WO2022048209A1 - 车牌识别方法、装置、电子设备及存储介质 - Google Patents

车牌识别方法、装置、电子设备及存储介质 Download PDF

Info

Publication number
WO2022048209A1
WO2022048209A1 PCT/CN2021/097068 CN2021097068W WO2022048209A1 WO 2022048209 A1 WO2022048209 A1 WO 2022048209A1 CN 2021097068 W CN2021097068 W CN 2021097068W WO 2022048209 A1 WO2022048209 A1 WO 2022048209A1
Authority
WO
WIPO (PCT)
Prior art keywords
license plate
target frame
detection model
recognized
trained
Prior art date
Application number
PCT/CN2021/097068
Other languages
English (en)
French (fr)
Inventor
徐国诚
Original Assignee
平安科技（深圳）有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 平安科技（深圳）有限公司 filed Critical 平安科技（深圳）有限公司
Publication of WO2022048209A1 publication Critical patent/WO2022048209A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/60—Type of objects
    • G06V20/62—Text, e.g. of license plates, overlay texts or captions on TV images
    • G06V20/63—Scene text, e.g. street names
    • 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
    • G06T5/00—Image enhancement or restoration
    • G06T5/80—Geometric correction
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
    • G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/60—Type of objects
    • G06V20/62—Text, e.g. of license plates, overlay texts or captions on TV images
    • G06V20/625—License plates
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V2201/00—Indexing scheme relating to image or video recognition or understanding
    • G06V2201/07—Target detection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
    • G06V30/10—Character recognition
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
    • Y02T10/00—Road transport of goods or passengers
    • Y02T10/10—Internal combustion engine [ICE] based vehicles
    • Y02T10/40—Engine management systems

