to the present invention;
FIG. 5 is a schematic structural diagram of a license plate recognition device provided in the present invention;
fig. 6 is a schematic structural diagram of an electronic device provided in the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is obvious that the described embodiments are some, but not all embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Fig. 1 is a schematic flow diagram of a license plate recognition method provided by the present invention, and referring to fig. 1, the license plate recognition method provided by the present invention includes:
s110, determining confidence coefficients of various license plate types corresponding to the license plate image to be recognized;
s120, determining the license plate type corresponding to the license plate image to be recognized according to the confidence coefficient;
s130, determining license plate characters corresponding to the license plate image to be recognized according to the license plate character prediction branch corresponding to the license plate type.
The execution main body of the license plate recognition method provided by the invention is a license plate recognition system applied to different scenes or complex environments, and the license plate recognition system can be computer equipment. The application scenario can be monitoring alarm, speeding violation punishment, vehicle access management, highway toll management and the like, and the invention is not particularly limited.
Optionally, in step S110, each license plate type may be a blue license plate, a new energy license plate, a double-row license plate, or other types of license plates, which is not limited in the present invention. And the license plate recognition system determines the confidence corresponding to each license plate type of the license plate image to be recognized according to the input license plate image to be recognized. The confidence level is used for representing the confidence level of a certain object, and in general, the greater the value of the confidence level, the higher the confidence level is; conversely, a smaller confidence value indicates a lower confidence.
In step S120, according to the confidence level of the license plate image to be recognized corresponding to each license plate type, the license plate type with the highest confidence level is selected and determined as the license plate type corresponding to the license plate image to be recognized.
In step S130, for each license plate type, a corresponding license plate character prediction branch is designed. And the license plate characters corresponding to the license plate image to be recognized can be determined through the license plate character prediction branch.
According to the license plate recognition method and device, the license plate type corresponding to the image to be recognized is confirmed by determining the confidence coefficient, so that the license plate recognition accuracy can be improved, and different types of license plate recognition can be recognized; and the corresponding license plate characters are determined through the license plate character prediction branch corresponding to the license plate type, so that the recognition of the multi-type license plate characters is realized, and the applicability of a license plate recognition network is improved.
In one embodiment, the determining the confidence of each license plate type corresponding to the license plate image to be recognized includes:
inputting the license plate image to be recognized into a trained license plate recognition network, and acquiring a corresponding feature map and a prediction result of a license plate position;
mapping the prediction result to the corresponding position of the characteristic diagram to obtain an interested area;
and determining the confidence of each license plate type corresponding to the features in the region of interest.
Optionally, the license plate recognition system may input the image with the recognized license plate to a trained license plate recognition network, and may obtain a corresponding feature map and a prediction result of the license plate position through feature extraction; and mapping the prediction result of the license plate position to a corresponding position in the characteristic diagram to obtain the region of interest. Wherein the region of interest is an image region selected from the image, which is the focus of interest for image analysis, and which is delineated for further processing. And determining the confidence degree of the license plate image to be recognized corresponding to each license plate type according to the features in the region of interest in the feature map.
According to the license plate recognition method provided by the invention, the confidence degrees of the license plate image to be recognized corresponding to different license plate types are determined through the region of interest obtained by mapping, so that the accuracy of license plate recognition is improved.
In one embodiment, the determining license plate characters corresponding to the license plate image to be recognized according to the license plate character prediction branch corresponding to the license plate type includes:
obtaining a character recognition result of the region of interest through a classifier in the license plate character prediction branch;
and combining the character recognition results to determine the license plate characters.
Optionally, the license plate recognition system performs feature data processing and output dimension adjustment on the features in the region of interest through a full connection layer and a reshape layer of the license plate recognition network, and transmits the feature data and the output dimension adjustment to the corresponding license plate character prediction branch according to the license plate type. The license plate character prediction branch can be a 1x1x 7-dimensional blue license plate prediction branch, a 1x1x 8-dimensional new energy license plate prediction branch and a 1x1x 7-dimensional double-row license plate prediction branch, and the invention is not particularly limited.
