WO2023272991A1 - 一种数据处理方法、装置、计算机设备和存储介质 - Google Patents

一种数据处理方法、装置、计算机设备和存储介质 Download PDF

Info

Publication number
WO2023272991A1
WO2023272991A1 PCT/CN2021/121759 CN2021121759W WO2023272991A1 WO 2023272991 A1 WO2023272991 A1 WO 2023272991A1 CN 2021121759 W CN2021121759 W CN 2021121759W WO 2023272991 A1 WO2023272991 A1 WO 2023272991A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
vehicle
information
sub
processed
Prior art date
Application number
PCT/CN2021/121759
Other languages
English (en)
French (fr)
Inventor
何智群
刘钢
武伟
闫俊杰
Original Assignee
深圳市商汤科技有限公司
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 深圳市商汤科技有限公司 filed Critical 深圳市商汤科技有限公司
Publication of WO2023272991A1 publication Critical patent/WO2023272991A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods

Definitions

  • the present disclosure relates to the field of computer technology and image processing, in particular, to a data processing method, device, computer equipment and storage medium.
  • analyzing vehicle feature information plays an important role in understanding digital traffic scenarios and retrieving traffic vehicles. For example, in the scene of traffic detection vehicle toll evasion, it is necessary to ensure that the obtained information is accurate and true, and can fully identify the characteristics of the current vehicle, so as to reduce the probability of failure or misjudgment of vehicle toll evasion detection caused by wrong vehicle characteristic information.
  • Embodiments of the present disclosure at least provide a data processing method, device, computer equipment, and storage medium.
  • an embodiment of the present disclosure provides a data processing method, including:
  • At least one vehicle image is determined; wherein each vehicle image corresponds to a vehicle;
  • the vehicle characteristic information of the target vehicle image is stored in a database.
  • using the first preset condition and image quality information to screen the target vehicle image can improve the accuracy of the extracted vehicle feature information, thereby improving the success rate of escape; in addition, only the screened target vehicle image Feature extraction can reduce the amount of calculation in the process of feature extraction, and can reduce the probability of storing invalid information in the database, saving the storage space of the database.
  • the determining the vehicle feature information of the target vehicle image based on the target vehicle image and the image to be processed includes:
  • vehicle characteristic information of the target vehicle image is determined.
  • the image of the target vehicle is expanded to obtain a more complete image of the vehicle in the image of the target vehicle, and further more comprehensive vehicle feature information can be extracted.
  • the image quality information includes the quality score of the captured vehicle image and the orientation of the vehicle in the vehicle image;
  • the vehicle images whose image quality information meets the first preset condition are screened as target vehicle images, including:
  • the vehicle images whose quality scores are greater than a preset threshold and whose orientation of the vehicle is within a preset orientation range are selected as the target vehicle images.
  • the target vehicle image with high shooting quality and comprehensive feature extraction can be accurately screened out from the vehicle image, so that accurate , Comprehensive vehicle characteristic information.
  • the storing the vehicle characteristic information of the target vehicle image in a library includes:
  • the vehicle feature information corresponding to the identity information can be conveniently and quickly found from the database by using the identity information, and the vehicle feature information can be improved. The efficiency of subsequent vehicle feature information queries.
  • the vehicle feature information includes first sub-information and second sub-information;
  • the mapping relationship includes a first sub-relation and a second sub-relationship;
  • the first sub-information includes at least one of the following: an initial detection frame of the vehicle in the image of the target vehicle, a confidence degree of the initial detection frame, the image quality information, a vehicle attribute, a confidence degree of the vehicle attribute, The detection frame of the license plate in the target vehicle image, the confidence degree of the detection frame of the license plate, the license plate number on the license plate and the confidence degree of the license plate number; the shooting time when the image to be processed is captured by the photographing device; Device information of the shooting device;
  • the second sub-information includes at least one of the following: the feature vector of the vehicle in the target vehicle image and the feature vector of the license plate in the target vehicle image;
  • the establishment of the mapping relationship between the identity information and the vehicle feature information includes:
  • the storing the vehicle feature information, the mapping relationship and the identification information into a database includes:
  • the above-mentioned vehicle feature information can comprehensively and accurately characterize the vehicle in the target vehicle image, using the above-mentioned vehicle feature information helps to improve the success rate of evasion.
  • the data types of the first sub-information and the second sub-information are different, classifying and storing the first sub-information and the second sub-information can facilitate data management and improve data query efficiency.
  • the first sub-relationship is stored in the attribute database, which is convenient to use the identity information to retrieve the first sub-information matching the identity information from the attribute database; the second relationship is stored in the feature database, which is convenient to use the identity
  • the identification information retrieves the second sub-information matching the identification information from the feature database, which can improve the efficiency of subsequent vehicle feature information query.
  • the storing the vehicle feature information, the mapping relationship, and the identity information in a library includes:
  • the vehicle characteristic information includes the confidence degree of the license plate number, it is judged whether the confidence degree of the license plate number is greater than a preset confidence degree;
  • the vehicle characteristic information, the mapping relationship and the identification information are stored in a database.
  • the license plate number when storing vehicle characteristic information, is stored only when the confidence of the license plate number is greater than the preset reliability, which can reduce the probability of storing invalid data and save storage space.
  • the device information includes identification information
  • the storing the vehicle feature information, the mapping relationship and the identification information into a database includes:
  • the vehicle feature information includes the shooting time and/or the device information of the shooting device, if the shooting time conforms to the first preset format, the vehicle feature information, the mapping relationship, and The identity identification information is stored in a database;
  • the identification information complies with the second preset format, storing the vehicle feature information, the mapping relationship, and the identification information into a library.
  • the vehicle characteristic information before storing the vehicle characteristic information, by rationalizing the shooting time and identification information, only when the shooting time conforms to the first preset format and/or the identification information conforms to the second preset format, will the The vehicle characteristic information, mapping relationship and identity information are stored in the database, which can ensure that the stored vehicle characteristic information is preset legal data, reduce the probability of storing invalid data, and save storage space.
  • the determining the vehicle characteristic information of the target vehicle image includes:
  • the storage request includes the shooting time of the image to be processed and/or the device information of the shooting device;
  • the storage request is analyzed to obtain the shooting time and/or device information of the shooting device.
  • the device information includes at least one of the following:
  • the location information of the shooting device The location information of the shooting device; the identification information of the shooting device.
  • the identifying the image to be processed in response to the storage request of the image to be processed includes:
  • the image to be processed meets the second preset condition, the image to be processed is identified.
  • the determining the vehicle feature information of the target vehicle image based on the image obtained by expanding the image to be processed based on the image of the target vehicle includes:
  • the vehicle characteristic information of the target vehicle image can be extracted in a more detailed manner, and then relatively comprehensive vehicle characteristic information can be obtained.
  • the object attribute includes at least one of the following: the color of the vehicle, the appearance information of the vehicle, and the type of the vehicle.
  • an embodiment of the present disclosure further provides a data processing device, including:
  • An image recognition module configured to identify the image to be processed in response to the storage request of the image to be processed
  • the first determination module is configured to determine at least one vehicle image when it is identified that the image to be processed contains a vehicle; wherein each vehicle image corresponds to a vehicle;
  • a second determining module configured to determine the image quality information of the vehicle image
  • An image screening module configured to screen vehicle images whose image quality information meets a first preset condition from the vehicle images as target vehicle images
  • a third determining module configured to determine vehicle characteristic information of the target vehicle image based on the target vehicle image and the image to be processed
  • a feature storage module configured to store the vehicle feature information of the target vehicle image into a library.
  • the third determination module is configured to determine the vehicle characteristic information of the target vehicle image based on an image obtained by expanding the target vehicle image on the image to be processed.
  • the image quality information includes the quality score of the captured vehicle image and the orientation of the vehicle in the vehicle image;
  • the image screening module is configured to select, from the vehicle images, vehicle images whose quality scores are greater than a preset threshold and whose orientation of the vehicle is within a preset orientation range, as the target vehicle image.
  • the feature storage module is configured to determine the identity information of the vehicle in the target vehicle image; establish a mapping relationship between the identity information and the vehicle feature information; store the The vehicle characteristic information, the mapping relationship and the identification information are stored in a database.
  • the vehicle feature information includes first sub-information and second sub-information;
  • the mapping relationship includes a first sub-relation and a second sub-relationship;
  • the first sub-information includes at least one of the following: an initial detection frame of the vehicle in the image of the target vehicle, a confidence degree of the initial detection frame, the image quality information, a vehicle attribute, a confidence degree of the vehicle attribute, The detection frame of the license plate in the target vehicle image, the confidence degree of the detection frame of the license plate, the license plate number on the license plate and the confidence degree of the license plate number; the shooting time when the image to be processed is captured by the photographing device; Device information of the shooting device;
  • the second sub-information includes at least one of the following: the feature vector of the vehicle in the target vehicle image and the feature vector of the license plate in the target vehicle image;
  • the feature storage module is configured to establish a first sub-relationship between the identity information and the first sub-information; establish a second sub-relationship between the identity information and the second sub-information; storing the first sub-information, the first sub-relationship and the identification information into an attribute database; storing the second sub-information, the second sub-relationship and the identification information into a feature database.
  • the characteristic storage module is configured to determine whether the confidence degree of the license plate number is greater than a preset confidence degree when the vehicle characteristic information includes the confidence degree of the license plate number; In the case that the confidence degree of the license plate number is greater than the preset reliability degree, the vehicle characteristic information, the mapping relationship and the identification information are stored in a database.
  • the device information includes identification information
  • the characteristic storage module is configured to, when the vehicle characteristic information includes the shooting time and/or device information of the shooting device, If the shooting time conforms to the first preset format, then store the vehicle characteristic information, the mapping relationship and the identification information into the database; or, if the identification information conforms to the second preset format, store the The vehicle characteristic information, the mapping relationship and the identification information are stored in a database.
  • the third determination module is configured to receive the storage request sent by the client; the storage request includes the shooting time and/or the The device information of the shooting device; analyzing the storage request to obtain the shooting time and/or the device information of the shooting device.
  • the device information includes at least one of the following:
  • the location information of the shooting device The location information of the shooting device; the identification information of the shooting device.
  • the image recognition module is configured to receive the storage request sent by the client, wherein the storage request includes the image to be processed; The image to be processed; if the image to be processed meets a second preset condition, identify the image to be processed.
  • the third determining module is configured to use an image obtained by expanding the target vehicle image on the image to be processed as a first sub-image; based on the first sub-image , determine the detection frame of the license plate in the first sub-image; based on the detection frame of the license plate, intercept the second sub-image from the first sub-image; based on the first sub-image and the vehicle feature detection model, determine The vehicle feature sub-information of the first sub-image; based on the second sub-image and the license plate feature detection model, determine the license plate feature sub-information of the second sub-image; based on the vehicle feature sub-information and the license plate feature
  • the sub-information is used to determine the vehicle feature information of the target vehicle image.