Definitions

  • the present application relates to the technical field of artificial intelligence, and in particular, to a license plate recognition method, device, electronic device, and computer-readable storage medium.
  • license plate recognition belongs to a core technology in the intelligent transportation system. It acquires images containing license plates and uses deep learning networks to obtain license plates Character recognition is performed on the image of the vehicle, so as to obtain the license plate information, which greatly facilitates the management of the vehicle by the relevant personnel and the system.
  • the license plate cannot be detected in many environments, for example, the angle of the license plate is not correct, the environment is too bright or too dark, and the license plate is too small in the picture.
  • the detection efficiency of the license plate position of the license plate image is low; 2.
  • the license plate characters in the license plate image need to be segmented before the license plate characters can be recognized, which leads to the low efficiency of the license plate character recognition in the license plate image.
  • the current license plate recognition method mainly has the problem of low license plate recognition efficiency.
  • a license plate recognition method provided by this application includes:
  • the license plate is intercepted on the to-be-recognized license plate picture to obtain the target license plate;
  • the license plate characters of the target license plate are recognized by using a pre-trained license plate character recognition model to obtain license plate information.
  • the present application also provides a license plate recognition device, the device comprising:
  • the detection module is used to detect the license plate target frame of the license plate image to be recognized by using the pre-trained license plate target frame detection model
  • the detection module is further configured to detect the license plate key points of the license plate target frame by using a pre-trained key point detection model
  • An interception module configured to intercept the license plate image of the to-be-recognized license plate according to the key points of the license plate to obtain the target license plate;
  • the recognition module is used for recognizing the license plate characters of the target license plate by using the pre-trained license plate character recognition model to obtain license plate information.
  • the present application also provides an electronic device, the electronic device comprising:
  • the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the steps of:
  • the license plate is intercepted on the to-be-recognized license plate picture to obtain the target license plate;
  • the license plate characters of the target license plate are recognized by using a pre-trained license plate character recognition model to obtain license plate information.
  • the present application also provides a computer-readable storage medium in which at least one instruction is stored, and the at least one instruction is executed by a processor in the electronic device to perform the following steps:
  • the license plate is intercepted on the to-be-recognized license plate picture to obtain the target license plate;
  • the license plate characters of the target license plate are recognized by using a pre-trained license plate character recognition model to obtain license plate information.
  • FIG. 1 is a detailed schematic flowchart of a model training method provided by an embodiment of the present application.
  • FIG. 2 is a detailed schematic flowchart of step S11 of the model training method provided in FIG. 1 in an embodiment of the present application;
  • FIG. 3 is a schematic flowchart of a license plate recognition method provided by an embodiment of the present application.
  • FIG. 4 is a detailed flowchart of the interception of the license plate in step S3 of the license plate recognition method provided in FIG. 3 in the embodiment of the application;
  • FIG. 5 is a schematic block diagram of a license plate recognition device provided by an embodiment of the present application.
  • FIG. 6 is a schematic diagram of the internal structure of an electronic device for implementing a license plate recognition method provided by an embodiment of the present application
  • the execution body of the license plate recognition method provided by the embodiments of the present application includes, but is not limited to, at least one of electronic devices that can be configured to execute the method provided by the embodiments of the present application, such as a server and a terminal.
  • the license plate recognition method can be executed by software or hardware installed in a terminal device or a server device, and the software can be a blockchain platform.
  • the server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, and the like.
  • the core of the license plate recognition method provided by the embodiment of the present application is to use three models to respectively identify the license plate target frame in the license plate image to be recognized, detect the key points of the license plate from the license plate target frame to intercept the target license plate, and identify the target license plate The license plate characters in , get the license plate information.
  • the three models are respectively a license plate target frame detection model, a key point detection model and a license plate character recognition model.
  • the model training method includes:
  • the license plate image set includes a plurality of license plate images, and the license plate character data set includes specific license plate information in each license plate image.
  • the license plate information is Yue B250ZZ.
  • the license plate image set and the license plate character data set are downloaded from a vehicle database.
  • the labeling of the target frame of the license plate and the key of the license plate of the license plate picture set is performed by a preset data labeling tool.
  • the labeling of the license plate target frame can be performed by labeling the target frame for the area of the license plate information in the license plate image, and the labeling of the key points of the license plate is performed by labeling the upper left, lower left, upper right and lower right of the license plate information in the target frame of the license plate Four points are marked.
  • the preset data labeling tool is labelImg, and the labelImg is written based on the Python language and can support cross-platform running on Windows and Linux.
  • the pre-built target frame detection model is generated based on the Single Shot MultiBox Detector (referred to as SSD) algorithm, and is used to detect the target location area in the picture.
  • SSD Single Shot MultiBox Detector
  • the pre-built key point detection model is generated based on the maskrcnn algorithm for performing key point detection in the picture
  • the pre-built character recognition model is generated based on a convolutional neural network (Convolutional Neural Networks, CNN), which uses for character recognition.
  • CNN convolutional Neural Networks
  • the pre-built target frame detection model is trained by using the license plate target frame picture set to obtain a pre-trained license plate target frame detection model, including:
  • S112 Calculate the loss value between the training value and the corresponding label value, adjust the parameters of the pre-built target frame detection model according to the loss value until the loss value is not greater than a preset loss value, and obtain the license plate Object box detection model.
  • the label value refers to the real position area of the license plate target frame in the license plate target frame picture
  • the training value refers to the position area of the license plate target frame predicted by the pre-built target frame detection model.
  • the following method is used to calculate the loss value of the training value and the corresponding label value:
  • L(s) represents the loss value
  • k represents the number of license plate target frame image sets
  • yi represents the label value of the ith license plate target frame image
  • y′ i represents the training value of the ith license plate target frame image
  • the parameters of the pre-built target frame detection model include: model weight and bias, and the preset loss value is 0.1.
  • the training principle of using the license plate key point image set to train the pre-built key point detection model and the training principle of using the license plate character data set to train the pre-built character recognition model and the above-mentioned pre-trained target The training principle of the box detection model is the same and will not be further elaborated here.
  • FIG. 3 is a schematic flowchart of a license plate recognition method provided by an embodiment of the present application.
  • the license plate recognition method includes:
  • the image of the license plate to be recognized refers to the image of the license plate for which license plate information needs to be obtained
  • the license plate target frame refers to the area containing the license plate information
  • the method before the S1, further includes: performing image equalization processing on the picture of the license plate to be recognized and/or performing angular rotation on the picture of the license plate to be recognized.
  • the license plate target frame detection model is used to detect the image of the to-be-recognized license plate after image equalization and/or angle rotation.
  • a histogram equalization algorithm may be used to perform image equalization processing on the to-be-recognized license plate image and to perform 90-degree rotation on the to-be-recognized license plate image.
  • using the pre-trained license plate target frame detection model to detect the license plate target frame of the image to be recognized after image equalization and/or angle rotation can improve the detection accuracy of the license plate target frame.
  • Environmental adaptability to increase license plate detection rate can improve the detection accuracy of the license plate target frame.
  • the picture of the license plate to be recognized can also be stored in a node of a blockchain.
  • the method may further include: identifying whether the to-be-recognized license plate image has multiple license plate target boxes, so as to determine that a license plate image has multiple license plate information.
  • the key point detection model is used to detect the license plate key points of the license plate target frame.
  • the license plate target frame is segmented on the picture of the license plate to be recognized to obtain a plurality of sub-license plate target frames, and use all the target frames of the license plate to be identified.
  • the key point detection model described above detects the license plate key points of each sub-license plate target frame.
  • the detection result of the license plate target frame detection model it is identified whether the to-be-recognized license plate picture has multiple license plate target frames.
  • an image region segmentation algorithm is used to segment the license plate target frame on the to-be-recognized license plate picture.
  • the key points of the license plate can be directly located, so that the position of the license plate can be directly extracted without locating the upper and lower boundaries and the left and right boundaries of the license plate. Detection efficiency of key points of license plate.
  • the S3 includes:
  • the distortion correction refers to a method of correcting an object with an oblique angle in a picture into a rectangle with a normal viewing angle by a method of perspective transformation, which is convenient for image processing.
  • the size of the preset standard license plate may refer to the size of the license plate in mainland China. Specifically, the size of the license plate in mainland China is 440 mm in length and 140 mm in height.
  • the preset picture cropping tool may be a picture cutter.
  • the target license plate is input into the license plate character recognition model to obtain the license plate characters of the target license plate, that is, the license plate information. Based on the license plate character recognition model generated by the CNN, the target license plate is directly recognized. For character recognition, no character segmentation is required, and the speed of license plate character recognition is improved.
  • the embodiment of the present application firstly uses a pre-trained license plate target frame detection model to detect the license plate target frame of the license plate image to be recognized, which can improve the environmental adaptability of license plate detection and increase the license plate detection rate;
  • the point detection model detects the key points of the license plate in the target frame of the license plate, and can directly locate the four corners of the license plate, so that the position of the license plate can be directly extracted, without locating the upper and lower boundaries and the left and right boundaries of the license plate, so that the key points of the license plate can be detected more accurately.
  • the detection efficiency of the key points of the license plate is also improved; further, according to the key points of the license plate, the embodiment of the present application performs the license plate interception on the to-be-recognized license plate picture to obtain the target license plate, and uses the pre-trained license plate character recognition model to recognize The license plate characters of the target license plate are obtained, the license plate information is obtained, and the character recognition of the target license plate can be performed directly without character segmentation, thereby improving the speed of license plate character recognition. Therefore, a license plate recognition method proposed in this application can improve the recognition efficiency of license plate recognition.
  • FIG. 5 it is a functional block diagram of the license plate recognition device of the present application.
  • the license plate recognition device 100 described in this application can be installed in an electronic device.
  • the license plate recognition device may include a detection module 101 , an interception module 102 and an identification module 103 .
  • the modules described in the present invention can also be called units, which refer to a series of computer program segments that can be executed by the electronic device processor and can perform fixed functions, and are stored in the memory of the electronic device.
  • each module/unit is as follows:
  • the detection module 101 is used to detect the license plate target frame of the license plate image to be recognized by using the pre-trained license plate target frame detection model.
  • the image of the license plate to be recognized refers to the image of the license plate for which license plate information needs to be obtained
  • the license plate target frame refers to the area containing the license plate information
  • the detection module 101 before using the pre-trained license plate target frame detection model to detect the license plate target frame of the to-be-recognized license plate picture, the detection module 101 is also used for: detecting the to-be-recognized license plate image Perform image equalization processing and/or perform angular rotation on the picture of the license plate to be recognized.
  • the license plate target frame detection model is used to detect the image of the to-be-recognized license plate after image equalization and/or angle rotation.
  • a histogram equalization algorithm may be used to perform image equalization processing on the to-be-recognized license plate image and to perform 90-degree rotation on the to-be-recognized license plate image.
  • using the pre-trained license plate target frame detection model to detect the license plate target frame of the image to be recognized after image equalization and/or angle rotation can improve the detection accuracy of the license plate target frame.
  • Environmental adaptability to increase license plate detection rate can improve the detection accuracy of the license plate target frame.
  • the picture of the license plate to be recognized can also be stored in a node of a blockchain.
  • the detection module 101 is further configured to detect the license plate key points of the license plate target frame by using a pre-trained key point detection model.
  • the detection module 101 can also be used to: identify whether the to-be-recognized license plate picture has multiple license plates before detecting the license plate key points of the license plate target frame by using a pre-trained key point detection model
  • the target frame is used to determine the existence of multiple license plate information in a license plate image.
  • the key point detection model is used to detect the license plate key points of the license plate target frame.
  • the license plate target frame is segmented on the picture of the license plate to be recognized to obtain a plurality of sub-license plate target frames, and use all the target frames of the license plate to be identified.