Optionally, the blue license plate, the new energy license plate and the double-row license plate are three multi-task license plate character prediction branches, the blue license plate is 7-character recognition, the new energy license plate is 8-character recognition, and the double-row license plate is also 7 characters. The network structure of the three license plate character prediction branches is consistent, and only the difference exists in the recognition number of characters. The blue license plate character prediction branch is taken as an example for detailed description, and referring to fig. 2, fig. 2 is a schematic diagram of the license plate character prediction branch of the license plate recognition method provided by the invention. The partial network is composed of 7 independent classification tasks, and the character recognition results are obtained through 7 classifiers responsible for respective positions of the license plate region of interest obtained through mapping. And finally combining the output results of the 7 classifiers to obtain a final identification result. The scheme adopts a Softmax classifier to be responsible for 7 classification tasks. And performing character recognition on each classification task and the corresponding classification parameters according to a preset rule, wherein the invention is not particularly limited.
Each classification task and its corresponding classification parameters are shown in the table:
| subtask | Corresponding position | Classification parameter |
| Task1 | First Chinese character recognition | 32 |
| Task2 | Second letter identification | 27 |
| Task3 | Third character (letter and number) recognition | 35 |
| Task4 | Fourth character (letter and number) recognition | 35 |
| Task5 | Fifth character (letter and number) recognition | 35 |
| Task6 | Sixth character (letter and number) recognition | 35 |
| Task7 | Seventh character (letter and number) recognition | 35 |
According to the license plate recognition method provided by the invention, different license plate character prediction branches are designed aiming at different types of license plates, so that the character recognition of multiple types of license plates is realized, the end-to-end multiple types of license plates are simultaneously realized, and the applicability of a license plate recognition network is improved.
In one embodiment, the license plate recognition network is trained by:
determining a license plate data set and license plate marking information as a first training set;
acquiring a pre-training model according to the first training set;
generating the license plate recognition network and corresponding network model parameters according to the license plate data set and the license plate sequence information based on the pre-training model;
the license plate data set is constructed based on a vehicle sampling image; the license plate labeling information comprises: and license plate position information and license plate type information corresponding to the vehicle sampling image.
Optionally, the license plate data set is from a vehicle image in a monitoring video, and a sampling period is set for sampling, so that three types of vehicle images including a blue single-line license plate, a new energy license plate and a double-line yellow license plate can be obtained. And the license plate recognition system marks the collected data set, and the marked data are used for marking the coordinate position, the license plate type and the license plate serial number of the license plate in each image and storing the marked data in a text form. And then enhancing the image data, wherein the enhancement strategy comprises the following steps: offset transform, contrast transform, color dither, noise dither, etc., while scaling the data to a uniform size image.
Optionally, the invention adopts a staged training mode to train the license plate recognition network, and the specific process is as follows:
stage1: and training license plate positioning and recognition branches. And training the recognition of the position and the type of the license plate by using the built license plate data set and a labeling file containing the position and the type of the license plate to obtain a license plate recognition Model _ A at the stage.
Stage2: and training an end-to-end license plate recognition network. In the training Stage, the Model _ A trained by Stage1 is used as a pre-training Model, the same license plate data set is used, and license plate sequence information is added for training. And obtaining a final network Model _ B. Training is performed on the basis of Model _ A. And the license plate recognition system is trained according to the established license plate data set to finally obtain a license plate recognition network and corresponding network model parameters.
According to the license plate identification method provided by the invention, end-to-end multi-type license plate identification is realized by constructing an integrated network structure integrating license plate position, license plate type and license plate sequence identification.
In one embodiment, the license plate recognition network is trained by:
and under the condition that the iteration times of the license plate recognition network are smaller than the preset iteration times, updating the weight corresponding to the network model parameter until the loss function value of the license plate recognition network is smaller than a preset threshold value.
Optionally, the same optimization strategy is adopted for Stage1 and Stage2, and the optimization strategy is as follows: when the number of iterations of the license plate recognition network is smaller than a preset number of iterations, updating the weight of the network model parameter by using an Adam optimization method until the loss function value of the license plate recognition network is lower than a preset threshold value; and finishing the training when the error of the license plate recognition network is smaller than a preset threshold value or the iteration times of the license plate recognition network is larger than the preset iteration times.