  • the object attribute includes at least one of the following: the color of the vehicle, the appearance information of the vehicle, and the type of the vehicle.
  • an embodiment of the present disclosure further provides a computer device, including: a processor, a memory, and a bus, the memory stores machine-readable instructions executable by the processor, and when the computer device is running, the processing The processor communicates with the memory through a bus, and when the machine-readable instructions are executed by the processor, the above-mentioned first aspect, or the steps in any possible implementation manner of the first aspect are executed.
  • a computer device including: a processor, a memory, and a bus
  • the memory stores machine-readable instructions executable by the processor, and when the computer device is running, the processing
  • the processor communicates with the memory through a bus, and when the machine-readable instructions are executed by the processor, the above-mentioned first aspect, or the steps in any possible implementation manner of the first aspect are executed.
  • embodiments of the present disclosure further provide a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the above-mentioned first aspect, or any of the first aspects of the first aspect, may be executed. Steps in one possible implementation.
  • the embodiments of the present disclosure further provide a computer program product, including computer readable codes, or a non-volatile computer readable storage medium bearing computer readable codes, when the computer readable codes are stored in an electronic device
  • the processor in the electronic device executes the steps for realizing the above first aspect, or any possible implementation manner of the first aspect.
  • FIG. 1 shows a flowchart of a data processing method provided by an embodiment of the present disclosure
  • FIG. 2 shows a flow chart of determining vehicle characteristic information of a target vehicle image provided by an embodiment of the present disclosure
  • FIG. 3 shows a flow chart of storing and storing vehicle feature information into a database according to an embodiment of the present disclosure
  • Fig. 4 shows a flow chart of storing vehicle feature information in an application scenario provided by an embodiment of the present disclosure
  • FIG. 5 shows a schematic diagram of corresponding storage and warehousing of vehicle structured information provided by an embodiment of the present disclosure
  • FIG. 6 shows a schematic diagram of a data processing device provided by an embodiment of the present disclosure
  • FIG. 7 shows a schematic structural diagram of a computer device provided by an embodiment of the present disclosure.
  • the execution subject of the data processing method provided by this embodiment of the present disclosure is generally a computer device with certain computing capabilities.
  • the data processing method may be implemented by a processor invoking computer-readable instructions stored in a memory.
  • FIG. 1 is a flow chart of a data processing method provided by an embodiment of the present disclosure, the method includes steps S101 to S106, wherein:
  • S101 Identify the image to be processed in response to the storage request of the image to be processed.
  • the storage request can be an http (hypertext transfer protocol) request initiated by the user, and the http request includes images to be processed. Specifically, in response to the http request, the image to be processed carried in the http request is identified.
  • http hypertext transfer protocol
  • a trained neural network model can be used to identify the image to be processed, and identify whether there is a vehicle in the image to be processed.
  • the neural network may include at least one of the following: convolutional neural network (Convolutional Neural Networks, CNN), regional convolutional neural network (Regions Region-based Convolutional Network, R-CNN), fast regional convolutional neural network (Fast Region-based Convolutional Network, Fast R-CNN), Faster Region-based Convolutional Network (Faster R-CNN), etc.
  • the storage request further includes the shooting time of the image to be processed and/or the device information of the shooting device.
  • the storage request sent by the user's client terminal is received; the storage request is analyzed to obtain the shooting time and/or device information of the shooting device.
  • the device information may include at least one of the following: location information of the shooting device; identification information of the shooting device.
  • the identification information of the shooting device may also be used to indicate the location of the shooting device.
  • S102 Determine at least one vehicle image when it is identified that the image to be processed contains a vehicle.
  • the convolutional neural network model of the detection frame of the vehicle can be used to perform structural detection on the image to be processed, and it can be judged whether there is a vehicle in the image to be processed. If there is a vehicle, the convolutional neural network model can be used to output The initial detection frame of the vehicle and the confidence of the initial detection frame. Afterwards, in the image to be processed, the image of the area marked with the initial detection frame of the vehicle is used as the vehicle image. Since one image to be processed may contain multiple vehicles, the initial detection frames of multiple vehicles can be identified, and then multiple vehicle images can be obtained.
  • step S101 is executed in a loop to wait for the next warehousing request.
  • the image quality information may be an index indicating the picture quality of the vehicle image and/or the orientation of the vehicle in the vehicle image.
  • the picture quality can include several aspects such as the clarity, sharpness, dispersion, resolution, color gamut, color purity, and color balance of the vehicle image.
  • a neural network model for detecting image quality information may be used to output image quality information.
  • the image quality information may include a quality score for evaluating the picture quality of the vehicle image and orientation information for evaluating the reasonableness of the position of the vehicle in the vehicle image.
  • the better the picture quality of the vehicle image the higher the quality score.
  • the captured high-speed moving vehicle may appear blurred; Therefore, it is necessary to perform quality inspection on the vehicle image to determine the quality information of the vehicle image in order to determine more accurate vehicle feature information in the future.
  • S104 Select vehicle images whose image quality information meets a first preset condition from vehicle images as target vehicle images.
  • the first preset condition may include a condition that the quality score is greater than a preset threshold; for the orientation information of a reasonable degree of the position of the vehicle indicated by the image quality information, The first preset condition may also include a preset orientation range of the orientation of the vehicle, such as the orientation of the vehicle in which the front or the rear of the vehicle can be seen.
  • the vehicle images that do not meet the first preset condition can be filtered, and only the vehicle images that meet the first preset condition are retained, and the vehicle image that meets the first preset condition is used as the target vehicle Image, extracting and storing subsequent vehicle feature information for the target vehicle image can obtain more accurate vehicle feature information.
  • vehicle images with quality scores greater than a preset threshold can be selected from vehicle images as target vehicle images; or, vehicle images can be selected with quality scores greater than a preset threshold and the orientation of the vehicle is within a predetermined
  • the vehicle image within the heading range is set as the target vehicle image.
  • the quality score of the vehicle image is less than or equal to the preset threshold, so the vehicle image does not meet the criteria for continued detection, and the vehicle image will not be Then carry out the processing of the subsequent process; for the case where the orientation of the captured vehicle is sideways, it can be determined that the orientation of the vehicle is not within the preset orientation range, so the vehicle does not meet the criteria for continued detection.
  • the vehicle image where the vehicle is located Subsequent processes will not be processed, which can relieve part of the pressure of invalid data feature detection for subsequent processes, thereby reducing the probability of storing invalid information in the database.
  • S105 Determine vehicle characteristic information of the target vehicle image based on the target vehicle image and the image to be processed.
  • the target vehicle image can be expanded based on the image to be processed to obtain the complete target vehicle image after expansion.
  • the vehicle feature information of the target vehicle image is determined based on an image obtained by expanding the target vehicle image on the image to be processed.
  • the external expansion process can be centered on the geometric center of the initial detection frame in the target vehicle image, and the long side and the wide side corresponding to the initial detection frame are respectively expanded by 1.2 times to determine the target detection frame, and the target detection frame framed
  • the image that is, the image obtained by the expansion process, can include a complete vehicle.
  • vehicle feature information includes feature information of different data types
  • neural network models of different feature extraction types for feature extraction.
  • vehicle feature detection models can be used to extract vehicle feature sub-information
  • license plate The feature detection model extracts the feature sub-information of the license plate.
  • FIG. 2 is a flow chart of determining the vehicle feature information of the target vehicle image, which includes steps S201-S206:
  • S201 Use an image obtained by expanding the image of the target vehicle on the image to be processed as a first sub-image.
  • the image framed by the target detection frame may be used as the first sub-image.
  • the vehicle position information indicated by the target detection frame may be used to intercept the first sub-image from the image to be processed.
  • S202 Based on the first sub-image, determine a detection frame of the license plate in the first sub-image.
  • a convolutional neural network model for detecting a license plate detection frame may be used to perform structured detection on the first sub-image, and output a license plate detection frame.
  • S203 Based on the detection frame of the license plate, intercept the second sub-image from the first sub-image.
  • the image framed by the detection frame of the license plate may be used as the second sub-image.
  • the second sub-image may be intercepted from the first sub-image by using the position information of the license plate indicated by the detection frame of the license plate.
  • S204 Based on the first sub-image and the vehicle feature detection model, determine vehicle feature sub-information of the first sub-image.
  • the vehicle sub-feature information may include vehicle attributes, confidence levels of vehicle attributes, and vehicle feature vectors.
  • the vehicle attribute includes at least one of the following: the color of the vehicle, the appearance information of the vehicle and the type of the vehicle.
  • the vehicle feature detection model may include a neural network model for detecting vehicle attributes and a neural network model for detecting vehicle feature vectors.
  • the neural network model for detecting vehicle attributes may be used to output the color of the vehicle in the first sub-image and the confidence of the color of the vehicle, the appearance information of the vehicle and the confidence of the appearance information of the vehicle, the type of the vehicle and the confidence of the type of the vehicle ;
  • the neural network model for detecting the feature vector of the vehicle may be used to output the feature vector of the vehicle in the first sub-image, where the feature vector of the vehicle may represent the feature of the vehicle.
  • S205 Based on the second sub-image and the license plate feature detection model, determine the license plate feature sub-information of the second sub-image.
  • the license plate feature sub-information may include the detection frame of the license plate, the confidence of the detection frame of the license plate, the license plate number on the license plate, the confidence of the license plate number, and the feature vector of the license plate.
  • the license plate feature detection model includes a neural network model for detecting the detection frame of the license plate, a neural network model for detecting the license plate number, and a neural network model for detecting the feature vector of the license plate.
  • the neural network model for detecting the feature vector of the license plate can be used to output the feature vector of the license plate in the second sub-image, where the feature vector of the license plate can represent the feature of the license plate.
  • S206 Determine vehicle feature information of the target vehicle image based on the vehicle feature sub-information and the license plate feature sub-information.
  • the vehicle feature information includes: vehicle attributes, the confidence of the vehicle attributes, and the feature vector of the vehicle; the detection frame of the license plate, the confidence of the detection frame of the license plate, the license plate number on the license plate, the confidence of the license plate number, and the feature vector of the license plate .
  • the vehicle feature information also includes: the initial detection frame of the vehicle, the confidence of the initial detection frame, image quality information, the shooting time when the image to be processed is captured by the shooting device, and the device information of the shooting device.
  • S106 Store the vehicle feature information of the target vehicle image into a library.