  • the key point detection model described above detects the license plate key points of each sub-license plate target frame.
  • the detection result of the license plate target frame detection model it is identified whether the to-be-recognized license plate picture has multiple license plate target frames.
  • an image region segmentation algorithm is used to segment the license plate target frame on the to-be-recognized license plate picture.
  • the key points of the license plate can be directly located, so that the position of the license plate can be directly extracted without locating the upper and lower boundaries and the left and right boundaries of the license plate. Detection efficiency of key points of license plate.
  • the intercepting module 102 is configured to perform license plate interception on the to-be-recognized license plate image according to the key points of the license plate to obtain a target license plate.
  • the intercepting module 102 performs license plate interception on the to-be-recognized license plate image according to the key points of the license plate to obtain the target license plate, including:
  • Step A based on the key points of the license plate, perform license plate distortion correction on the to-be-recognized license plate picture to obtain an initial license plate;
  • Step B based on the size of the preset standard license plate, size configuration is performed on the initial license plate to obtain a standard license plate;
  • Step C using a preset image cropping tool to crop the standard license plate to obtain the target license plate.
  • the distortion correction refers to a method of correcting an object with an oblique angle in a picture into a rectangle with a normal viewing angle through a method of perspective transformation, which is convenient for image processing.
  • the size of the preset standard license plate may refer to the size of the license plate in mainland China. Specifically, the size of the license plate in mainland China is 440 mm in length and 140 mm in height.
  • the preset picture cropping tool may be a picture cutter.
  • the recognition module 103 is configured to recognize the license plate characters of the target license plate by using a pre-trained license plate character recognition model to obtain license plate information.
  • the target license plate is input into the license plate character recognition model to obtain the license plate characters of the target license plate, that is, the license plate information. Based on the license plate character recognition model generated by the CNN, the target license plate is directly recognized. For character recognition, no character segmentation is required, and the speed of license plate character recognition is improved.
  • the license plate recognition device further includes a model training module for training the license plate target frame detection model, the key point detection model, and the license plate character recognition model.
  • model training module trains the license plate target frame detection model by the following method:
  • Step 1 Obtain a license plate image set and a license plate character data set, label the license plate target frame and license plate key points on the license plate image set, and generate a license plate target frame image set and a license plate key point image set respectively;
  • Step 2 using the license plate target frame picture set to train the pre-built target frame detection model to obtain a pre-trained license plate target frame detection model;
  • Step 3 using the license plate key point image set to train the pre-built key point detection model to obtain a pre-trained license plate key point detection model;
  • Step 4 Using the license plate character data set to train the pre-built character recognition model to obtain a pre-trained license plate character recognition model.
  • the license plate image set includes a plurality of license plate images, and the license plate character data set includes specific license plate information in each license plate image.
  • the license plate information is Yue B250ZZ.
  • the license plate image set and the license plate character data set are downloaded from a vehicle database.
  • the labeling of the target frame of the license plate and the key of the license plate of the license plate picture set is performed by a preset data labeling tool.
  • the labeling of the license plate target frame can be performed by labeling the target frame for the area of the license plate information in the license plate image, and the labeling of the key points of the license plate is performed by labeling the upper left, lower left, upper right and lower right of the license plate information in the target frame of the license plate Four points are marked.
  • the preset data labeling tool is labelImg, and the labelImg is written based on the Python language and can support cross-platform running on Windows and Linux.
  • the pre-built target frame detection model is generated based on the Single Shot MultiBox Detector (referred to as SSD) algorithm, and is used to detect the target location area in the picture.
  • SSD Single Shot MultiBox Detector
  • the pre-built key point detection model is generated based on the maskrcnn algorithm for performing key point detection in the picture
  • the pre-built character recognition model is generated based on a convolutional neural network (Convolutional Neural Networks, CNN), which uses for character recognition.
  • CNN convolutional Neural Networks
  • the pre-built target frame detection model is trained by using the license plate target frame picture set to obtain a pre-trained license plate target frame detection model, including:
  • Step A obtain the label value corresponding to the picture set of the license plate target frame
  • Step B using the convolution layer in the target frame detection model to perform a convolution operation on the license plate target frame picture set to obtain the feature vector of the license plate target frame picture set, and using the pool in the target frame detection model.
  • the layer performs a pooling operation on the feature vector, and uses the activation layer in the target frame detection model to calculate the pooled feature vector to obtain the training value of the license plate target frame picture set;
  • Step C Calculate the loss value of the training value and the corresponding label value, adjust the parameters of the pre-built target frame detection model according to the loss value, until the loss value is not greater than the preset loss value, and obtain the License plate target box detection model.
  • the label value refers to the real position area of the license plate target frame in the license plate target frame picture
  • the training value refers to the position area of the license plate target frame predicted by the pre-built target frame detection model.
  • the following method is used to calculate the loss value of the training value and the corresponding label value:
  • L(s) represents the loss value
  • k represents the number of license plate target frame image sets
  • yi represents the label value of the ith license plate target frame image
  • y′ i represents the training value of the ith license plate target frame image
  • the parameters of the pre-built target frame detection model include: model weight and bias, and the preset loss value is 0.1.
  • the training principle of using the license plate key point image set to train the pre-built key point detection model and the training principle of using the license plate character data set to train the pre-built character recognition model and the above-mentioned pre-trained target The training principle of the box detection model is the same and will not be further elaborated here.
  • the embodiment of the present application firstly uses a pre-trained license plate target frame detection model to detect the license plate target frame of the license plate image to be recognized, which can improve the environmental adaptability of license plate detection and increase the license plate detection rate;
  • the point detection model detects the key points of the license plate in the target frame of the license plate, and can directly locate the four corners of the license plate, so that the position of the license plate can be directly extracted, without locating the upper and lower boundaries and the left and right boundaries of the license plate, so that the key points of the license plate can be detected more accurately.
  • the detection efficiency of the key points of the license plate is also improved; further, according to the key points of the license plate, the embodiment of the present application performs the license plate interception on the to-be-recognized license plate picture to obtain the target license plate, and uses the pre-trained license plate character recognition model to recognize The license plate characters of the target license plate are obtained, the license plate information is obtained, and the character recognition of the target license plate can be performed directly without character segmentation, thereby improving the speed of license plate character recognition. Therefore, the license plate recognition device proposed in the present application can improve the recognition efficiency of license plate recognition.
  • FIG. 6 it is a schematic structural diagram of an electronic device implementing the license plate recognition method of the present application.
  • the electronic device 1 may include a processor 10, a memory 11 and a bus, and may also include a computer program stored in the memory 11 and executable on the processor 10, such as a license plate recognition program 12.
  • the memory 11 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, mobile hard disk, multimedia card, card-type memory (for example: SD or DX memory, etc.), magnetic memory, magnetic disk, CD etc.
  • the memory 11 may be an internal storage unit of the electronic device 1 in some embodiments, such as a mobile hard disk of the electronic device 1 .
  • the memory 11 may also be an external storage device of the electronic device 1, such as a pluggable mobile hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital) equipped on the electronic device 1. , SD) card, flash memory card (Flash Card), etc.
  • the memory 11 may also include both an internal storage unit of the electronic device 1 and an external storage device.
  • the memory 11 can not only be used to store application software installed in the electronic device 1 and various data, such as the code of the license plate recognition program, etc., but also can be used to temporarily store the data that has been output or will be output.
  • the processor 10 may be composed of integrated circuits, for example, may be composed of a single packaged integrated circuit, or may be composed of multiple integrated circuits packaged with the same function or different functions, including one or more integrated circuits.
  • Central Processing Unit CPU
  • microprocessor digital processing chip
  • graphics processor and combination of various control chips, etc.
  • the processor 10 is the control core (Control Unit) of the electronic device, and uses various interfaces and lines to connect the various components of the entire electronic device, by running or executing the program or module (for example, executing the program) stored in the memory 11. license plate recognition program, etc.), and call the data stored in the memory 11 to execute various functions of the electronic device 1 and process data.
  • the bus may be a peripheral component interconnect (PCI for short) bus or an extended industry standard architecture (Extended industry standard architecture, EISA for short) bus or the like.
  • PCI peripheral component interconnect
  • EISA Extended industry standard architecture
  • the bus can be divided into address bus, data bus, control bus and so on.
  • the bus is configured to implement connection communication between the memory 11 and at least one processor 10 and the like.
  • FIG. 6 only shows an electronic device with components. Those skilled in the art can understand that the structure shown in FIG. 6 does not constitute a limitation on the electronic device 1, and may include fewer or more components than those shown in the drawings. components, or a combination of certain components, or a different arrangement of components.
  • the electronic device 1 may also include a power supply (such as a battery) for powering the various components, preferably, the power supply may be logically connected to the at least one processor 10 through a power management device, so that the power management
  • the device implements functions such as charge management, discharge management, and power consumption management.
  • the power source may also include one or more DC or AC power sources, recharging devices, power failure detection circuits, power converters or inverters, power status indicators, and any other components.
  • the electronic device 1 may further include various sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be repeated here.
  • the electronic device 1 may also include a network interface, optionally, the network interface may include a wired interface and/or a wireless interface (such as a WI-FI interface, a Bluetooth interface, etc.), which is usually used in the electronic device 1 Establish a communication connection with other electronic devices.
  • a network interface optionally, the network interface may include a wired interface and/or a wireless interface (such as a WI-FI interface, a Bluetooth interface, etc.), which is usually used in the electronic device 1 Establish a communication connection with other electronic devices.
  • the electronic device 1 may further include a user interface, and the user interface may be a display (Display), an input unit (eg, a keyboard (Keyboard)), optionally, the user interface may also be a standard wired interface or a wireless interface.
  • the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode, organic light-emitting diode) touch device, and the like.
  • the display may also be appropriately called a display screen or a display unit, which is used for displaying information processed in the electronic device 1 and for displaying a visualized user interface.
  • the license plate recognition program 12 stored in the memory 11 in the electronic device 1 is a combination of multiple instructions, and when running in the processor 10, it can realize:
  • the license plate is intercepted on the to-be-recognized license plate picture to obtain the target license plate;
  • the license plate characters of the target license plate are recognized by using a pre-trained license plate character recognition model to obtain license plate information.
  • a pre-trained license plate character recognition model to obtain license plate information.
  • the modules/units integrated in the electronic device 1 may be stored in a computer-readable storage medium.
  • the computer-readable storage medium may be non-volatile or non-volatile.
  • the computer-readable storage medium may include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a removable hard disk, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM, Read-Only Memory) Only Memory).
  • the computer-readable storage medium stores a license plate recognition program
  • the license plate recognition program includes at least one instruction, which, when executed by the processor, can realize:
  • the license plate is intercepted on the to-be-recognized license plate picture to obtain the target license plate;
  • the license plate characters of the target license plate are recognized by using a pre-trained license plate character recognition model to obtain license plate information.
  • modules described as separate components may or may not be physically separated, and components shown as modules may or may not be physical units, that is, may be located in one place, or may be distributed to multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution in this embodiment.
  • each functional module in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically alone, or two or more units may be integrated into one unit.
  • the above-mentioned integrated units can be implemented in the form of hardware, or can be implemented in the form of hardware plus software function modules.
  • the blockchain referred to in this application is a new application mode of computer technologies such as distributed data storage, point-to-point transmission, consensus mechanism, and encryption algorithm.
  • Blockchain essentially a decentralized database, is a series of data blocks associated with cryptographic methods. Each data block contains a batch of network transaction information to verify its Validity of information (anti-counterfeiting) and generation of the next block.
  • the blockchain can include the underlying platform of the blockchain, the platform product service layer, and the application service layer.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • General Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Biomedical Technology (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Biophysics (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Character Discrimination (AREA)
  • Image Analysis (AREA)