The preset iteration number may be an Epoch, and the Epoch may be set according to the accuracy requirement. When a complete data set passes through the neural network once and back once, the process is called an Epoch. I.e. all training samples have been propagated in the neural network in one forward and one backward direction.
Optionally, the loss function part of the model training of the license plate recognition network mainly includes license plate position loss, license plate category loss, confidence loss and license plate character loss.
The license plate position loss function formula is as follows:
where d represents the euclidean distance between the center points of the prediction frame and the actual frame, c represents the diagonal distance of the minimum closure area that can contain both the prediction frame and the actual frame, w represents the width of the prediction frame, h represents the width of the prediction frame, and IoU represents the intersection ratio of the prediction frame and the actual frame.
It should be noted that the prediction result of the license plate position includes the length and width of the prediction frame, a center point coordinate, and the confidence of each prediction frame.
The license plate category loss function is:
wherein the license plate classification loss function is a cross entropy loss function, p i (c) The probability of a certain license plate is predicted in the ith prediction box,
the license plate label is actually framed.
The confidence loss function is:
wherein the confidence coefficient is a binary cross entropy loss function, C i And predicting the probability of containing the license plate for the ith prediction frame.
The license plate character loss function is a cross entropy loss function, and is specifically represented as:
the task _ classes are categories of corresponding character recognition tasks, and the classification parameters of different tasks corresponding to different character categories include: 27. 32 or 35, etc. p is a radical of i (t) predicting the probability of a character in the ith prediction box for the network,
is actually a character label.
The overall loss function is:
the license plate identification method provided by the invention has the advantages that the detection and identification performance of the license plate identification network is effectively improved by carrying out combined optimization on the position and the type of the license plate and the license plate sequence information.
In one embodiment, the inputting the license plate image to be recognized into a trained license plate recognition network to obtain a corresponding feature map and a prediction result of a license plate position includes:
inputting the license plate image to be recognized into the trained license plate recognition network, and extracting the corresponding features of the license plate image to be recognized through convolution operation and concat stacking;
and fusing the feature layer of the feature and the feature layer of the residual volume block to obtain the fused feature map and the prediction result.
Optionally, the license plate recognition network is built on the basis of yolov4 network, and includes: a darknet convolutional layer, a residual convolutional block, a pooling layer, and a full-link layer. Fig. 3 is a schematic overall network diagram of the license plate recognition method provided by the invention. Referring to fig. 3, the license plate recognition network model is built by the following steps:
firstly, inputting a preprocessed vehicle image with a fixed size, continuously compressing the length and the width of the final vehicle image through one-time darknet convolution and processing of residual convolution blocks layer by layer, and extracting the characteristics of the vehicle image layer by layer to be recognized. And performing convolution operation for a plurality of times, such as 3 times, on the features after passing through the residual convolution block 5. The input feature layer is then subjected to maximum pooling of three different sizes of convolution kernels at the pooling level, such as using the commonly used 13x13, 9x9, 5x5 sizes of convolution kernels. And performing concat stacking on the pooling results, and performing convolution for multiple times to increase the receptive field and separate out the context characteristics. And then performing convolution and upsampling processing on the context features to enlarge the feature image. And performing concat stacking on the first up-sampling result and the feature layer of the residual convolution block 4 of the backbone network to realize feature fusion. And after fusion, performing convolution and upsampling again to obtain a second upsampling result. And stacking and fusing the second up-sampling result and the feature layer concat of the residual volume block 3 to obtain a fused feature map. It should be noted that, because the number plate is smaller than the whole vehicle image, the extraction of the bottom layer features is more beneficial to the detection of small targets.
Optionally, the license plate position prediction is output by passing the second upsampled result through two different convolutional layers. For example, a commonly used convolution layer with a size of 3x3 and 1x1 is used to output the prediction result of the license plate position.
According to the license plate prediction recognition method provided by the invention, the characteristics with the license plate information are extracted by fusing the characteristics of the high layer and the bottom layer, and the characteristics of the license plate area are highlighted, so that the license plate recognition is more comprehensive and accurate.