  • the vehicle feature information includes information of different data types
  • the information of different data types needs to be classified and stored.
  • the determined vehicle feature information may be classified and processed, that is, the first sub-information and the second sub-information, and correspondingly store different types of vehicle feature information in different databases.
  • the first sub-information is stored in the attribute database
  • the second sub-information is stored in the feature database.
  • the first sub-information may include at least one of the following: the initial detection frame of the vehicle in the target vehicle image, the confidence degree of the initial detection frame, image quality information, the vehicle attribute, the confidence degree of the vehicle attribute, the detection frame of the license plate in the target vehicle image, The confidence of the detection frame of the license plate, the license plate number on the license plate, and the confidence of the license plate number; the shooting time when the shooting device captures the image to be processed; the device information of the shooting device.
  • the second sub-information may include at least one of the following: a feature vector of the vehicle in the target vehicle image and a feature vector of the license plate in the target vehicle image.
  • the above-mentioned first sub-information and second sub-information can comprehensively and accurately characterize the vehicle in the target vehicle image, so using the vehicle feature information composed of the first sub-information and the second sub-information helps to improve the success rate of evasion.
  • FIG. 3 is a flow chart of storing and storing vehicle characteristic information. Among them, including steps S301-S303:
  • S301 Determine the identification information of the vehicle in the target vehicle image.
  • determining the identity information of the vehicle in the image of the target vehicle facilitates the use of the identity information to search the database for the vehicle represented by the identity information.
  • the identification information may be the license plate number of the vehicle or a preset number.
  • S302 Establish a mapping relationship between identity information and vehicle characteristic information.
  • a first sub-relationship between the identity information and the first sub-information may be established; a second sub-relationship between the identity information and the second sub-information may be established.
  • S303 Store the vehicle characteristic information, mapping relationship and identification information into a database.
  • the vehicle in the target vehicle image can be marked with a preset number A, and all the information included in the vehicle feature information of the target vehicle image is represented by the number A.
  • the vehicle feature of vehicle A is retrieved from the database When it comes to information, you only need to enter the number A to query all the vehicle characteristic information about the vehicle A.
  • the first sub-information, the first sub-relationship and the identification information may be stored in the attribute database; the second sub-information, the second sub-relationship and the identification information may be stored in the feature database.
  • the attribute database may include a structure database StructDB; the feature database may include a static database (engine-static-feature-db, sfd).
  • the first sub-relationship is stored in the attribute database, which is convenient to use the identity information to retrieve the first sub-information matching the identity information from the attribute database;
  • the second relationship is stored in the feature database, which is convenient to use the identity
  • the identification information retrieves the second sub-information matching the identification information from the feature database, which can improve the efficiency of subsequent vehicle feature information query.
  • the stored vehicle characteristic information before storing the vehicle characteristic information, it is also necessary to determine whether the stored vehicle characteristic information meets the storage requirements.
  • the preset reliability can be specifically set according to different application scenarios, and is not limited here.
  • the first preset format may be XXXX (year)-XX (month)-XX (day).
  • the first preset format may include several different formats representing time, which is not limited here. Since the identification information of different shooting devices is different, the second preset format of the identification information is not specifically limited here.
  • the vehicle feature information when the vehicle feature information includes the shooting time and/or the device information of the shooting device, if the shooting time does not conform to the first preset format, the shooting time will not be stored in the storage, and the shooting time will be saved.
  • the external vehicle characteristic information is stored in the warehouse. If the identification information does not conform to the second preset format, the identification information is not stored in the storage, and the vehicle characteristic information other than the identification information is stored in the storage.
  • the shooting time conforming to the first preset format may be stored in the attribute database together with other vehicle characteristic information. If the shooting time is XXXXXX-XXXXX, etc. that do not conform to the first preset format, the shooting time will not be allowed to be stored in the attribute database, and the vehicle characteristic information other than the shooting time can be stored in the database.
  • the initial detection frame of the vehicle identified in step S102 and the confidence level of the initial detection frame are stored as the first sub-information in the attribute database.
  • the image quality information determined in step S103 is stored as the first sub-information in the attribute database.
  • the image to be processed before the image to be processed is identified in step S101 above, the image to be processed can also be verified, and if the image to be processed meets the second preset condition, the image to be processed can be identified , if the image to be processed does not meet the second preset condition, the image to be processed does not need to be identified, and the next storage request is waited for.
  • the image to be processed of the captured vehicle may be obtained through an http request. Firstly, verification processing is performed on the image to be processed, and when the image to be processed meets the second preset condition, the image to be processed can be identified.
  • the second preset condition may include that the image to be processed has image information, and if the image to be processed is garbled or has no image, the image to be processed does not meet the second preset condition.
  • the second preset condition may also be set according to a specific task, which is not limited here.
  • FIG. 4 is a flow chart of storing vehicle feature information in an application scenario, including steps S401-S408:
  • the content of the http request may include the captured image of the vehicle to be processed, the shooting time when the shooting device captures the image to be processed, and the device information of the shooting device.
  • the device information includes identification information.
  • step S403 verification processing is performed on the image to be processed, and if the image to be processed meets the second preset condition, continue to execute step S403; otherwise, end the process.
  • the shooting time of the image to be processed captured by the shooting device if the shooting time conforms to the first preset format, the shooting time passes the verification and is waiting to be stored; when the identification information of the shooting device conforms to the second preset format Next, the identification information of the shooting device has passed the verification and is waiting to be stored.
  • base64 decoding can be performed on the image to be processed.
  • S404 Perform structured detection on vehicles in the image to be processed.
  • the structured detection may be to input the image to be processed into the convolutional neural network model for processing, so as to determine whether there is a vehicle in the image to be processed.
  • step S405 Determine whether there is a vehicle in the image to be processed, if yes, execute step S306; if not, end the process.
  • step S404 based on the result output by the convolutional neural network model in step S404, that is, the initial detection frame of the output vehicle and the confidence degree of the initial detection frame, if there is an initial detection frame of the vehicle, and the confidence degree of the initial detection frame is greater than preset value, it is determined that there is a vehicle in the image to be processed.
  • step S46 Determine whether the image quality information meets the first preset condition, if so, execute step S407; if not, end the process.
  • step S407 for the quality score in the image quality information, if the image quality score is greater than the preset threshold, and the orientation of the vehicle in the vehicle image is within the preset range, then determine the target vehicle image, and perform step S407; if the quality score If it is less than or equal to the preset threshold, or the orientation of the vehicle in the vehicle image is not within the preset range, the process ends.
  • S408 Extract and store the structural information of the vehicle in the expanded target vehicle image.
  • the structured information of the vehicle may include at least one of the following: the initial detection frame of the vehicle, the confidence degree of the initial detection frame of the vehicle, image quality information, vehicle attributes, the confidence degree of the vehicle attributes, the detection frame of the license plate, the license plate The confidence of the detection frame, the license plate number, the confidence of the license plate number, the feature vector of the vehicle, the feature vector of the license plate, the shooting time when the verification passed, and the identification information of the shooting device that passed the verification.
  • the storage of the structural information of the vehicle may refer to the storage of the vehicle characteristic information in the above step S106, and the repeated part will not be repeated here.
  • using the first preset condition and image quality information to filter the target vehicle image can improve the accuracy of the extracted vehicle feature information, thereby improving the success rate of vehicle toll evasion detection; in addition, only for The feature extraction of the screened target vehicle images can reduce the amount of computation in the process of feature extraction, reduce the probability of storing invalid information in the database, and save the storage space of the database.
  • the writing order of each step does not mean a strict execution order and constitutes any limitation on the implementation process.
  • the specific execution order of each step should be based on its function and possible
  • the inner logic is OK.
  • the embodiment of the present disclosure also provides a data processing device corresponding to the data processing method. Since the problem-solving principle of the device in the embodiment of the present disclosure is similar to the above-mentioned data processing method of the embodiment of the present disclosure, the implementation of the device Reference can be made to the implementation of the method, and repeated descriptions will not be repeated.
  • the device includes: an image recognition module 601, a first determination module 602, a second determination module 603, an image screening module 604, a third Determination module 605 and feature storage module 606; Wherein,
  • An image identification module 601, configured to identify the image to be processed in response to the storage request of the image to be processed
  • the first determination module 602 is configured to determine at least one vehicle image when it is identified that the image to be processed contains a vehicle; wherein, each vehicle image corresponds to a vehicle;
  • the second determination module 603 is configured to determine the image quality information of the vehicle image
  • An image screening module 604 configured to screen vehicle images whose image quality information meets a first preset condition from the vehicle images as target vehicle images;
  • a third determining module 605, configured to determine vehicle characteristic information of the target vehicle image based on the target vehicle image and the image to be processed;
  • the feature storage module 606 is configured to store the vehicle feature information of the target vehicle image into a library.
  • the third determination module 605 is configured to determine the vehicle feature information of the target vehicle image based on an image obtained by expanding the target vehicle image on the image to be processed.
  • the image quality information includes the quality score of the captured vehicle image and the orientation of the vehicle in the vehicle image;
  • the image screening module 604 is configured to select, from the vehicle images, vehicle images whose quality scores are greater than a preset threshold and whose orientation of the vehicle is within a preset orientation range, as the target vehicle image.
  • the feature storage module 606 is configured to determine the identity information of the vehicle in the target vehicle image; establish a mapping relationship between the identity information and the vehicle feature information; The vehicle characteristic information, the mapping relationship and the identification information are stored in a database.
  • the vehicle feature information includes first sub-information and second sub-information;
  • the mapping relationship includes a first sub-relation and a second sub-relationship;
  • the first sub-information includes at least one of the following: an initial detection frame of the vehicle in the image of the target vehicle, a confidence degree of the initial detection frame, the image quality information, a vehicle attribute, a confidence degree of the vehicle attribute, The detection frame of the license plate in the target vehicle image, the confidence degree of the detection frame of the license plate, the license plate number on the license plate and the confidence degree of the license plate number; the shooting time when the image to be processed is captured by the photographing device; Device information of the shooting device;
  • the second sub-information includes at least one of the following: the feature vector of the vehicle in the target vehicle image and the feature vector of the license plate in the target vehicle image;
  • the feature storage module 606 is configured to establish a first sub-relationship between the identity information and the first sub-information; establish a second sub-relationship between the identity information and the second sub-information ; Store the first sub-information, the first sub-relationship and the identification information into an attribute database; store the second sub-information, the second sub-relationship and the identification information into a feature database .
  • the characteristic storage module 606 is configured to determine whether the confidence degree of the license plate number is greater than a preset confidence degree under the condition that the vehicle characteristic information includes the confidence degree of the license plate number ; when the confidence of the license plate number is greater than the preset reliability, storing the vehicle characteristic information, the mapping relationship and the identification information into a database.