Abstract

一种车牌识别方法，包括：利用预先训练的车牌目标框检测模型检测待识别车牌图片的车牌目标框（S1）；利用预先训练的关键点检测模型检测所述车牌目标框的车牌关键点（S2）；根据所述车牌关键点，对所述待识别车牌图片进行车牌截取，得到目标车牌（S3）；利用预先训练的车牌字符识别模型识别出所述目标车牌的车牌字符，得到车牌信息（S4）。所述方法可以提高车牌识别的识别效率。

Description

车牌识别方法、装置、电子设备及存储介质
本申请要求于2020年9月3日提交中国专利局、申请号为CN202010916527.4、名称为“车牌识别方法、装置、电子设备及存储介质”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请涉及人工智能技术领域，尤其涉及一种车牌识别方法、装置、电子设备及计算机可读存储介质。
背景技术
随着AI技术的日益发展，智能化技术已经涉及到生活的各个层面中，其中，车牌识别属于智能交通***中的一项核心技术，其通过获取包含车牌的图像，利用深度学习网络对获取车牌的图像进行字符识别，从而得到车牌信息，大大方便了相关人员及***对于车辆的管理。
发明人意识到目前关于车牌识别主要存在如下问题：1、对于获取包含车牌图像，在很多环境下检测不到车牌，比如说车牌角度不正，环境过亮或者过暗、车牌在图片中过小过大等，导致车牌图像的车牌位置检测效率低；2、需要对车牌图像中车牌字符进行分割后才能进行车牌字符的识别，导致车牌图像的车牌字符识别效率低。
因此，目前车牌识别方法主要存在车牌识别效率低下的问题。
发明内容
本申请提供的一种车牌识别方法，包括：
利用预先训练的车牌目标框检测模型检测待识别车牌图片的车牌目标框；
利用预先训练的关键点检测模型检测所述车牌目标框的车牌关键点；
根据所述车牌关键点，对所述待识别车牌图片进行车牌截取，得到目标车牌；
利用预先训练的车牌字符识别模型识别出所述目标车牌的车牌字符，得到车牌信息。
本申请还提供一种车牌识别装置，所述装置包括：
检测模块，用于利用预先训练的车牌目标框检测模型检测待识别车牌图片的车牌目标框；
所述检测模块，还用于利用预先训练的关键点检测模型检测所述车牌目标框的车牌关键点；
截取模块，用于根据所述车牌关键点，对所述待识别车牌图片进行车牌截取，得到目标车牌；
识别模块，用于利用预先训练的车牌字符识别模型识别出所述目标车牌的车牌字符，得到车牌信息。
本申请还提供一种电子设备，所述电子设备包括：
至少一个处理器；以及，
与所述至少一个处理器通信连接的存储器；其中，
所述存储器存储有可被所述至少一个处理器执行的指令，所述指令被所述至少一个处理器执行，以使所述至少一个处理器能够执行如下步骤：
利用预先训练的车牌目标框检测模型检测待识别车牌图片的车牌目标框；
利用预先训练的关键点检测模型检测所述车牌目标框的车牌关键点；
根据所述车牌关键点，对所述待识别车牌图片进行车牌截取，得到目标车牌；
利用预先训练的车牌字符识别模型识别出所述目标车牌的车牌字符，得到车牌信息。
本申请还提供一种计算机可读存储介质，所述计算机可读存储介质中存储有至少一个 指令，所述至少一个指令被电子设备中的处理器执行以执行如下步骤：
利用预先训练的车牌目标框检测模型检测待识别车牌图片的车牌目标框；
利用预先训练的关键点检测模型检测所述车牌目标框的车牌关键点；
根据所述车牌关键点，对所述待识别车牌图片进行车牌截取，得到目标车牌；
利用预先训练的车牌字符识别模型识别出所述目标车牌的车牌字符，得到车牌信息。
附图说明
图1为本申请实施例提供的模型训练方法的详细流程示意图；
图2为本申请实施例中图1提供的模型训练方法步骤S11的详细流程示意图；
图3为本申请实施例提供的车牌识别方法的流程示意图；
图4为本申请实施例中图3提供的车牌识别方法步骤S3中车牌截取的详细流程示意图；
图5为本申请实施例提供的车牌识别装置的模块示意图；
图6为本申请实施例提供的实现车牌识别方法的电子设备的内部结构示意图；
本申请目的的实现、功能特点及优点将结合实施例，参照附图做进一步说明。
具体实施方式
应当理解，此处所描述的具体实施例仅仅用以解释本申请，并不用于限定本申请。
本申请实施例提供的车牌识别方法的执行主体包括但不限于服务端、终端等能够被配置为执行本申请实施例提供的该方法的电子设备中的至少一种。换言之，所述车牌识别方法可以由安装在终端设备或服务端设备的软件或硬件来执行，所述软件可以是区块链平台。所述服务端包括但不限于：单台服务器、服务器集群、云端服务器或云端服务器集群等。
本申请实施例提供的车牌识别方法核心在于利用三个模型分别识别待识别车牌图片中的车牌目标框、从所述车牌目标框中检测车牌关键点从而截取到目标车牌，并识别所述目标车牌中的车牌字符，得到车牌信息。
详细地，本申请其中一个优选实施例中，所述三个模型分别为车牌目标框检测模型、关键点检测模型以及车牌字符识别模型。
本申请实施例中，该三个模型需要预先训练，参阅图1所示，所述模型训练方法包括：
S10、获取车牌图片集及车牌字符数据集，对所述车牌图片集进行车牌目标框和车牌关键点的标注，分别生成车牌目标框图片集和车牌关键点图片集；
S11、利用所述车牌目标框图片集对预构建的目标框检测模型进行训练，得到预先训练的车牌目标框检测模型；
S12、利用所述车牌关键点图片集对预构建的关键点检测模型进行训练，得到预先训练的车牌关键点检测模型；
S13、利用车牌字符数据集对预构建的字符识别模型进行训练，得到预先训练的车牌字符识别模型。
其中，所述车牌图片集包括多张包含车牌图片，所述车牌字符数据集包含每个车牌图片中具体的车牌信息，例如，所述车牌信息为粤B250ZZ。
一个优选实施例中，所述车牌图片集及车牌字符数据集通过车辆数据库下载。
一个优选实施例中，所述车牌图片集的车牌目标框和车牌关键的标注通过预设的数据标注工具执行。
其中，所述车牌目标框的标注可以通过对车牌图片中车牌信息的区域进行目标框标注，所述车牌关键点的标注通过对所述车牌目标框中车牌信息的左上、左下、右上以及右下四个点进行标注。
一个可选实施例中，所述预设的数据标注工具为labelImg，所述labelImg基于Python语言编写，可以支持在Windows、Linux的跨平台运行。
进一步地，所述预构建的目标框检测模型基于Single Shot MultiBox Detector(简称SSD) 算法生成，用于检测图片中的目标位置区域。
此外，所述预构建的关键点检测模型基于maskrcnn算法生成，用于执行图片中的关键点检测，所述预构建的字符识别模型基于卷积神经网络(Convolutional Neural Networks，CNN)生成，其用于字符识别。
详细地，参阅图2所示，所述利用所述车牌目标框图片集对预构建的目标框检测模型进行训练，得到预先训练的车牌目标框检测模型，包括：
S110、获取所述车牌目标框图片集对应的标签值；
S111、利用所述目标框检测模型中的卷积层对所述车牌目标框图片集进行卷积操作，得到所述车牌目标框图片集的特征向量，利用所述目标框检测模型中的池化层对所述特征向量进行池化操作，利用所述目标框检测模型中的激活层对池化后的所述特征向量进行计算，得到所述车牌目标框图片集的训练值；
S112、计算所述训练值与对应标签值的损失值，根据所述损失值调整所述预构建的目标框检测模型的参数，直至所述损失值不大于预设的损失值，得到所述车牌目标框检测模型。
所述标签值指的是车牌目标框图片中车牌目标框的真实位置区域，所述训练值指的是通过所述预构建的目标框检测模型预测的车牌目标框的位置区域。
一个优选实施例中，利用下述方法计算所述训练值与对应标签值的损失值：
Figure PCTCN2021097068-appb-000001
其中，其中，L(s)表示损失值，k表示车牌目标框图片集的数量，y i表示第i个车牌目标框图片的标签值，y′ i表示第i个车牌目标框图片的训练值。
一个可选实施例中，所述预构建的目标框检测模型的参数包括：模型权重和偏置，所述预设的损失值为0.1。
进一步地，所述利用所述车牌关键点图片集对预构建的关键点检测模型进行训练以及所述利用车牌字符数据集对预构建的字符识别模型进行训练的训练原理与上述预训练好的目标框检测模型的训练原理相同，在此不再进一步阐述。
进一步地，本申请实施例可以利用上述训练完成的三个模型执行车牌优化识别。详细地，图3是本申请实施例提供的车牌识别方法的流程示意图。在本申请实施例中，所述车牌识别方法包括：
S1、利用预先训练的车牌目标框检测模型检测待识别车牌图片的车牌目标框。
本申请较佳实施例中，所述待识别车牌图片指的是需要获取车牌信息的车牌图片，所述车牌目标框指的是包含车牌信息的区域。
进一步地，本申请另一个较佳实施例中，在所述S1之前，还包括：对所述待识别车牌图片进行图像均衡化处理和/或对所述待识别车牌图片进行角度旋转。
对所述待识别车牌图片进行图像均衡化处理和/或对所述待识别车牌图片进行角度旋转，可以调整所述待识别车牌图片的亮度或角度，避免出现因图片模糊不清晰或图片角度而导致车牌目标框无法检测的情况，进一步地，利用所述车牌目标框检测模型检测图像均衡化和/或角度旋转后的所述待识别车牌图片。
一个优选实施例中，可以利用直方图均衡化算法对所述待识别车牌图片进行图像均衡化处理以及对所述待识别车牌图片进行90度旋转。
进一步地，本申请较佳实施例中，利用所述预先训练的车牌目标框检测模型检测图像均衡化处理和/或角度旋转后的待识别车牌图片的车牌目标框，可以提高车牌目标框检测的环境适应性，增加车牌检出率。