In one embodiment, fig. 4 is a schematic flowchart of an embodiment of a license plate recognition method provided by the present invention; referring to fig. 4, the license plate recognition method comprises the following steps:
step 1: collecting data and constructing a license plate data set;
step 2: building a license plate recognition network;
and step 3: training a license plate recognition network;
and 4, step 4: and inputting the license plate image to be recognized into a trained license plate recognition network to obtain the position of the license plate, the type of the license plate and the characters of the license plate.
The invention also provides a license plate recognition device, which can be correspondingly referred to the license plate recognition method.
Fig. 5 is a schematic structural diagram of a license plate recognition device provided in the present invention, and as shown in fig. 5, the device includes:
the first determining module 510 is configured to determine confidence levels of license plate types corresponding to the license plate image to be recognized;
the second determining module 520 is configured to determine the license plate type corresponding to the license plate image to be recognized according to the confidence;
and a third determining module 530, configured to determine license plate characters corresponding to the license plate image to be recognized according to the license plate character prediction branch corresponding to the license plate type.
According to the license plate recognition device, the license plate type corresponding to the image to be recognized is confirmed by determining the confidence coefficient, so that the license plate recognition accuracy can be improved, and different types of license plate recognition can be recognized; and the corresponding license plate characters are determined through the license plate character prediction branch corresponding to the license plate type, so that the recognition of the multi-type license plate characters is realized, and the applicability of a license plate recognition network is improved.
In an embodiment, the first determining module 510 is further specifically configured to:
inputting the license plate image to be recognized into a trained license plate recognition network, and acquiring a corresponding feature map and a prediction result of a license plate position;
mapping the prediction result to the corresponding position of the characteristic diagram to obtain an interested area;
and determining the confidence of each license plate type corresponding to the features in the region of interest.
In an embodiment, the first determining module 510 is further specifically configured to:
obtaining a character recognition result of the region of interest through a classifier in the license plate character prediction branch;
and combining the character recognition results to determine the license plate characters.
In one embodiment, the license plate recognition network is trained by:
determining a license plate data set and license plate marking information as a first training set;
acquiring a pre-training model according to the first training set;
generating the license plate recognition network and corresponding network model parameters according to the license plate data set and the license plate sequence information based on the pre-training model;
the license plate data set is constructed based on a vehicle sampling image; the license plate labeling information comprises: and license plate position information and license plate type information corresponding to the vehicle sampling image.
In one embodiment, the license plate recognition network is trained by:
and under the condition that the number of iterations of the license plate recognition network is smaller than a preset number of iterations, updating the weight corresponding to the network model parameter until the loss function value of the license plate recognition network is smaller than a preset threshold value.
In an embodiment, the first determining module 510 is further specifically configured to:
inputting the license plate image to be recognized into the trained license plate recognition network, and extracting the corresponding features of the license plate image to be recognized through convolution operation and concat stacking;
and fusing the characteristic layer of the characteristic and the characteristic layer of the residual volume block to obtain the fused characteristic graph and the prediction result.
Fig. 6 illustrates a schematic structural diagram of an electronic device, which may include, as shown in fig. 6: a processor (processor) 610, a Communication Interface (Communication Interface) 620, a memory (memory) 630 and a Communication bus 640, wherein the processor 610, the Communication Interface 620 and the memory 630 complete Communication with each other through the Communication bus 640. The processor 610 may invoke computer programs in the memory 630 to perform the steps of the license plate recognition method, including, for example:
determining the confidence of each license plate type corresponding to the license plate image to be recognized;
determining the license plate type corresponding to the license plate image to be recognized according to the confidence coefficient;
and determining license plate characters corresponding to the license plate image to be recognized according to the license plate character prediction branch corresponding to the license plate type.