  • the device information includes identification information
  • the feature storage module 606 is configured to, when the vehicle feature information includes the shooting time and/or the device information of the shooting device , if the shooting time conforms to the first preset format, then store the vehicle feature information, the mapping relationship, and the identification information into a library; or, if the identification information conforms to the second preset format, then storing the vehicle characteristic information, the mapping relationship and the identification information in a database.
  • the third determination module 605 is configured to receive the storage request sent by the client; the storage request includes the shooting time and/or the The device information of the shooting device; the storage request is analyzed to obtain the shooting time and/or the device information of the shooting device.
  • the device information includes at least one of the following:
  • the location information of the shooting device The location information of the shooting device; the identification information of the shooting device.
  • the image recognition module 601 is configured to receive the storage request sent by the client, wherein the storage request includes the image to be processed; The carried image to be processed; if the image to be processed meets a second preset condition, identify the image to be processed.
  • the third determination module 605 is configured to use an image obtained by expanding the target vehicle image on the image to be processed as a first sub-image; based on the first sub-image image, determining the detection frame of the license plate in the first sub-image; based on the detection frame of the license plate, intercepting the second sub-image from the first sub-image; based on the first sub-image and the vehicle feature detection model, Determine the vehicle feature sub-information of the first sub-image; determine the license plate feature sub-information of the second sub-image based on the second sub-image and the license plate feature detection model; based on the vehicle feature sub-information and the license plate
  • the feature sub-information is used to determine the vehicle feature information of the target vehicle image.
  • the object attribute includes at least one of the following: the color of the vehicle, the appearance information of the vehicle, and the type of the vehicle.
  • the embodiment of the present application also provides a computer device.
  • FIG 7 it is a schematic structural diagram of a computer device provided in the embodiment of the present application, including:
  • processor 71 memory 72 and bus 73 .
  • the memory 72 stores machine-readable instructions executable by the processor 71
  • the processor 71 is used to execute the machine-readable instructions stored in the memory 72.
  • the processor 71 executes follows the steps below:
  • S101 Responding to the storage request of the image to be processed, identify the image to be processed;
  • S102 In the case of identifying that the image to be processed contains a vehicle, determine at least one vehicle image; wherein, each vehicle image corresponds to a vehicle;
  • S104 Screen the vehicle images whose image quality information meets the first preset condition from the vehicle images as the target vehicle image;
  • S105 Determine vehicle characteristic information of the target vehicle image based on the target vehicle image and the image to be processed
  • S106 Store the vehicle feature information of the target vehicle image into a library.
  • memory 72 comprises memory 721 and external memory 722;
  • Memory 721 here is also called internal memory, is used for temporarily storing computing data in processor 71, and the data exchanged with external memory 722 such as hard disk, processor 71 communicates with memory 721 through memory 721.
  • the external memory 722 performs data exchange.
  • the processor 71 communicates with the memory 72 through the bus 73, so that the processor 71 executes the execution instructions mentioned in the above method embodiments.
  • Embodiments of the present disclosure further provide a computer-readable storage medium, on which a computer program is stored, and when the computer program is run by a processor, the steps of the data processing method described in the foregoing method embodiments are executed.
  • the storage medium may be a volatile or non-volatile computer-readable storage medium.
  • Embodiments of the present disclosure also provide a computer program product, the computer program product carries a program code, and the instructions included in the program code can be used to execute the steps of the data processing method described in the above method embodiment, for details, please refer to the above The method embodiment will not be repeated here.
  • the above-mentioned computer program product may be specifically implemented by means of hardware, software or a combination thereof.
  • the computer program product is embodied as a computer storage medium, and in another optional embodiment, the computer program product is embodied as a software product, such as a software development kit (Software Development Kit, SDK) etc. Wait.