较佳地，为进一步保证所述待识别车牌图片的私密和安全性，所述待识别车牌图片还可以存储于一区块链的节点中。
S2、利用预先训练的关键点检测模型检测所述车牌目标框的车牌关键点。
本申请较佳实施例中，所述S2之前，还可以包括：识别所述待识别车牌图片是否具有多个车牌目标框，以确定出一个车牌图片存在多个车牌信息。
进一步地，若识别出所述待识别车牌图片不具有多个车牌目标框，则利用所述关键点检测模型检测所述车牌目标框的车牌关键点。
进一步地，若识别出所述待识别车牌图片具有多个车牌目标框，根据识别的车牌目标框，对所述待识别车牌图片进行车牌目标框切分，得到多个子车牌目标框，并利用所述关键点检测模型检测每一个子车牌目标框的车牌关键点。
一个优选实施例中，根据车牌目标框检测模型的检测结果，识别出所述待识别车牌图片是否具有多个车牌目标框。
一个优选实施例中，利用图像区域分割算法对所述待识别车牌图片进行车牌目标框切分。
基于上述通过maskrcnn算法生成的车牌关键点检测模型直接定位车牌关键点，从而可以直接提取出车牌位置，不用再定位车牌的上下边界和左右边界，以使车牌关键点检测更加准确的同时也提高了车牌关键点的检测效率。
S3、根据所述车牌关键点，对所述待识别车牌图片进行车牌截取，得到目标车牌。
本申请较佳实施例中，参阅图4所示，所述S3，包括：
S30、基于所述车牌关键点，对所述待识别车牌图片进行车牌畸变纠正，得到初始车牌；
S31、基于预设标准车牌的尺寸，对所述初始车牌进行尺寸配置，得到标准车牌；
S32、利用预设图片裁剪工具对所述标准车牌进行裁剪，得到所述目标车牌。
其中，所述畸变纠正指的是一种通过透视变换的方法把图片中角度倾斜的物体纠正为正视角度的矩形，方便进行图像处理方法。
所述预设标准车牌的尺寸可参考中国内地车牌尺寸，具体的，所述中国内地车牌尺寸为长440mm，高140mm。
所述预设图片裁剪工具可以为图片切割器。
S4、利用预先训练的车牌字符识别模型识别出所述目标车牌的车牌字符，得到车牌信息
本申请实施例中，将所述目标车牌输入至所述车牌字符识别模型中，得到所述目标车牌的车牌字符，即所述车牌信息，基于上述通过CNN生成的车牌字符识别模型直接对目标车牌进行字符识别，不需要进行字符分割，提高车牌字符识别的速度。
本申请实施例首先利用预先训练的车牌目标框检测模型检测待识别车牌图片的车牌目标框，可以提高车牌检测的环境适应性，增加车牌检出率；其次，本申请实施例利用预先训练的关键点检测模型检测所述车牌目标框的车牌关键点，可以直接定位车牌四个角，从而可以直接提取出车牌位置，不用再定位车牌的上下边界和左右边界，以使车牌关键点检测更加准确的同时也提高了车牌关键点的检测效率；进一步地，本申请实施例根据所述车牌关键点，对所述待识别车牌图片进行车牌截取，得到目标车牌，并利用预先训练的车牌字符识别模型识别出所述目标车牌的车牌字符，得到车牌信息，可以直接对目标车牌进行字符识别，不需要进行字符分割，提高车牌字符识别的速度。因此，本申请提出的一种车牌识别方法可以提高车牌识别的识别效率。
如图5所示，是本申请车牌识别装置的功能模块图。
本申请所述车牌识别装置100可以安装于电子设备中。根据实现的功能，所述车牌识别装置可以包括检测模块101、截取模块102以及识别模块103。本发所述模块也可以称之为单元，是指一种能够被电子设备处理器所执行，并且能够完成固定功能的一系列计算机程序段，其存储在电子设备的存储器中。
在本实施例中，关于各模块/单元的功能如下：
所述检测模块101，用于利用预先训练的车牌目标框检测模型检测待识别车牌图片的车牌目标框。
本申请较佳实施例中，所述待识别车牌图片指的是需要获取车牌信息的车牌图片，所述车牌目标框指的是包含车牌信息的区域。
进一步地，本申请另一个较佳实施例中，所述检测模块101在利用预先训练的车牌目标框检测模型检测待识别车牌图片的车牌目标框之前，还用于：对所述待识别车牌图片进行图像均衡化处理和/或对所述待识别车牌图片进行角度旋转。
对所述待识别车牌图片进行图像均衡化处理和/或对所述待识别车牌图片进行角度旋转，可以调整所述待识别车牌图片的亮度或角度，避免出现因图片模糊不清晰或图片角度而导致车牌目标框无法检测的情况，进一步地，利用所述车牌目标框检测模型检测图像均衡化和/或角度旋转后的所述待识别车牌图片。
一个优选实施例中，可以利用直方图均衡化算法对所述待识别车牌图片进行图像均衡化处理以及对所述待识别车牌图片进行90度旋转。
进一步地，本申请较佳实施例中，利用所述预先训练的车牌目标框检测模型检测图像均衡化处理和/或角度旋转后的待识别车牌图片的车牌目标框，可以提高车牌目标框检测的环境适应性，增加车牌检出率。
较佳地，为进一步保证所述待识别车牌图片的私密和安全性，所述待识别车牌图片还可以存储于一区块链的节点中。
所述检测模块101，还用于利用预先训练的关键点检测模型检测所述车牌目标框的车牌关键点。
本申请较佳实施例中，所述检测模块101利用预先训练的关键点检测模型检测所述车牌目标框的车牌关键点之前，还可以用于：识别所述待识别车牌图片是否具有多个车牌目标框，以确定出一个车牌图片存在多个车牌信息。
进一步地，若识别出所述待识别车牌图片不具有多个车牌目标框，则利用所述关键点检测模型检测所述车牌目标框的车牌关键点。
进一步地，若识别出所述待识别车牌图片具有多个车牌目标框，根据识别的车牌目标框，对所述待识别车牌图片进行车牌目标框切分，得到多个子车牌目标框，并利用所述关键点检测模型检测每一个子车牌目标框的车牌关键点。
一个优选实施例中，根据车牌目标框检测模型的检测结果，识别出所述待识别车牌图片是否具有多个车牌目标框。
一个优选实施例中，利用图像区域分割算法对所述待识别车牌图片进行车牌目标框切分。
基于上述通过maskrcnn算法生成的车牌关键点检测模型直接定位车牌关键点，从而可以直接提取出车牌位置，不用再定位车牌的上下边界和左右边界，以使车牌关键点检测更加准确的同时也提高了车牌关键点的检测效率。
所述截取模块102，用于根据所述车牌关键点，对所述待识别车牌图片进行车牌截取，得到目标车牌。
本申请较佳实施例中，所述截取模块102根据所述车牌关键点，对所述待识别车牌图片进行车牌截取，得到目标车牌，包括：
步骤A、基于所述车牌关键点，对所述待识别车牌图片进行车牌畸变纠正，得到初始车牌；
步骤B、基于预设标准车牌的尺寸，对所述初始车牌进行尺寸配置，得到标准车牌；
步骤C、利用预设图片裁剪工具对所述标准车牌进行裁剪，得到所述目标车牌。
其中，所述畸变纠正指的是一种通过透视变换的方法把图片中角度倾斜的物体纠正为 正视角度的矩形，方便进行图像处理方法。
所述预设标准车牌的尺寸可参考中国内地车牌尺寸，具体的，所述中国内地车牌尺寸为长440mm，高140mm。
所述预设图片裁剪工具可以为图片切割器。
所述识别模块103，用于利用预先训练的车牌字符识别模型识别出所述目标车牌的车牌字符，得到车牌信息。
本申请实施例中，将所述目标车牌输入至所述车牌字符识别模型中，得到所述目标车牌的车牌字符，即所述车牌信息，基于上述通过CNN生成的车牌字符识别模型直接对目标车牌进行字符识别，不需要进行字符分割，提高车牌字符识别的速度。
优选地，本申请其他实施例中，所述车牌识别装置还包括模型训练模块，用于训练所述车牌目标框检测模型、关键点检测模型以及车牌字符识别模型。
详细地，所述模型训练模块通过下述方法训练所述车牌目标框检测模型：
步骤一、获取车牌图片集及车牌字符数据集，对所述车牌图片集进行车牌目标框和车牌关键点的标注，分别生成车牌目标框图片集和车牌关键点图片集；
步骤二、利用所述车牌目标框图片集对预构建的目标框检测模型进行训练，得到预先训练的车牌目标框检测模型；
步骤三、利用所述车牌关键点图片集对预构建的关键点检测模型进行训练，得到预先训练的车牌关键点检测模型；
步骤四、利用车牌字符数据集对预构建的字符识别模型进行训练，得到预先训练的车牌字符识别模型。
其中，所述车牌图片集包括多张包含车牌图片，所述车牌字符数据集包含每个车牌图片中具体的车牌信息，例如，所述车牌信息为粤B250ZZ。
一个优选实施例中，所述车牌图片集及车牌字符数据集通过车辆数据库下载。
一个优选实施例中，所述车牌图片集的车牌目标框和车牌关键的标注通过预设的数据标注工具执行。
其中，所述车牌目标框的标注可以通过对车牌图片中车牌信息的区域进行目标框标注，所述车牌关键点的标注通过对所述车牌目标框中车牌信息的左上、左下、右上以及右下四个点进行标注。
一个可选实施例中，所述预设的数据标注工具为labelImg，所述labelImg基于Python语言编写，可以支持在Windows、Linux的跨平台运行。
进一步地，所述预构建的目标框检测模型基于Single Shot MultiBox Detector(简称SSD)算法生成，用于检测图片中的目标位置区域。
此外，所述预构建的关键点检测模型基于maskrcnn算法生成，用于执行图片中的关键点检测，所述预构建的字符识别模型基于卷积神经网络(Convolutional Neural Networks，CNN)生成，其用于字符识别。
详细地，参阅图2所示，所述利用所述车牌目标框图片集对预构建的目标框检测模型进行训练，得到预先训练的车牌目标框检测模型，包括：
步骤A、获取所述车牌目标框图片集对应的标签值；
步骤B、利用所述目标框检测模型中的卷积层对所述车牌目标框图片集进行卷积操作，得到所述车牌目标框图片集的特征向量，利用所述目标框检测模型中的池化层对所述特征向量进行池化操作，利用所述目标框检测模型中的激活层对池化后的所述特征向量进行计算，得到所述车牌目标框图片集的训练值；