In addition, the logic instructions in the memory 630 may be implemented in the form of software functional units and stored in a computer readable storage medium when the logic instructions are sold or used as independent products. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) 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: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
In another aspect, the present invention also provides a computer program product, the computer program product comprising a computer program stored on a non-transitory computer-readable storage medium, the computer program comprising program instructions, which when executed by a computer, enable the computer to perform the license plate recognition method provided by the above methods, the method comprising:
determining the confidence of each license plate type corresponding to the license plate image to be recognized;
determining the license plate type corresponding to the license plate image to be recognized according to the confidence coefficient;
and determining license plate characters corresponding to the license plate image to be recognized according to the license plate character prediction branch corresponding to the license plate type.
On the other hand, an embodiment of the present application further provides a processor-readable storage medium, where the processor-readable storage medium stores a computer program, where the computer program is configured to cause the processor to execute the method provided in each of the foregoing embodiments, for example, the method includes:
determining the confidence of each license plate type corresponding to the license plate image to be recognized;
determining the license plate type corresponding to the license plate image to be recognized according to the confidence coefficient;
and determining license plate characters corresponding to the license plate image to be recognized according to the license plate character prediction branch corresponding to the license plate type.
The processor-readable storage medium can be any available medium or data storage device that can be accessed by a processor, including, but not limited to, magnetic memory (e.g., floppy disks, hard disks, magnetic tape, magneto-optical disks (MOs), etc.), optical memory (e.g., CDs, DVDs, BDs, HVDs, etc.), and semiconductor memory (e.g., ROMs, EPROMs, EEPROMs, non-volatile memories (NAND FLASH), solid State Disks (SSDs)), etc.
The above-described embodiments of the apparatus are merely illustrative, and 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 modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware. Based on the understanding, the above technical solutions substantially or otherwise contributing to the prior art may be embodied in the form of a software product, which may be stored in a computer-readable storage medium, such as ROM/RAM, magnetic disk, optical disk, etc., and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method according to the various embodiments or some parts of the embodiments.
Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, and not to limit it; 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; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (10)
1. A license plate recognition method is characterized by comprising the following steps:
determining the confidence of each license plate type corresponding to the license plate image to be recognized;
determining the license plate type corresponding to the license plate image to be recognized according to the confidence coefficient;
and determining license plate characters corresponding to the license plate image to be recognized according to the license plate character prediction branch corresponding to the license plate type.
2. The license plate recognition method of claim 1, wherein the determining the confidence level of each license plate type corresponding to the license plate image to be recognized comprises:
inputting the license plate image to be recognized into a trained license plate recognition network, and acquiring a corresponding feature map and a prediction result of a license plate position;
mapping the prediction result to the corresponding position of the characteristic diagram to obtain an interested area;
and determining the confidence of each license plate type corresponding to the features in the region of interest.
3. The license plate recognition method of claim 2, wherein the determining license plate characters corresponding to the license plate image to be recognized according to the license plate character prediction branch corresponding to the license plate type comprises:
obtaining a character recognition result of the region of interest through a classifier in the license plate character prediction branch;
and combining the character recognition results to determine the license plate characters.
4. The license plate recognition method of claim 1, wherein the license plate recognition network is trained by:
determining a license plate data set and license plate marking information as a first training set;
acquiring a pre-training model according to the first training set;
generating the license plate recognition network and corresponding network model parameters according to the license plate data set and the license plate sequence information based on the pre-training model;
wherein the license plate data set is constructed based on a vehicle sampling image; the license plate labeling information comprises: and the vehicle sampling image corresponds to the license plate position information and the license plate type information.
5. The license plate recognition method of claim 4, further comprising:
and under the condition that the iteration times of the license plate recognition network are smaller than the preset iteration times, updating the weight corresponding to the network model parameter until the loss function value of the license plate recognition network is smaller than a preset threshold value.
6. The license plate recognition method of claim 2, wherein the step of inputting the license plate image to be recognized into a trained license plate recognition network to obtain a corresponding feature map and a prediction result of a license plate position comprises the steps of:
inputting the license plate image to be recognized into the trained license plate recognition network, and extracting the corresponding features of the license plate image to be recognized through convolution operation and concat stacking;
and fusing the feature layer of the feature and the feature layer of the residual volume block to obtain the fused feature map and the prediction result.