  • the disclosed devices and methods may be implemented in other ways.
  • the device embodiments described above are only illustrative.
  • the division of the modules is only a logical function division.
  • multiple modules or components can be combined or Some features may be ignored, or not implemented.
  • the mutual coupling or direct coupling or communication connection shown or discussed may be through some communication interfaces, and the indirect coupling or communication connection of devices or modules may be in electrical, mechanical or other forms.
  • the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.
  • each functional module in each embodiment of the present disclosure may be integrated into one processing module, each module may exist separately physically, or two or more modules may be integrated into one module.
  • the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium executable by a processor.
  • the technical solution of the present disclosure is essentially or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including Several instructions are used to make a computer device (which may be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the methods described in various embodiments of the present disclosure.
  • the aforementioned storage media include: U disk, mobile hard disk, read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), magnetic disk or optical disc and other media that can store program codes. .

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Traffic Control Systems (AREA)

Abstract

本公开提供了一种数据处理方法、装置、计算机设备和存储介质，其中，该方法包括：响应于对待处理图像的入库请求，对待处理图像进行识别；在识别待处理图像中包含车辆的情况下，确定至少一个车辆图像；其中，每个车辆图像分别对应一个车辆；确定车辆图像的图像质量信息；从车辆图像中筛选图像质量信息达到第一预设条件的车辆图像，作为目标车辆图像；基于目标车辆图像和待处理图像，确定目标车辆图像的车辆特征信息；将目标车辆图像的车辆特征信息存储入库。

Description

一种数据处理方法、装置、计算机设备和存储介质
本申请要求2021年06月28日提交、申请号为202110718804.5，发明名称为“一种数据处理方法、装置、计算机设备和存储介质”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
本公开涉及计算机技术以及图像处理领域，具体而言，涉及一种数据处理方法、装置、计算机设备和存储介质。
背景技术
在智慧交通领域中，分析车辆特征信息对数字化交通场景的理解，以及交通车辆检索等有着重要作用。比如，在交通检测车辆逃费的场景中，需要保证得到的信息准确真实，且能完全指证当前车辆特征，降低错误的车辆特征信息导致车辆逃费检测失败或者误判等情况发生的概率。
发明内容
本公开实施例至少提供一种数据处理方法、装置、计算机设备和存储介质。
第一方面，本公开实施例提供了一种数据处理方法，包括：
响应于对待处理图像的入库请求，对所述待处理图像进行识别；
在识别所述待处理图像中包含车辆的情况下，确定至少一个车辆图像；其中，每个车辆图像分别对应一个车辆；
确定所述车辆图像的图像质量信息；
从所述车辆图像中筛选所述图像质量信息达到第一预设条件的车辆图像，作为目标车辆图像；
基于所述目标车辆图像和所述待处理图像，确定所述目标车辆图像的车辆特征信息；
将所述目标车辆图像的车辆特征信息存储入库。
该方面，利用第一预设条件和图像质量信息，来筛选目标车辆图像，能够提高提取的车辆特征信息的准确性，从而能够提高稽逃成功率；另外，只对筛选出的目标车辆图像进行特征提取，能够减少特征提取过程中的计算量，并且能够降低数据库存储无效信息的概率，节省数据库的存储空间。
一种可选的实施方式中，所述基于所述目标车辆图像和所述待处理图像，确定所述目标车辆图像的车辆特征信息，包括：
基于所述目标车辆图像在所述待处理图像上外扩处理得到的图像，确定所述目标车辆图像的车辆特征信息。
该实施方式，对目标车辆图像进行外扩处理，能够得到更为完整的目标车辆图像中车辆的图像，进而能够提取到更为全面的车辆特征信息。
一种可选的实施方式中，所述图像质量信息包括拍摄的所述车辆图像的质量分数和所述车辆图像中车辆的朝向；
从所述车辆图像中，筛选所述图像质量信息达到第一预设条件的车辆图像，作为目标车辆图像，包括：
从所述车辆图像中，筛选所述质量分数大于预设阈值、且所述车辆的朝向在预设朝向范围内的车辆图像，作为所述目标车辆图像。
该实施方式，通过检测车辆图像的质量分数以及车辆图像中车辆的朝向，能够精准地从车辆图像中筛选出拍摄质量较高的、能够较为全面的提取特征的目标车辆图像，从而能够提取到准确、全面的车辆特征信息。
一种可选的实施方式中，所述将所述目标车辆图像的车辆特征信息存储入库，包括：
确定所述目标车辆图像中车辆的身份标识信息；
建立所述身份标识信息和所述车辆特征信息之间的映射关系；
将所述车辆特征信息、所述映射关系以及所述身份标识信息存储入库。
该实施方式中，由于数据库中存储有身份标识信息和车辆特征信息之间的映射关系，因此，利用身份标识信息能够方便快捷的从数据库中找出与该身份标识信息对应的车辆特征信息，提高后续车辆特征信息查询的效率。
一种可选的实施方式中，所述车辆特征信息包括第一子信息和第二子信息；所述映射关系包括第一子关系和第二子关系；
所述第一子信息包括以下至少一项：所述目标车辆图像中车辆的初始检测框、所述初始检测框的置信度、所述图像质量信息、车辆属性、所述车辆属性的置信度、所述目标车辆图像中车牌的检测框、所述车牌的检测框的置信度、所述车牌上的车牌号和所述车牌号的置信度；拍摄设备拍摄到所述待处理图像的拍摄时间；所述拍摄设备的设备信息；
所述第二子信息包括以下至少一项：所述目标车辆图像中车辆的特征向量和所述目标车辆图像中车牌的特征向量；
所述建立所述身份标识信息和所述车辆特征信息之间的映射关系，包括：
建立所述身份标识信息和所述第一子信息之间的第一子关系；
建立所述身份标识信息和所述第二子信息之间的第二子关系；
所述将所述车辆特征信息、所述映射关系以及所述身份标识信息存储入库，包括：
将所述第一子信息、所述第一子关系和所述身份标识信息存入属性数据库；
将所述第二子信息、所述第二子关系和所述身份标识信息存入特征数据库。
该实施方式中，由于上述车辆特征信息能够全面、准确的表征目标车辆图像中的车辆，因此利用上述车辆特征信息有助于提高稽逃成功率。另外，由于第一子信息与第二子信息的数据类型不同，因此，将第一子信息与第二子信息分类存储，能够方便数据管理，同时还能够提高数据查询效率。另外，将第一子关系存入属性数据库中，方便利用身份标识信息从属性数据库中调取与该身份标识信息相匹配的第一子信息；将第二关系存入特征数据库中，方便利用身份标识信息从特征数据库中调取与该身份标识信息相匹配的第二子信息，能够提高后续车辆特征信息查询的效率。
一种可选的实施方式中，所述将所述车辆特征信息、所述映射关系以及所述身份标识信息存储入库，包括：
在所述车辆特征信息包括所述车牌号的置信度的情况下，判断所述车牌号的置信度是否大于预设置信度；
在所述车牌号的置信度大于所述预设置信度的情况下，将所述车辆特征信息、所述映射关系以及所述身份标识信息存储入库。
该实施方式中，在存储车辆特征信息时，只有在车牌号的置信度大于预设置信度的情况下，才会对车牌号进行存储，能够降低存储无效数据的概率，节省存储空间。
一种可选的实施方式中，所述设备信息包括标识信息；
所述将所述车辆特征信息、所述映射关系以及所述身份标识信息存储入库，包括：
在所述车辆特征信息包括所述拍摄时间和/或所述拍摄设备的设备信息的情况下，若所述拍摄时间符合第一预设格式，则将所述车辆特征信息、所述映射关系以及所述身份标识信息存储入库；
或者，若所述标识信息符合第二预设格式，则将所述车辆特征信息、所述映射关系以及所述身份标识信息存储入库。
该实施方式中，在存储车辆特征信息之前，通过对拍摄时间和标识信息进行合理化检测，在拍摄时间符合第一预设格式和/或标识信息符合第二预设格式的情况下，才会将车辆特征信息、映射关系以及身份标识信息存储入库，能够确保存储的车辆特征信息为预设的合法的数据，降低存储无效数据的概率，节省存储空间。
一种可选的实施方式中，所述确定所述目标车辆图像的车辆特征信息，包括：
接收客户端发送的所述入库请求；所述入库请求包括所述待处理图像的所述拍摄时间和/或所述拍摄设备的设备信息；
对所述入库请求进行解析，得到所述拍摄时间和/或所述拍摄设备的设备信息。
一种可选的实施方式中，所述设备信息包括以下至少一项：
所述拍摄设备的位置信息；所述拍摄设备的标识信息。
一种可选的实施方式中，所述响应于对待处理图像的入库请求，对所述待处理图像进行识别，包括：
接收客户端发送的所述入库请求，其中，所述入库请求包括所述待处理图像；
获取所述入库请求中携带的所述待处理图像；
在所述待处理图像符合第二预设条件的情况下，对所述待处理图像进行识别。
该实施方式中，通过对获取到的待处理图像进行合理化检测，能够确保识别的待处理图像为预设的需要进行特征提取的图像。
一种可选的实施方式中，所述基于所述目标车辆图像在所述待处理图像上外扩处理得到的图像，确定所述目标车辆图像的车辆特征信息，包括：
将所述目标车辆图像在所述待处理图像上外扩处理得到的图像作为第一子图像；
基于所述第一子图像，确定所述第一子图像中车牌的检测框；
基于所述车牌的检测框，从所述第一子图像中截取第二子图像；
基于所述第一子图像和车辆特征检测模型，确定所述第一子图像的车辆特征子信息；
基于所述第二子图像和车牌特征检测模型，确定所述第二子图像的车牌特征子信息；
基于所述车辆特征子信息和所述车牌特征子信息，确定所述目标车辆图像的车辆特征信息。
该实施方式中，通过第一子图像和第二子图像，能够更为细致化的提取到目标车辆图像的车辆特征信息，进而得到较为全面的车辆特征信息。
一种可选的实施方式中，所述对象属性包括以下至少一项：车辆的颜色、车辆的外观信息和车辆的类型。