步骤C、计算所述训练值与对应标签值的损失值，根据所述损失值调整所述预构建的目标框检测模型的参数，直至所述损失值不大于预设的损失值，得到所述车牌目标框检测模型。
所述标签值指的是车牌目标框图片中车牌目标框的真实位置区域，所述训练值指的是通过所述预构建的目标框检测模型预测的车牌目标框的位置区域。
一个优选实施例中，利用下述方法计算所述训练值与对应标签值的损失值：
Figure PCTCN2021097068-appb-000002
其中，其中，L(s)表示损失值，k表示车牌目标框图片集的数量，y i表示第i个车牌目标框图片的标签值，y′ i表示第i个车牌目标框图片的训练值。
一个可选实施例中，所述预构建的目标框检测模型的参数包括：模型权重和偏置，所述预设的损失值为0.1。
进一步地，所述利用所述车牌关键点图片集对预构建的关键点检测模型进行训练以及所述利用车牌字符数据集对预构建的字符识别模型进行训练的训练原理与上述预训练好的目标框检测模型的训练原理相同，在此不再进一步阐述。
本申请实施例首先利用预先训练的车牌目标框检测模型检测待识别车牌图片的车牌目标框，可以提高车牌检测的环境适应性，增加车牌检出率；其次，本申请实施例利用预先训练的关键点检测模型检测所述车牌目标框的车牌关键点，可以直接定位车牌四个角，从而可以直接提取出车牌位置，不用再定位车牌的上下边界和左右边界，以使车牌关键点检测更加准确的同时也提高了车牌关键点的检测效率；进一步地，本申请实施例根据所述车牌关键点，对所述待识别车牌图片进行车牌截取，得到目标车牌，并利用预先训练的车牌字符识别模型识别出所述目标车牌的车牌字符，得到车牌信息，可以直接对目标车牌进行字符识别，不需要进行字符分割，提高车牌字符识别的速度。因此，本申请提出的一种车牌识别装置可以提高车牌识别的识别效率。
如图6所示，是本申请实现车牌识别方法的电子设备的结构示意图。
所述电子设备1可以包括处理器10、存储器11和总线，还可以包括存储在所述存储器11中并可在所述处理器10上运行的计算机程序，如车牌识别程序12。
其中，所述存储器11至少包括一种类型的可读存储介质，所述可读存储介质包括闪存、移动硬盘、多媒体卡、卡型存储器(例如：SD或DX存储器等)、磁性存储器、磁盘、光盘等。所述存储器11在一些实施例中可以是电子设备1的内部存储单元，例如该电子设备1的移动硬盘。所述存储器11在另一些实施例中也可以是电子设备1的外部存储设备，例如电子设备1上配备的插接式移动硬盘、智能存储卡(Smart Media Card，SMC)、安全数字(Secure Digital，SD)卡、闪存卡(Flash Card)等。进一步地，所述存储器11还可以既包括电子设备1的内部存储单元也包括外部存储设备。所述存储器11不仅可以用于存储安装于电子设备1的应用软件及各类数据，例如车牌识别程序的代码等，还可以用于暂时地存储已经输出或者将要输出的数据。
所述处理器10在一些实施例中可以由集成电路组成，例如可以由单个封装的集成电路所组成，也可以是由多个相同功能或不同功能封装的集成电路所组成，包括一个或者多个中央处理器(Central Processing unit，CPU)、微处理器、数字处理芯片、图形处理器及各种控制芯片的组合等。所述处理器10是所述电子设备的控制核心(Control Unit)，利用各种接口和线路连接整个电子设备的各个部件，通过运行或执行存储在所述存储器11内的程序或者模块(例如执行车牌识别程序等)，以及调用存储在所述存储器11内的数据，以执行电子设备1的各种功能和处理数据。
所述总线可以是外设部件互连标准(peripheral component interconnect，简称PCI)总线或扩展工业标准结构(extended industry standard architecture，简称EISA)总线等。该总线可以分为地址总线、数据总线、控制总线等。所述总线被设置为实现所述存储器11以及至少一个处理器10等之间的连接通信。
图6仅示出了具有部件的电子设备，本领域技术人员可以理解的是，图6示出的结构 并不构成对所述电子设备1的限定，可以包括比图示更少或者更多的部件，或者组合某些部件，或者不同的部件布置。
例如，尽管未示出，所述电子设备1还可以包括给各个部件供电的电源(比如电池)，优选地，电源可以通过电源管理装置与所述至少一个处理器10逻辑相连，从而通过电源管理装置实现充电管理、放电管理、以及功耗管理等功能。电源还可以包括一个或一个以上的直流或交流电源、再充电装置、电源故障检测电路、电源转换器或者逆变器、电源状态指示器等任意组件。所述电子设备1还可以包括多种传感器、蓝牙模块、Wi-Fi模块等，在此不再赘述。
进一步地，所述电子设备1还可以包括网络接口，可选地，所述网络接口可以包括有线接口和/或无线接口(如WI-FI接口、蓝牙接口等)，通常用于在该电子设备1与其他电子设备之间建立通信连接。
可选地，该电子设备1还可以包括用户接口，用户接口可以是显示器(Display)、输入单元(比如键盘(Keyboard))，可选地，用户接口还可以是标准的有线接口、无线接口。可选地，在一些实施例中，显示器可以是LED显示器、液晶显示器、触控式液晶显示器以及OLED(Organic Light-Emitting Diode，有机发光二极管)触摸器等。其中，显示器也可以适当的称为显示屏或显示单元，用于显示在电子设备1中处理的信息以及用于显示可视化的用户界面。
应该了解，所述实施例仅为说明之用，在专利申请范围上并不受此结构的限制。
所述电子设备1中的所述存储器11存储的车牌识别程序12是多个指令的组合，在所述处理器10中运行时，可以实现：
利用预先训练的车牌目标框检测模型检测待识别车牌图片的车牌目标框；
利用预先训练的关键点检测模型检测所述车牌目标框的车牌关键点；
根据所述车牌关键点，对所述待识别车牌图片进行车牌截取，得到目标车牌；
利用预先训练的车牌字符识别模型识别出所述目标车牌的车牌字符，得到车牌信息。具体地，所述处理器10对上述指令的具体实现方法可参考图1对应实施例中相关步骤的描述，在此不赘述。
进一步地，所述电子设备1集成的模块/单元如果以软件功能单元的形式实现并作为独立的产品销售或使用时，可以存储在一个计算机可读存储介质中。所述计算机可读存储介质可以是非易失性的，也可以是非易失性的。例如：所述计算机可读存储介质可以包括：能够携带所述计算机程序代码的任何实体或装置、记录介质、U盘、移动硬盘、磁碟、光盘、计算机存储器、只读存储器(ROM，Read-Only Memory)。
本申请实施例中，所述计算机可读存储介质存储有车牌识别程序，所述车牌识别程序包括至少一个指令，该指令在被处理器执行时，可以实现：
利用预先训练的车牌目标框检测模型检测待识别车牌图片的车牌目标框；
利用预先训练的关键点检测模型检测所述车牌目标框的车牌关键点；
根据所述车牌关键点，对所述待识别车牌图片进行车牌截取，得到目标车牌；
利用预先训练的车牌字符识别模型识别出所述目标车牌的车牌字符，得到车牌信息。
具体地，所述处理器对上述至少一个车牌识别指令的具体实现方法可参考图1对应实施例中相关步骤的描述，在此不赘述。
在本申请所提供的几个实施例中，应该理解到，所揭露的设备，装置和方法，可以通过其它的方式实现。例如，以上所描述的装置实施例仅仅是示意性的，例如，所述模块的划分，仅仅为一种逻辑功能划分，实际实现时可以有另外的划分方式。
所述作为分离部件说明的模块可以是或者也可以不是物理上分开的，作为模块显示的部件可以是或者也可以不是物理单元，即可以位于一个地方，或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部模块来实现本实施例方案的目的。
另外，在本申请各个实施例中的各功能模块可以集成在一个处理单元中，也可以是各个单元单独物理存在，也可以两个或两个以上单元集成在一个单元中。上述集成的单元既可以采用硬件的形式实现，也可以采用硬件加软件功能模块的形式实现。
对于本领域技术人员而言，显然本申请不限于上述示范性实施例的细节，而且在不背离本申请的精神或基本特征的情况下，能够以其他的具体形式实现本申请。
因此，无论从哪一点来看，均应将实施例看作是示范性的，而且是非限制性的，本申请的范围由所附权利要求而不是上述说明限定，因此旨在将落在权利要求的等同要件的含义和范围内的所有变化涵括在本申请内。不应将权利要求中的任何附关联图标记视为限制所涉及的权利要求。
本申请所指区块链是分布式数据存储、点对点传输、共识机制、加密算法等计算机技术的新型应用模式。区块链(Blockchain)，本质上是一个去中心化的数据库，是一串使用密码学方法相关联产生的数据块，每一个数据块中包含了一批次网络交易的信息，用于验证其信息的有效性(防伪)和生成下一个区块。区块链可以包括区块链底层平台、平台产品服务层以及应用服务层等。
此外，显然“包括”一词不排除其他单元或步骤，单数不排除复数。***权利要求中陈述的多个单元或装置也可以由一个单元或装置通过软件或者硬件来实现。第二等词语用来表示名称，而并不表示任何特定的顺序。
最后应说明的是，以上实施例仅用以说明本申请的技术方案而非限制，尽管参照较佳实施例对本申请进行了详细说明，本领域的普通技术人员应当理解，可以对本申请的技术方案进行修改或等同替换，而不脱离本申请技术方案的精神和范围。