7. A license plate recognition device, comprising:
the first determining module is used for determining the confidence coefficient of each license plate type corresponding to the license plate image to be recognized;
the second determining module is used for determining the license plate type corresponding to the license plate image to be recognized according to the confidence coefficient;
and the third determining module is used for determining license plate characters corresponding to the license plate image to be recognized according to the license plate character prediction branch corresponding to the license plate type.
8. The license plate recognition device of claim 7,
the first determining module is used for inputting the license plate image to be recognized into a trained license plate recognition network and acquiring a corresponding feature map and a prediction result of a license plate position;
mapping the prediction result to the corresponding position of the characteristic diagram to obtain an interested area;
and determining the confidence of each license plate type corresponding to the features in the region of interest.
9. An electronic device comprising a processor and a memory storing a computer program, wherein the processor implements the steps of the license plate recognition method according to any one of claims 1 to 6 when executing the computer program.
10. A processor-readable storage medium, characterized in that the processor-readable storage medium stores a computer program for causing a processor to execute the steps of the license plate recognition method according to any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110518506.1A CN115424250A (en) | 2021-05-12 | 2021-05-12 | License plate recognition method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110518506.1A CN115424250A (en) | 2021-05-12 | 2021-05-12 | License plate recognition method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115424250A true CN115424250A (en) | 2022-12-02 |
Family
ID=84230458
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110518506.1A Pending CN115424250A (en) | 2021-05-12 | 2021-05-12 | License plate recognition method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115424250A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116311214A (en) * | 2023-05-22 | 2023-06-23 | 珠海亿智电子科技有限公司 | License plate recognition method and device |
-
2021
- 2021-05-12 CN CN202110518506.1A patent/CN115424250A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116311214A (en) * | 2023-05-22 | 2023-06-23 | 珠海亿智电子科技有限公司 | License plate recognition method and device |
| CN116311214B (en) * | 2023-05-22 | 2023-08-22 | 珠海亿智电子科技有限公司 | License plate recognition method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110363182B (en) | Deep learning-based lane line detection method | |
| CN110020592B (en) | Object detection model training method, device, computer equipment and storage medium | |
| CN111027493B (en) | Pedestrian detection method based on deep learning multi-network soft fusion | |
| CN111950453A (en) | Optional-shape text recognition method based on selective attention mechanism | |
| CN112116599B (en) | Sputum smear tubercle bacillus semantic segmentation method and system based on weak supervised learning | |
| CN111898432B (en) | Pedestrian detection system and method based on improved YOLOv3 algorithm | |
| CN112232371B (en) | American license plate recognition method based on YOLOv3 and text recognition | |
| CN111783819B (en) | Improved target detection method based on region of interest training on small-scale data set | |
| CN112287941B (en) | License plate recognition method based on automatic character region perception | |
| CN110781980B (en) | Training method of target detection model, target detection method and device | |
| CN114758288A (en) | Power distribution network engineering safety control detection method and device | |
| CN113642390A (en) | Street view image semantic segmentation method based on local attention network | |
| CN110852327A (en) | Image processing method, image processing device, electronic equipment and storage medium | |
| CN109858327B (en) | Character segmentation method based on deep learning | |
| CN111444986A (en) | Building drawing component classification method and device, electronic equipment and storage medium | |
| CN114332133A (en) | New coronary pneumonia CT image infected area segmentation method and system based on improved CE-Net | |
| CN110991374B (en) | Fingerprint singular point detection method based on RCNN | |
| CN113487610B (en) | Herpes image recognition method and device, computer equipment and storage medium | |
| CN112784494B (en) | Training method of false positive recognition model, target recognition method and device | |
| CN115424250A (en) | License plate recognition method and device | |
| KR102026280B1 (en) | Method and system for scene text detection using deep learning | |
| CN114663731B (en) | Training method and system of license plate detection model, and license plate detection method and system | |
| CN116189130A (en) | Lane line segmentation method and device based on image annotation model | |
| CN112950652B (en) | Robot and hand image segmentation method and device thereof | |
| CN114863442A (en) | Character recognition method, system, device and storage medium based on deep learning |
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 |