第二方面，本公开实施例还提供一种数据处理装置，包括：
图像识别模块，用于响应于对待处理图像的入库请求，对所述待处理图像进行识别；
第一确定模块，用于在识别所述待处理图像中包含车辆的情况下，确定至少一个车辆图像；其中，每个车辆图像分别对应一个车辆；
第二确定模块，用于确定所述车辆图像的图像质量信息；
图像筛选模块，用于从所述车辆图像中筛选所述图像质量信息达到第一预设条件的车辆图像，作为目标车辆图像；
第三确定模块，用于基于所述目标车辆图像和所述待处理图像，确定所述目标车辆图像的车辆特征信息；
特征存储模块，用于将所述目标车辆图像的车辆特征信息存储入库。
一种可选的实施方式中，所述第三确定模块，用于基于所述目标车辆图像在所述待处理图像上外扩处理得到的图像，确定所述目标车辆图像的车辆特征信息。
一种可选的实施方式中，所述图像质量信息包括拍摄的所述车辆图像的质量分数和所述车辆图像中车辆的朝向；
所述图像筛选模块，用于从所述车辆图像中，筛选所述质量分数大于预设阈值、且所述车辆的朝向在预设朝向范围内的车辆图像，作为所述目标车辆图像。
一种可选的实施方式中，所述特征存储模块，用于确定所述目标车辆图像中车辆的身份标识信息；建立所述身份标识信息和所述车辆特征信息之间的映射关系；将所述车辆特征信息、所述映射关系以及所述身份标识信息存储入库。
一种可选的实施方式中，所述车辆特征信息包括第一子信息和第二子信息；所述映射关系包括第一子关系和第二子关系；
所述第一子信息包括以下至少一项：所述目标车辆图像中车辆的初始检测框、所述初始检测框的置信度、所述图像质量信息、车辆属性、所述车辆属性的置信度、所述目标车辆图像中车牌的检测框、所述车牌的检测框的置信度、所述车牌上的车牌号和所述车牌号的置信度；拍摄设备拍摄到所述待处理图像的拍摄时间；所述拍摄设备的设备信息；
所述第二子信息包括以下至少一项：所述目标车辆图像中车辆的特征向量和所述目标车辆图像中车牌的特征向量；
所述特征存储模块，用于建立所述身份标识信息和所述第一子信息之间的第一子关系；建立所述身份标识信息和所述第二子信息之间的第二子关系；将所述第一子信息、所述第一子关系和所述身份标识信息存入属性数据库；将所述第二子信息、所述第二子关系和所述身份标识信息存入特征数据库。
一种可选的实施方式中，所述特征存储模块，用于在所述车辆特征信息包括所述车牌号的置信度的情况下，判断所述车牌号的置信度是否大于预设置信度；在所述车牌号的置信度大于所述预设置信度的情况下，将所述车辆特征信息、所述映射关系以及所述身份标识信息存储入库。
一种可选的实施方式中，所述设备信息包括标识信息；所述特征存储模块，用于在所述车辆特征信息包括所述拍摄时间和/或所述拍摄设备的设备信息的情况下，若所述拍摄时间符合第一预设格式，则将所述车辆特征信息、所述映射关系以及所述身份标识信息存储入库；或者，若所述标识信息符合第二预设格式，则将所述车辆特征信息、所述映射关系以及所述身份标识信息存储入库。
一种可选的实施方式中，所述第三确定模块，用于接收客户端发送的所述入库请求；所述入库请求包括所述待处理图像的所述拍摄时间和/或所述拍摄设备的设备信息；对所述入库请求进行解析，得到所述拍摄时间和/或所述拍摄设备的设备信息。
一种可选的实施方式中，所述设备信息包括以下至少一项：
所述拍摄设备的位置信息；所述拍摄设备的标识信息。
一种可选的实施方式中，所述图像识别模块，用于接收客户端发送的所述入库请求，其中，所述入库请求包括所述待处理图像；获取所述入库请求中携带的所述待处理图像；在所述待处理图像符合第二预设条件的情况下，对所述待处理图像进行识别。
一种可选的实施方式中，所述第三确定模块，用于将所述目标车辆图像在所述待处理图像上外扩处理得到的图像作为第一子图像；基于所述第一子图像，确定所述第一子图像中车牌的检测框；基于所述车牌的检测框，从所述第一子图像中截取第二子图像；基于所述第一子图像和车辆特征检测模型，确定所述第一子图像的车辆特征子信息；基于所述第二子图像和车牌特征检测模型，确定所述第二子图像的车牌特征子信息；基于所述车辆特征子信息和所述车牌特征子信息，确定所述目标车辆图像的车辆特征信息。
一种可选的实施方式中，所述对象属性包括以下至少一项：车辆的颜色、车辆的外观信息和车辆的类型。
第三方面，本公开实施例还提供一种计算机设备，包括：处理器、存储器和总线，所述存储器存储有所述处理器可执行的机器可读指令，当计算机设备运行时，所述处理器与所述存储器之间通过总线通信，所述机器可读指令被所述处理器执行时执行上述第一方面，或第一方面中任一种可能的实施方式中的步骤。
第四方面，本公开实施例还提供一种计算机可读存储介质，该计算机可读存储介质上存储有计算机程序，该计算机程序被处理器运行时执行上述第一方面，或第一方面中任一种可能的实施方式中的步骤。
第五方面，本公开实施例还提供一种计算机程序产品，包括计算机可读代码，或者承载有计算 机可读代码的非易失性计算机可读存储介质，当所述计算机可读代码在电子设备的处理器中运行时，所述电子设备中的处理器执行用于实现上述第一方面，或第一方面中任一种可能的实施方式中的步骤。
关于上述数据处理装置、计算机设备、存储介质和计算机程序产品的效果描述参见上述数据处理方法的说明，这里不再赘述。
为使本公开的上述目的、特征和优点能更明显易懂，下文特举可选实施例，并配合所附附图，作详细说明如下。
附图说明
为了更清楚地说明本公开实施例的技术方案，下面将对实施例中所需要使用的附图作简单地介绍，此处的附图被并入说明书中并构成本说明书中的一部分，这些附图示出了符合本公开的实施例，并与说明书一起用于说明本公开的技术方案。应当理解，以下附图仅示出了本公开的某些实施例，因此不应被看作是对范围的限定，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他相关的附图。
图1示出了本公开实施例所提供的一种数据处理方法的流程图；
图2示出了本公开实施例所提供的确定目标车辆图像的车辆特征信息的流程图；
图3示出了本公开实施例所提供的车辆特征信息存储入库的流程图；
图4示出了本公开实施例所提供的一种应用场景下存储车辆特征信息的流程图；
图5示出了本公开实施例所提供的车辆的结构化信息对应存储入库的示意图；
图6示出了本公开实施例所提供的一种数据处理装置的示意图；
图7示出了本公开实施例所提供的一种计算机设备的结构示意图。
具体实施方式
为使本公开实施例的目的、技术方案和优点更加清楚，下面将结合本公开实施例中附图，对本公开实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本公开一部分实施例，而不是全部的实施例。通常在此处附图中描述和示出的本公开实施例的组件可以以一种或多种不同的配置来布置和设计。因此，以下对在附图中提供的本公开的实施例的详细描述并非旨在限制要求保护的本公开的范围，而是仅仅表示本公开的选定实施例。基于本公开的实施例，本领域技术人员在没有做出创造性劳动的前提下所获得的所有其他实施例，都属于本公开保护的范围。
另外，本公开实施例中的说明书和权利要求书及上述附图中的术语“第一”、“第二”等是用于区别类似的对象，而不必用于描述特定的顺序或先后次序。应该理解这样使用的数据在适当情况下可以互换，以便这里描述的实施例能够以除了在这里图示或描述的内容以外的顺序实施。
在本文中提及的“多个或者若干个”是指两个或两个以上。“和/或”，描述关联对象的关联关系，表示可以存在三种关系，例如，A和/或B，可以表示：单独存在A，同时存在A和B，单独存在B这三种情况。字符“/”一般表示前后关联对象是一种“或”的关系。
针对以上方案所存在的缺陷，均是发明人在经过实践并仔细研究后得出的结果，因此，上述问题的发现过程以及下文中本公开针对上述问题所提出的解决方案，都应该是发明人在本公开过程中 对本公开做出的贡献。
应注意到：相似的标号和字母在下面的附图中表示类似项，因此，一旦某一项在一个附图中被定义，则在随后的附图中不需要对其进行进一步定义和解释。
为便于对本实施例进行理解，首先对本公开实施例所公开的一种数据处理方法进行详细介绍，本公开实施例所提供的数据处理方法的执行主体一般为具有一定计算能力的计算机设备。在一些可能的实现方式中，该数据处理方法可以通过处理器调用存储器中存储的计算机可读指令的方式来实现。
下面以执行主体为计算机设备为例对本公开实施例提供的数据处理方法加以说明。
参见图1所示，为本公开实施例提供的一种数据处理方法的流程图，所述方法包括步骤S101～S106，其中：
S101：响应于对待处理图像的入库请求，对待处理图像进行识别。
本步骤中，入库请求可以为用户发起的http(hypertext transfer protocol，超文本传输协议)请求，http请求中包括待处理图像。具体的，响应该http请求，对该http请求中所携带的待处理图像进行识别。
示例性的，可以利用训练好的神经网络模型对待处理图像进行识别，识别待处理图像中是否存在车辆。其中，神经网络可以包括下述至少一种：卷积神经网络(Convolutional Neural Networks，CNN)、区域卷积神经网络(Regions Region-based Convolutional Network，R-CNN)、快速区域卷积神经网络(Fast Region-based Convolutional Network，Fast R-CNN)、更快速区域卷积神经网络(Faster Region-based Convolutional Network，Faster R-CNN)等。
在一些实施例中，入库请求还包括待处理图像的拍摄时间和/或拍摄设备的设备信息。具体的，接收用户的客户端发送的入库请求；对入库请求进行解析，得到拍摄时间和/或拍摄设备的设备信息。其中，设备信息可以包括以下至少一项：拍摄设备的位置信息；拍摄设备的标识信息。示例性的，拍摄设备的标识信息还可以用于指示拍摄设备的位置。
S102：在识别待处理图像中包含车辆的情况下，确定至少一个车辆图像。
示例性的，可以通过检测车辆检测框的卷积神经网络模型对待处理图像进行结构化检测，能够判断出该待处理图像中是否存在车辆，在存在车辆的情况下，利用卷积神经网络模型输出车辆的初始检测框和初始检测框的置信度。之后，在待处理图像中将标注有车辆的初始检测框的区域的图像作为车辆图像。由于一张待处理图像中可能包含有多个车辆，因此，可以识别出多个车辆的初始检测框，进而能够得到多个车辆图像。
另外，待处理图像还可以不存在车辆，则循环执行步骤S101，等待下一次的入库请求。
S103：确定车辆图像的图像质量信息。
本步骤中，图像质量信息可以为指示车辆图像的画面质量和/或车辆图像中车辆的朝向的指标。
其中，画面质量可以包括车辆图像的清晰度、锐度、色散度、解析度、色域范围、色彩纯度、色彩平衡等几个方面。
具体地，可以利用检测图像质量信息的神经网络模型输出图像质量信息。图像质量信息可以包括评价车辆图像的画面质量的质量分数和评价车辆图像中车辆所处方位合理程度的朝向信息。这里， 车辆图像的画面质量越好，质量分数越高。
示例性的，车辆在运动过程中，由于受速度、环境、光照等因素的影响，可能会导致被抓拍到的高速运动车辆呈现模糊状态；或者，还可能抓拍到的车辆朝向为侧面(不能看见车头和车尾的车辆朝向)，因此需要对车辆图像进行质量检测，确定车辆图像的质量信息，以为后续确定更为准确的车辆特征信息。
S104：从车辆图像中筛选图像质量信息达到第一预设条件的车辆图像，作为目标车辆图像。
本步骤中，针对图像质量信息所指示的画面质量的质量分数，第一预设条件可以包括质量分数大于预设阈值的条件；针对图像质量信息所指示的车辆所处位置合理程度的朝向信息，第一预设条件还可以包括车辆的朝向的预设朝向范围，比如能看见车头或车尾的车辆的朝向。
基于上述步骤S102中的质量检测，可以将不符合第一预设条件的车辆图像进行过滤，只保留符合第一预设条件的车辆图像，并将符合第一预设条件的车辆图像作为目标车辆图像，针对目标车辆图像进行后续车辆特征信息的提取和存储，能够得到更为准确的车辆特征信息。
在一个实施例中，可以从车辆图像中，筛选质量分数大于预设阈值的车辆图像，作为目标车辆图像；或者，可以从车辆图像中，筛选质量分数大于预设阈值、且车辆的朝向在预设朝向范围内的车辆图像，作为目标车辆图像。这里，通过检测车辆图像的质量分数和/或车辆图像中车辆的朝向，能够精准地从车辆图像中筛选出拍摄质量较高的、能够较为全面的提取特征的目标车辆图像，从而能够提取到准确、全面的车辆特征信息。