Claims (20)

  1. 一种车牌识别方法，其中，所述方法包括：
    利用预先训练的车牌目标框检测模型检测待识别车牌图片的车牌目标框；
    利用预先训练的关键点检测模型检测所述车牌目标框的车牌关键点；
    根据所述车牌关键点，对所述待识别车牌图片进行车牌截取，得到目标车牌；
    利用预先训练的车牌字符识别模型识别出所述目标车牌的车牌字符，得到车牌信息。
  2. 如权利要求1所述的车牌识别方法，其中，在所述利用预先训练的车牌目标框检测模型检测待识别车牌图片的车牌目标框之前，该方法还包括：
    获取车牌图片集及车牌字符数据集，对所述车牌图片集进行车牌目标框和车牌关键点的标注，分别生成车牌目标框图片集和车牌关键点图片集；
    利用所述车牌目标框图片集对预构建的目标框检测模型进行训练，得到所述预先训练的车牌目标框检测模型；
    利用所述车牌关键点图片集对预构建的关键点检测模型进行训练，得到所述预先训练的车牌关键点检测模型；
    利用所述车牌字符数据集对预构建的字符识别模型进行训练，得到所述预先训练的车牌字符识别模型。
  3. 如权利要求2所述的车牌识别方法，其中，所述利用所述车牌目标框图片集对预构建的目标框检测模型进行训练，得到所述预先训练的车牌目标框检测模型，包括：
    获取所述车牌目标框图片集对应的标签值；
    利用所述目标框检测模型中的卷积层对所述车牌目标框图片集进行卷积操作，得到所述车牌目标框图片集的特征向量，利用所述目标框检测模型中的池化层对所述特征向量进行池化操作，利用所述目标框检测模型中的激活层对池化后的所述特征向量进行计算，得到所述车牌目标框图片集的训练值；
    计算所述训练值与对应标签值的损失值，根据所述损失值调整所述预构建的目标框检测模型的参数，直至所述损失值不大于预设的损失值，得到所述车牌目标框检测模型。
  4. 如权利要求3所述的车牌识别方法，其中，所述计算所述训练值与对应标签值的损失值包括：
    利用下述方法计算所述损失值：
    Figure PCTCN2021097068-appb-100001
    其中，L(s)表示损失值，k表示车牌目标框图片集的数量，y i表示第i个车牌目标框图片的标签值，y′ i表示第i个车牌目标框图片的训练值。
  5. 如权利要求1所述的车牌识别方法，其中，在所述利用预先训练的车牌目标框检测模型检测待识别车牌图片的车牌目标框之前，该方法还包括：
    对所述待识别车牌图片进行图像均衡化处理和/或对所述待识别车牌图片进行角度旋转。
  6. 如权利要求1所述的车牌识别方法，其中，在所述利用预先训练的关键点检测模型检测所述车牌目标框的车牌关键点之前，该方法还包括：
    识别所述待识别车牌图片是否具有多个车牌目标框；
    若所述待识别车牌图片不具有多个车牌目标框，则利用所述关键点检测模型检测所述待识别车牌图片的车牌关键点；
    若所述待识别车牌图片具有多个车牌目标框，则根据识别的车牌目标框，对所述待识别车牌图片进行车牌目标框切分，得到多个子车牌目标框，并利用所述关键点检测模型检测每一个子车牌目标框的车牌关键点。
  7. 如权利要求1至6中任意一项所述的车牌识别方法，其中，所述根据所述车牌关键点，对所述待识别车牌图片进行车牌截取，得到目标车牌，包括：
    基于所述车牌关键点，对所述待识别车牌图片进行车牌畸变纠正，得到初始车牌；
    基于预设标准车牌的尺寸，对所述初始车牌进行尺寸配置，得到标准车牌；
    利用预设图片裁剪工具对所述标准车牌进行裁剪，得到所述目标车牌。
  8. 一种车牌识别装置，其中，所述装置包括：
    检测模块，用于利用预先训练的车牌目标框检测模型检测待识别车牌图片的车牌目标框；
    所述检测模块，还用于利用预先训练的关键点检测模型检测所述车牌目标框的车牌关键点；
    截取模块，用于根据所述车牌关键点，对所述待识别车牌图片进行车牌截取，得到目标车牌；
    识别模块，用于利用预先训练的车牌字符识别模型识别出所述目标车牌的车牌字符，得到车牌信息。
  9. 一种电子设备，其中，所述电子设备包括：
    至少一个处理器；以及，
    与所述至少一个处理器通信连接的存储器；其中，
    所述存储器存储有可被所述至少一个处理器执行的指令，所述指令被所述至少一个处理器执行，以使所述至少一个处理器能够执行如下步骤：
    利用预先训练的车牌目标框检测模型检测待识别车牌图片的车牌目标框；
    利用预先训练的关键点检测模型检测所述车牌目标框的车牌关键点；
    根据所述车牌关键点，对所述待识别车牌图片进行车牌截取，得到目标车牌；
    利用预先训练的车牌字符识别模型识别出所述目标车牌的车牌字符，得到车牌信息。
  10. 如权利要求9所述的电子设备，其中，在所述利用预先训练的车牌目标框检测模型检测待识别车牌图片的车牌目标框之前，所述指令被所述至少一个处理器执行时还实现如下步骤：
    获取车牌图片集及车牌字符数据集，对所述车牌图片集进行车牌目标框和车牌关键点的标注，分别生成车牌目标框图片集和车牌关键点图片集；
    利用所述车牌目标框图片集对预构建的目标框检测模型进行训练，得到所述预先训练的车牌目标框检测模型；
    利用所述车牌关键点图片集对预构建的关键点检测模型进行训练，得到所述预先训练的车牌关键点检测模型；
    利用所述车牌字符数据集对预构建的字符识别模型进行训练，得到所述预先训练的车牌字符识别模型。
  11. 如权利要求10所述的电子设备，其中，所述利用所述车牌目标框图片集对预构建的目标框检测模型进行训练，得到所述预先训练的车牌目标框检测模型，包括：
    获取所述车牌目标框图片集对应的标签值；
    利用所述目标框检测模型中的卷积层对所述车牌目标框图片集进行卷积操作，得到所述车牌目标框图片集的特征向量，利用所述目标框检测模型中的池化层对所述特征向量进行池化操作，利用所述目标框检测模型中的激活层对池化后的所述特征向量进行计算，得到所述车牌目标框图片集的训练值；
    计算所述训练值与对应标签值的损失值，根据所述损失值调整所述预构建的目标框检测模型的参数，直至所述损失值不大于预设的损失值，得到所述车牌目标框检测模型。
  12. 如权利要求11所述的电子设备，其中，所述计算所述训练值与对应标签值的损失值包括：
    利用下述方法计算所述损失值：
    Figure PCTCN2021097068-appb-100002
    其中，L(s)表示损失值，k表示车牌目标框图片集的数量，y i表示第i个车牌目标框图片的标签值，y′ i表示第i个车牌目标框图片的训练值。
  13. 如权利要求9所述的电子设备，其中，在所述利用预先训练的车牌目标框检测模型检测待识别车牌图片的车牌目标框之前，所述指令被所述至少一个处理器执行时还实现如下步骤：
    对所述待识别车牌图片进行图像均衡化处理和/或对所述待识别车牌图片进行角度旋转。
  14. 如权利要求9所述的电子设备，其中，在所述利用预先训练的关键点检测模型检测所述车牌目标框的车牌关键点之前，所述指令被所述至少一个处理器执行时还实现如下步骤：
    识别所述待识别车牌图片是否具有多个车牌目标框；
    若所述待识别车牌图片不具有多个车牌目标框，则利用所述关键点检测模型检测所述待识别车牌图片的车牌关键点；
    若所述待识别车牌图片具有多个车牌目标框，则根据识别的车牌目标框，对所述待识别车牌图片进行车牌目标框切分，得到多个子车牌目标框，并利用所述关键点检测模型检测每一个子车牌目标框的车牌关键点。
  15. 如权利要求9至14中任意一项所述的电子设备，其中，所述根据所述车牌关键点，对所述待识别车牌图片进行车牌截取，得到目标车牌，包括：
    基于所述车牌关键点，对所述待识别车牌图片进行车牌畸变纠正，得到初始车牌；
    基于预设标准车牌的尺寸，对所述初始车牌进行尺寸配置，得到标准车牌；
    利用预设图片裁剪工具对所述标准车牌进行裁剪，得到所述目标车牌。
  16. 一种计算机可读存储介质，存储有至少一个指令，其中，所述至少一个指令被电子设备中的处理器执行以实现如下步骤：
    利用预先训练的车牌目标框检测模型检测待识别车牌图片的车牌目标框；
    利用预先训练的关键点检测模型检测所述车牌目标框的车牌关键点；
    根据所述车牌关键点，对所述待识别车牌图片进行车牌截取，得到目标车牌；
    利用预先训练的车牌字符识别模型识别出所述目标车牌的车牌字符，得到车牌信息。
  17. 如权利要求16所述的计算机可读存储介质，其中，在所述利用预先训练的车牌目标框检测模型检测待识别车牌图片的车牌目标框之前，所述至少一个指令被电子设备中的处理器执行时还实现如下步骤：
    获取车牌图片集及车牌字符数据集，对所述车牌图片集进行车牌目标框和车牌关键点的标注，分别生成车牌目标框图片集和车牌关键点图片集；
    利用所述车牌目标框图片集对预构建的目标框检测模型进行训练，得到所述预先训练的车牌目标框检测模型；
    利用所述车牌关键点图片集对预构建的关键点检测模型进行训练，得到所述预先训练的车牌关键点检测模型；
    利用所述车牌字符数据集对预构建的字符识别模型进行训练，得到所述预先训练的车牌字符识别模型。
  18. 如权利要求17所述的计算机可读存储介质，其中，所述利用所述车牌目标框图片集对预构建的目标框检测模型进行训练，得到所述预先训练的车牌目标框检测模型，包括：
    获取所述车牌目标框图片集对应的标签值；
    利用所述目标框检测模型中的卷积层对所述车牌目标框图片集进行卷积操作，得到所述车牌目标框图片集的特征向量，利用所述目标框检测模型中的池化层对所述特征向量进行池化操作，利用所述目标框检测模型中的激活层对池化后的所述特征向量进行计算，得到所述车牌目标框图片集的训练值；
    计算所述训练值与对应标签值的损失值，根据所述损失值调整所述预构建的目标框检测模型的参数，直至所述损失值不大于预设的损失值，得到所述车牌目标框检测模型。
  19. 如权利要求18所述的计算机可读存储介质，其中，所述计算所述训练值与对应标签值的损失值包括：
    利用下述方法计算所述损失值：
    Figure PCTCN2021097068-appb-100003
    其中，L(s)表示损失值，k表示车牌目标框图片集的数量，y i表示第i个车牌目标框图片的标签值，y′ i表示第i个车牌目标框图片的训练值。
  20. 如权利要求16所述的计算机可读存储介质，其中，在所述利用预先训练的车牌目标框检测模型检测待识别车牌图片的车牌目标框之前，所述至少一个指令被电子设备中的处理器执行时还实现如下步骤：
    对所述待识别车牌图片进行图像均衡化处理和/或对所述待识别车牌图片进行角度旋转。
PCT/CN2021/097068 2020-09-03 2021-05-30 车牌识别方法、装置、电子设备及存储介质 WO2022048209A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202010916527.4A CN112052850B (zh) 2020-09-03 2020-09-03 车牌识别方法、装置、电子设备及存储介质
CN202010916527.4 2020-09-03

Publications (1)

Publication Number Publication Date
WO2022048209A1 true WO2022048209A1 (zh) 2022-03-10

Family

ID=73607337

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/097068 WO2022048209A1 (zh) 2020-09-03 2021-05-30 车牌识别方法、装置、电子设备及存储介质

Country Status (2)

Country Link
CN (1) CN112052850B (zh)
WO (1) WO2022048209A1 (zh)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115001992A (zh) * 2022-05-16 2022-09-02 成都华迈通信技术有限公司 道闸数据采集方法、***、可读存储介质、电子设备
CN115019297A (zh) * 2022-08-04 2022-09-06 之江实验室 一种基于颜色增广的实时车牌检测识别方法和装置
CN115049895A (zh) * 2022-06-15 2022-09-13 北京百度网讯科技有限公司 一种图像属性识别方法、属性识别模型训练方法及装置
CN116958954A (zh) * 2023-07-27 2023-10-27 匀熵智能科技(无锡)有限公司 基于关键点与旁路矫正的车牌识别方法、装置及存储介质

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111191653A (zh) * 2019-12-25 2020-05-22 北京精英路通科技有限公司 车牌识别方法、装置、计算机设备及存储介质
CN112052850B (zh) * 2020-09-03 2024-05-10 平安科技（深圳）有限公司 车牌识别方法、装置、电子设备及存储介质
CN112906643A (zh) * 2021-03-22 2021-06-04 高新兴科技集团股份有限公司 车牌号码识别方法及装置
CN114399766B (zh) * 2022-01-18 2024-05-10 平安科技（深圳）有限公司 光学字符识别模型训练方法、装置、设备及介质

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107679531A (zh) * 2017-06-23 2018-02-09 平安科技（深圳）有限公司 基于深度学习的车牌识别方法、装置、设备及存储介质
CN108229474A (zh) * 2017-12-29 2018-06-29 北京旷视科技有限公司 车牌识别方法、装置及电子设备
US20190251369A1 (en) * 2018-02-11 2019-08-15 Ilya Popov License plate detection and recognition system
CN110866524A (zh) * 2019-11-15 2020-03-06 北京字节跳动网络技术有限公司 车牌检测方法、装置、设备及存储介质
CN112052850A (zh) * 2020-09-03 2020-12-08 平安科技（深圳）有限公司 车牌识别方法、装置、电子设备及存储介质

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110298278B (zh) * 2019-06-19 2021-06-04 中国计量大学 一种基于人工智能的地下停车库行人车辆监测方法
CN110781882A (zh) * 2019-09-11 2020-02-11 南京钰质智能科技有限公司 一种基于yolo模型的车牌定位和识别方法
CN111191653A (zh) * 2019-12-25 2020-05-22 北京精英路通科技有限公司 车牌识别方法、装置、计算机设备及存储介质
CN111435446A (zh) * 2019-12-25 2020-07-21 珠海大横琴科技发展有限公司 一种基于LeNet车牌识别方法及装置

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107679531A (zh) * 2017-06-23 2018-02-09 平安科技（深圳）有限公司 基于深度学习的车牌识别方法、装置、设备及存储介质
CN108229474A (zh) * 2017-12-29 2018-06-29 北京旷视科技有限公司 车牌识别方法、装置及电子设备
US20190251369A1 (en) * 2018-02-11 2019-08-15 Ilya Popov License plate detection and recognition system
CN110866524A (zh) * 2019-11-15 2020-03-06 北京字节跳动网络技术有限公司 车牌检测方法、装置、设备及存储介质
CN112052850A (zh) * 2020-09-03 2020-12-08 平安科技（深圳）有限公司 车牌识别方法、装置、电子设备及存储介质

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115001992A (zh) * 2022-05-16 2022-09-02 成都华迈通信技术有限公司 道闸数据采集方法、***、可读存储介质、电子设备
CN115001992B (zh) * 2022-05-16 2024-04-05 成都华迈通信技术有限公司 道闸数据采集方法、***、可读存储介质、电子设备
CN115049895A (zh) * 2022-06-15 2022-09-13 北京百度网讯科技有限公司 一种图像属性识别方法、属性识别模型训练方法及装置
CN115049895B (zh) * 2022-06-15 2024-01-05 北京百度网讯科技有限公司 一种图像属性识别方法、属性识别模型训练方法及装置
CN115019297A (zh) * 2022-08-04 2022-09-06 之江实验室 一种基于颜色增广的实时车牌检测识别方法和装置
CN115019297B (zh) * 2022-08-04 2022-12-09 之江实验室 一种基于颜色增广的实时车牌检测识别方法和装置
CN116958954A (zh) * 2023-07-27 2023-10-27 匀熵智能科技(无锡)有限公司 基于关键点与旁路矫正的车牌识别方法、装置及存储介质
CN116958954B (zh) * 2023-07-27 2024-03-22 匀熵智能科技(无锡)有限公司 基于关键点与旁路矫正的车牌识别方法、装置及存储介质

Also Published As

Publication number Publication date
CN112052850A (zh) 2020-12-08
CN112052850B (zh) 2024-05-10

Similar Documents

Publication Publication Date Title
WO2022048209A1 (zh) 车牌识别方法、装置、电子设备及存储介质
WO2021189906A1 (zh) 基于联邦学习的目标检测方法、装置、设备及存储介质
WO2018205467A1 (zh) 车损部位的识别方法、***、电子装置及存储介质
WO2022105179A1 (zh) 生物特征图像识别方法、装置、电子设备及可读存储介质
WO2021189855A1 (zh) 基于ct序列的图像识别方法、装置、电子设备及介质
CN112699775A (zh) 基于深度学习的证件识别方法、装置、设备及存储介质
WO2021151338A1 (zh) 医学影像图片分析方法、装置、电子设备及可读存储介质
CN112801718B (zh) 用户行为预测方法、装置、设备及介质
US11734954B2 (en) Face recognition method, device and electronic equipment, and computer non-volatile readable storage medium
CN111739016B (zh) 目标检测模型训练方法、装置、电子设备及存储介质
CN111738212B (zh) 基于人工智能的交通信号灯识别方法、装置、设备及介质
CN113033543B (zh) 曲形文本识别方法、装置、设备及介质
WO2022126914A1 (zh) 活体检测方法、装置、电子设备及存储介质
EP3944145A2 (en) Method and device for training image recognition model, equipment and medium
CN111178357B (zh) 车牌识别方法、***、设备和存储介质
WO2022141858A1 (zh) 行人检测方法、装置、电子设备及存储介质
US11106933B2 (en) Method, device and system for processing image tagging information
US11481577B2 (en) Machine learning (ML) quality assurance for data curation
CN113313114B (zh) 证件信息获取方法、装置、设备以及存储介质
CN112528903B (zh) 人脸图像获取方法、装置、电子设备及介质
WO2023134080A1 (zh) 相机作弊识别方法、装置、设备及存储介质
WO2022227191A1 (zh) 非主动活体检测方法、装置、电子设备及存储介质
CN112434601B (zh) 基于行车视频的车辆违法检测方法、装置、设备及介质
US20220277540A1 (en) System and method for generating an optimized image with scribble-based annotation of images using a machine learning model
CN114882420A (zh) 接待人数统计方法、装置、电子设备及可读存储介质

Legal Events

Date Code Title Description
NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205 DATED 23/06/2023)

122 Ep: pct application non-entry in european phase

Ref document number: 21863279

Country of ref document: EP

Kind code of ref document: A1