延续上例，对于上述被抓拍到的高速运动车辆呈现模糊状态，可以确定该车辆图像的质量分数为小于或等于预设阈值，因此该车辆图像不符合继续检测的标准，针对该车辆图像将不再进行后续流程的处理；对于被抓拍到的车辆的朝向为侧面的情况，可以确定该车辆的朝向不在预设朝向范围内，因此该车辆不符合继续检测的标准，针对该车辆所在的车辆图像将不再进行后续流程的处理，能够为后续流程减轻一部分无效数据特征检测的压力，进而能够降低数据库存储无效信息的概率。
S105：基于目标车辆图像和待处理图像，确定目标车辆图像的车辆特征信息。
为了能够提取到较为全面的车辆特征信息，降低目标车辆图像中的车辆信息不全的概率，因此，可以基于待处理图像，外扩目标车辆图像，得到外扩后完整的目标车辆图像中的车辆的图像。具体实施时，基于目标车辆图像在待处理图像上外扩处理得到的图像，确定目标车辆图像的车辆特征信息。
示例性的，外扩处理可以为以目标车辆图像中的初始检测框的几何中心为中心，初始检测框对应的长边和宽边分别扩大1.2倍，确定目标检测框，该目标检测框框出的图像，即为外扩处理得到的图像，能够包括完整的车辆。
这里，由于车辆特征信息中包括不同数据类型的特征信息，因此，需要利用不同特征提取类型的神经网络模型进行特征提取，示例性的，可以利用车辆特征检测模型提取车辆特征子信息，可以利用车牌特征检测模型提取车牌特征子信息。
可以参见图2所示，其为确定目标车辆图像的车辆特征信息的流程图，其中，包括步骤S201～S206：
S201：将目标车辆图像在待处理图像上外扩处理得到的图像作为第一子图像。
具体的，基于上述目标车辆图像外扩处理的具体描述，可以将目标检测框框出的图像作为第一子图像。示例性的，可以利用目标检测框所指示的车辆位置信息，从待处理图像中截取第一子图像。
S202：基于第一子图像，确定第一子图像中车牌的检测框。
具体的，可以利用检测车牌检测框的卷积神经网络模型对第一子图像进行结构化检测，输出车牌的检测框。
S203：基于车牌的检测框，从第一子图像中截取第二子图像。
具体的，可以将车牌的检测框框出的图像作为第二子图像。示例性的，可以利用车牌的检测框所指示的车牌位置信息，从第一子图像中截取第二子图像。
S204：基于第一子图像和车辆特征检测模型，确定第一子图像的车辆特征子信息。
本步骤中，车辆子特征信息可以包括车辆属性、车辆属性的置信度和车辆的特征向量。其中，车辆属性包括以下至少一项：车辆的颜色、车辆的外观信息和车辆的类型。
示例性的，车辆特征检测模型可以包括检测车辆属性的神经网络模型和检测车辆的特征向量的神经网络模型。可以利用检测车辆属性的神经网络模型，输出第一子图像中车辆的颜色和车辆的颜色的置信度、车辆的外观信息和车辆的外观信息的置信度、车辆的类型和车辆的类型的置信度；可以利用检测车辆的特征向量的神经网络模型，输出第一子图像中车辆的特征向量，这里，车辆的特征向量可以表示车辆的特征。
S205：基于第二子图像和车牌特征检测模型，确定第二子图像的车牌特征子信息。
本步骤中，车牌特征子信息可以包括车牌的检测框、车牌的检测框的置信度、车牌上的车牌号、车牌号的置信度和车牌的特征向量。
示例性的，车牌特征检测模型包括检测车牌的检测框的神经网络模型、检测车牌号的神经网络模型和检测车牌的特征向量的神经网络模型。可以利用检测车牌的检测框的神经网络模型，输出第二子图像中车牌的检测框和车牌的检测框的置信度；可以利用检测车牌号的神经网络模型，输出第二子图像中车牌号和车牌号的置信度；可以利用检测车牌的特征向量的神经网络模型，输出第二子图像中车牌的特征向量，这里，车牌的特征向量可以表示车牌的特征。
S206：基于车辆特征子信息和车牌特征子信息，确定目标车辆图像的车辆特征信息。
这里，车辆特征信息包括：车辆属性、车辆属性的置信度和车辆的特征向量；车牌的检测框、车牌的检测框的置信度、车牌上的车牌号、车牌号的置信度和车牌的特征向量。
基于上述S101～S105，车辆特征信息还包括：车辆的初始检测框、初始检测框的置信度、图像质量信息、拍摄设备拍摄到待处理图像的拍摄时间、拍摄设备的设备信息。
S106：将目标车辆图像的车辆特征信息存储入库。
本步骤中，由于车辆特征信息中包括不同数据类型的信息，因此，需要将不同数据类型的信息分类存储。
可以将确定出的车辆特征信息分类处理，即第一子信息和第二子信息，将不同类别的车辆特征信息对应存储在不同的数据库中。比如，将第一子信息存入属性数据库中，将第二子信息存入特征数据库中。
第一子信息可以包括以下至少一项：目标车辆图像中车辆的初始检测框、初始检测框的置信度、图像质量信息、车辆属性、车辆属性的置信度、目标车辆图像中车牌的检测框、车牌的检测框的置信度、车牌上的车牌号和车牌号的置信度；拍摄设备拍摄到待处理图像的拍摄时间；拍摄设备的设备信息。
第二子信息可以包括以下至少一项：目标车辆图像中车辆的特征向量和目标车辆图像中车牌的特征向量。
上述第一子信息和第二子信息能够全面、准确的表征目标车辆图像中的车辆，因此利用该第一子信息和第二子信息所组成的车辆特征信息有助于提高稽逃成功率。
由于第一子信息与第二子信息的数据类型不同，因此，需要将第一子信息与第二子信息分类存储，可以参见图3所示，其为车辆特征信息存储入库的流程图，其中，包括步骤S301～S303：
S301：确定目标车辆图像中车辆的身份标识信息。
这里，确定目标车辆图像中车辆的身份标识信息方便利用身份标识信息从数据库中查找该身份标识信息表示的车辆。身份标识信息可以为车辆的车牌号或者为预设的编号等。
S302：建立身份标识信息和车辆特征信息之间的映射关系。
具体的，可以建立身份标识信息和第一子信息之间的第一子关系；建立身份标识信息和第二子信息之间的第二子关系。
S303：将车辆特征信息、映射关系以及身份标识信息存储入库。
示例性的，可以将目标车辆图像中的车辆标记一个预设的编号A，该目标车辆图像的车辆特征信息包括的全部信息用编号A来表示，这样，从数据库中调取车辆A的车辆特征信息的时候，只需要输入该编号A就可以查询出有关该车辆A的全部车辆特征信息。
具体实施时，可以将第一子信息、第一子关系和身份标识信息存入属性数据库；将第二子信息、第二子关系和身份标识信息存入特征数据库。
示例性的，属性数据库可以包括结构体数据库StructDB；特征数据库可以包括静态数据库(engine-static-feature-db，sfd)。
上述，将第一子关系存入属性数据库中，方便利用身份标识信息从属性数据库中调取与该身份标识信息相匹配的第一子信息；将第二关系存入特征数据库中，方便利用身份标识信息从特征数据库中调取与该身份标识信息相匹配的第二子信息，能够提高后续车辆特征信息查询的效率。
在一些实施例中，在存储车辆特征信息之前，还需要判断存储的车辆特征信息是否符合存储要求。
第一种判断方式、在车辆特征信息包括车牌号的置信度的情况下，判断车牌号的置信度是否大于预设置信度；在车牌号的置信度大于预设置信度的情况下，将车辆特征信息、映射关系以及身份标识信息存储入库。其中，预设置信度可以根据不同应用场景进行具体设置，在此不进行限定。
第二种判断方式、在车辆特征信息包括拍摄时间和/或拍摄设备的设备信息的情况下，若拍摄时间符合第一预设格式，则将车辆特征信息、映射关系以及所述身份标识信息存储入库；或者，若标识信息符合第二预设格式，则将车辆特征信息、映射关系以及身份标识信息存储入库。
示例性的，第一预设格式可以为XXXX(年)-XX(月)-XX(日)。这里，第一预设格式可以包括若干种表示时间的不同格式，在此不进行限定。由于不同拍摄设备的标识信息不同，所以在此不具体限定标识信息的第二预设格式。
在一些实施例中，在车辆特征信息包括拍摄时间和/或拍摄设备的设备信息的情况下，若拍摄时间不符合第一预设格式，则不将拍摄时间存出入库，将除拍摄时间之外的车辆特征信息存储入库。若标识信息不符合第二预设格式的情况下，则不将标识信息存储入库，将除标识信息之外的车辆特征信息存储入库。
示例性的，在车辆特征信息准备存入数据库的情况下，可以将符合第一预设格式的拍摄时间与其他车辆特征信息一同存入属性数据库。如果出现拍摄时间为XXXXXX-XXXXX等不符合第一预设格式的情况，将不允许该拍摄时间存入属性数据库，可以将除拍摄时间之外的车辆特征信息入库。
在确定车辆特征信息符合存储要求的情况下，将步骤S102中识别出的车辆的初始检测框和初始检测框的置信度作为第一子信息，存储在属性数据库中。
在确定车辆特征信息符合存储要求的情况下，将从步骤S103中确定的图像质量信息作为第一子信息，存储在属性数据库中。
在一些实施例中，在上述步骤S101所述的对待处理图像进行识别之前，还可以对待处理图像进行校验处理，如果待处理图像符合第二预设条件，则可以对该待处理图像进行识别，如果待处理图像不符合第二预设条件，则不需要对该待处理图像进行识别，等待下一入库请求。
示例性的，可以通过http请求获取抓拍车辆的待处理图像。首先对该待处理图像进行校验处理，在待处理图像符合第二预设条件的情况下，可以对该待处理图像进行识别。这里，第二预设条件可以包括待处理图像存在图像信息，如果待处理图像乱码或无图像，则该待处理图像不符合第二预设条件。或者，第二预设条件还可以根据具体任务进行设定，在此不进行限定。
示例性的，针对车辆检测场景，参见图4所示，其为一种应用场景下存储车辆特征信息的流程图，其中，包括步骤S401～S408：
S401：获取http请求内容。
本步骤中，获取客户发送的http格式的入库请求。其中，http请求内容中可以包括抓拍车辆的待处理图像、拍摄设备拍摄待处理图像的拍摄时间和拍摄设备的设备信息。其中，设备信息包括标识信息。
S402：针对http请求内容进行校验处理。
本步骤中，针对待处理图像进行校验处理，在待处理图像符合第二预设条件的情况下，继续执行步骤S403；否则，结束流程。
其中，针对拍摄设备拍摄到待处理图像的拍摄时间，在拍摄时间符合第一预设格式的情况下，拍摄时间校验通过，等待存储；在拍摄设备的标识信息符合第二预设格式的情况下，拍摄设备的标识信息校验通过，等待存储。
S403：对待处理图像进行解码。
比如，可以对待处理图像进行base64解码。
S404：对待处理图像中的车辆进行结构化检测。
本步骤中，可以通过对车辆进行结构化检测，判断出该待处理图像中是否存在车辆。结构化检测可以是将待处理图像输入卷积神经网络模型中进行处理，以判断该待处理图像中是否存在车辆。
S405：判断待处理图像中是否存在车辆，如果存在，则执行步骤S306；如果不存在，则结束流程。
本步骤中，基于步骤S404中卷积神经网络模型输出的结果，即输出的车辆的初始检测框和初始检测框的置信度，如果存在车辆的初始检测框，且该初始检测框的置信度大于预设值，则确定待处理图像中存在车辆。
S46：判断图像质量信息是否达到第一预设条件，如果达到，则执行步骤S407；如果没达到，则结束流程。
本步骤中，针对图像质量信息中的质量分数，如果该图像质量分数大于预设阈值，且车辆图像中车辆的朝向在预设范围内，则确定目标车辆图像，并执行步骤S407；如果质量分数小于或等于预设阈值，或者车辆图像中车辆的朝向不在预设范围内，则结束流程。
S407：外扩目标车辆图像。
本步骤中，可以参照上述对目标车辆图像在待处理图像上外扩处理的过程，重复之处不再赘述。
S408：提取并存储外扩后的目标车辆图像中车辆的结构化信息。
本步骤中，车辆的结构化信息可以包括以下至少一项：车辆的初始检测框、车辆的初始检测框的置信度、图像质量信息、车辆属性、车辆属性的置信度、车牌的检测框、车牌的检测框的置信度、车牌号、车牌号的置信度、车辆的特征向量、车牌的特征向量、校验通过的拍摄时间和校验通过拍摄设备的标识信息。
这里，车辆的结构化信息的存储可以参照上述步骤S106中车辆特征信息的存储，重复部分在此不再赘述。具体的，可以参见图5所示的车辆的结构化信息对应存储入库的示意图。
通过上述步骤S101～S106，利用第一预设条件和图像质量信息，来筛选目标车辆图像，能够提高提取的车辆特征信息的准确性，从而能够提高车辆逃费检测的成功率；另外，只对筛选出的目标车辆图像进行特征提取，能够减少特征提取过程中的计算量，并且能够降低数据库存储无效信息的概率，节省数据库的存储空间。
本领域技术人员可以理解，在具体实施方式的上述方法中，各步骤的撰写顺序并不意味着严格的执行顺序而对实施过程构成任何限定，各步骤的具体执行顺序应当以其功能和可能的内在逻辑确定。
基于同一发明构思，本公开实施例中还提供了与数据处理方法对应的数据处理装置，由于本公开实施例中的装置解决问题的原理与本公开实施例上述数据处理方法相似，因此装置的实施可以参见方法的实施，重复之处不再赘述。
参照图6所示，为本公开实施例提供的一种数据处理装置的示意图，所述装置包括：图像识别模块601、第一确定模块602、第二确定模块603、图像筛选模块604、第三确定模块605和特征存储模块606；其中，
图像识别模块601，用于响应于对待处理图像的入库请求，对所述待处理图像进行识别；
第一确定模块602，用于在识别所述待处理图像中包含车辆的情况下，确定至少一个车辆图像； 其中，每个车辆图像分别对应一个车辆；
第二确定模块603，用于确定所述车辆图像的图像质量信息；
图像筛选模块604，用于从所述车辆图像中筛选所述图像质量信息达到第一预设条件的车辆图像，作为目标车辆图像；
第三确定模块605，用于基于所述目标车辆图像和所述待处理图像，确定所述目标车辆图像的车辆特征信息；
特征存储模块606，用于将所述目标车辆图像的车辆特征信息存储入库。
一种可选的实施方式中，所述第三确定模块605，用于基于所述目标车辆图像在所述待处理图像上外扩处理得到的图像，确定所述目标车辆图像的车辆特征信息。
一种可选的实施方式中，所述图像质量信息包括拍摄的所述车辆图像的质量分数和所述车辆图像中车辆的朝向；
所述图像筛选模块604，用于从所述车辆图像中，筛选所述质量分数大于预设阈值、且所述车辆的朝向在预设朝向范围内的车辆图像，作为所述目标车辆图像。
一种可选的实施方式中，所述特征存储模块606，用于确定所述目标车辆图像中车辆的身份标识信息；建立所述身份标识信息和所述车辆特征信息之间的映射关系；将所述车辆特征信息、所述映射关系以及所述身份标识信息存储入库。
一种可选的实施方式中，所述车辆特征信息包括第一子信息和第二子信息；所述映射关系包括第一子关系和第二子关系；
所述第一子信息包括以下至少一项：所述目标车辆图像中车辆的初始检测框、所述初始检测框的置信度、所述图像质量信息、车辆属性、所述车辆属性的置信度、所述目标车辆图像中车牌的检测框、所述车牌的检测框的置信度、所述车牌上的车牌号和所述车牌号的置信度；拍摄设备拍摄到所述待处理图像的拍摄时间；所述拍摄设备的设备信息；
所述第二子信息包括以下至少一项：所述目标车辆图像中车辆的特征向量和所述目标车辆图像中车牌的特征向量；
所述特征存储模块606，用于建立所述身份标识信息和所述第一子信息之间的第一子关系；建立所述身份标识信息和所述第二子信息之间的第二子关系；将所述第一子信息、所述第一子关系和所述身份标识信息存入属性数据库；将所述第二子信息、所述第二子关系和所述身份标识信息存入特征数据库。
一种可选的实施方式中，所述特征存储模块606，用于在所述车辆特征信息包括所述车牌号的置信度的情况下，判断所述车牌号的置信度是否大于预设置信度；在所述车牌号的置信度大于所述预设置信度的情况下，将所述车辆特征信息、所述映射关系以及所述身份标识信息存储入库。
一种可选的实施方式中，所述设备信息包括标识信息；所述特征存储模块606，用于在所述车辆特征信息包括所述拍摄时间和/或所述拍摄设备的设备信息的情况下，若所述拍摄时间符合第一预设格式，则将所述车辆特征信息、所述映射关系以及所述身份标识信息存储入库；或者，若所述标识信息符合第二预设格式，则将所述车辆特征信息、所述映射关系以及所述身份标识信息存储入库。
一种可选的实施方式中，所述第三确定模块605，用于接收客户端发送的所述入库请求；所述 入库请求包括所述待处理图像的所述拍摄时间和/或所述拍摄设备的设备信息；对所述入库请求进行解析，得到所述拍摄时间和/或所述拍摄设备的设备信息。
一种可选的实施方式中，所述设备信息包括以下至少一项：
所述拍摄设备的位置信息；所述拍摄设备的标识信息。
一种可选的实施方式中，所述图像识别模块601，用于接收客户端发送的所述入库请求，其中，所述入库请求包括所述待处理图像；获取所述入库请求中携带的所述待处理图像；在所述待处理图像符合第二预设条件的情况下，对所述待处理图像进行识别。
一种可选的实施方式中，所述第三确定模块605，用于将所述目标车辆图像在所述待处理图像上外扩处理得到的图像作为第一子图像；基于所述第一子图像，确定所述第一子图像中车牌的检测框；基于所述车牌的检测框，从所述第一子图像中截取第二子图像；基于所述第一子图像和车辆特征检测模型，确定所述第一子图像的车辆特征子信息；基于所述第二子图像和车牌特征检测模型，确定所述第二子图像的车牌特征子信息；基于所述车辆特征子信息和所述车牌特征子信息，确定所述目标车辆图像的车辆特征信息。
一种可选的实施方式中，所述对象属性包括以下至少一项：车辆的颜色、车辆的外观信息和车辆的类型。
关于装置中的各模块的处理流程、以及各模块之间的交互流程的描述可以参照上述数据处理方法实施例中的相关说明，这里不再详述。
基于同一技术构思，本申请实施例还提供了一种计算机设备。参照图7所示，为本申请实施例提供的计算机设备的结构示意图，包括：
处理器71、存储器72和总线73。其中，存储器72存储有处理器71可执行的机器可读指令，处理器71用于执行存储器72中存储的机器可读指令，所述机器可读指令被处理器71执行时，处理器71执行下述步骤：
S101：响应于对待处理图像的入库请求，对待处理图像进行识别；
S102：在识别待处理图像中包含车辆的情况下，确定至少一个车辆图像；其中，每个车辆图像分别对应一个车辆；
S103：确定车辆图像的图像质量信息；
S104：从车辆图像中筛选图像质量信息达到第一预设条件的车辆图像，作为目标车辆图像；
S105：基于目标车辆图像和待处理图像，确定目标车辆图像的车辆特征信息；
S106：将目标车辆图像的车辆特征信息存储入库。
上述存储器72包括内存721和外部存储器722；这里的内存721也称内存储器，用于暂时存放处理器71中的运算数据，以及与硬盘等外部存储器722交换的数据，处理器71通过内存721与外部存储器722进行数据交换，当计算机设备运行时，处理器71与存储器72之间通过总线73通信，使得处理器71在执行上述方法实施例中所提及的执行指令。
本公开实施例还提供一种计算机可读存储介质，该计算机可读存储介质上存储有计算机程序，该计算机程序被处理器运行时执行上述方法实施例中所述的数据处理方法的步骤。其中，该存储介 质可以是易失性或非易失的计算机可读取存储介质。
本公开实施例所还提供一种计算机程序产品，该计算机程序产品承载有程序代码，所述程序代码包括的指令可用于执行上述方法实施例中所述的数据处理方法的步骤，具体可参见上述方法实施例，在此不再赘述。
其中，上述计算机程序产品可以具体通过硬件、软件或其结合的方式实现。在一个可选实施例中，所述计算机程序产品具体体现为计算机存储介质，在另一个可选实施例中，计算机程序产品具体体现为软件产品，例如软件开发包(Software Development Kit，SDK)等等。
所属领域的技术人员可以清楚地了解到，为描述的方便和简洁，上述描述的装置的具体工作过程，可以参考前述方法实施例中的对应过程，在此不再赘述。在本公开所提供的几个实施例中，应该理解到，所揭露的装置和方法，可以通过其它的方式实现。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，又例如，多个模块或组件可以结合或一些特征可以忽略，或不执行。另一点，所显示或讨论的相互之间的耦合或直接耦合或通信连接可以是通过一些通信接口，装置或模块的间接耦合或通信连接，可以是电性，机械或其它的形式。
所述作为分离部件说明的单元可以是或者也可以不是物理上分开的，作为单元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单元来实现本实施例方案的目的。
另外，在本公开各个实施例中的各功能模块可以集成在一个处理模块中，也可以是各个模块单独物理存在，也可以两个或两个以上模块集成在一个模块中。
所述功能如果以软件功能模块的形式实现并作为独立的产品销售或使用时，可以存储在一个处理器可执行的非易失的计算机可读取存储介质中。基于这样的理解，本公开的技术方案本质上或者说对现有技术做出贡献的部分或者该技术方案的部分可以以软件产品的形式体现出来，该计算机软件产品存储在一个存储介质中，包括若干指令用以使得一台计算机设备(可以是个人计算机，服务器，或者网络设备等)执行本公开各个实施例所述方法的全部或部分步骤。而前述的存储介质包括：U盘、移动硬盘、只读存储器(Read-Only Memory，ROM)、随机存取存储器(Random Access Memory，RAM)、磁碟或者光盘等各种可以存储程序代码的介质。
最后应说明的是：以上所述实施例，仅为本公开的具体实施方式，用以说明本公开的技术方案，而非对其限制，本公开的保护范围并不局限于此，尽管参照前述实施例对本公开进行了详细的说明，本领域的普通技术人员应当理解：任何熟悉本技术领域的技术人员在本公开揭露的技术范围内，其依然可以对前述实施例所记载的技术方案进行修改或可轻易想到变化，或者对其中部分技术特征进行等同替换；而这些修改、变化或者替换，并不使相应技术方案的本质脱离本公开实施例技术方案的精神和范围，都应涵盖在本公开的保护范围之内。因此，本公开的保护范围应所述以权利要求的保护范围为准。

Claims (15)

  1. 一种数据处理方法，其特征在于，包括：
    响应于对待处理图像的入库请求，对所述待处理图像进行识别；
    在识别所述待处理图像中包含车辆的情况下，确定至少一个车辆图像；其中，每个车辆图像分别对应一个车辆；
    确定所述车辆图像的图像质量信息；
    从所述车辆图像中筛选所述图像质量信息达到第一预设条件的车辆图像，作为目标车辆图像；
    基于所述目标车辆图像和所述待处理图像，确定所述目标车辆图像的车辆特征信息；
    将所述目标车辆图像的车辆特征信息存储入库。
  2. 根据权利要求1所述的数据处理方法，其特征在于，所述基于所述目标车辆图像和所述待处理图像，确定所述目标车辆图像的车辆特征信息，包括：
    基于所述目标车辆图像在所述待处理图像上外扩处理得到的图像，确定所述目标车辆图像的车辆特征信息。
  3. 根据权利要求1或2所述的数据处理方法，其特征在于，所述图像质量信息包括拍摄的所述车辆图像的质量分数和所述车辆图像中车辆的朝向；
    从所述车辆图像中，筛选所述图像质量信息达到第一预设条件的车辆图像，作为目标车辆图像，包括：
    从所述车辆图像中，筛选所述质量分数大于预设阈值和/或所述车辆的朝向在预设朝向范围内的车辆图像，作为所述目标车辆图像。
  4. 根据权利要求1所述的数据处理方法，其特征在于，所述将所述目标车辆图像的车辆特征信息存储入库，包括：
    确定所述目标车辆图像中车辆的身份标识信息；
    建立所述身份标识信息和所述车辆特征信息之间的映射关系；
    将所述车辆特征信息、所述映射关系以及所述身份标识信息存储入库。
  5. 根据权利要求4所述的数据处理方法，其特征在于，所述车辆特征信息包括第一子信息和第二子信息；所述映射关系包括第一子关系和第二子关系；
    所述第一子信息包括以下至少一项：所述目标车辆图像中车辆的初始检测框、所述初始检测框的置信度、所述图像质量信息、车辆属性、所述车辆属性的置信度、所述目标车辆图像中车牌的检测框、所述车牌的检测框的置信度、所述车牌上的车牌号和所述车牌号的置信度；拍摄设备拍摄到所述待处理图像的拍摄时间；所述拍摄设备的设备信息；
    所述第二子信息包括以下至少一项：所述目标车辆图像中车辆的特征向量和所述目标车辆图像中车牌的特征向量；
    所述建立所述身份标识信息和所述车辆特征信息之间的映射关系，包括：
    建立所述身份标识信息和所述第一子信息之间的第一子关系；
    建立所述身份标识信息和所述第二子信息之间的第二子关系；
    所述将所述车辆特征信息、所述映射关系以及所述身份标识信息存储入库，包括：
    将所述第一子信息、所述第一子关系和所述身份标识信息存入第一数据库；
    将所述第二子信息、所述第二子关系和所述身份标识信息存入第二数据库。
  6. 根据权利要求4所述的数据处理方法，其特征在于，所述将所述车辆特征信息、所述映射关系以及所述身份标识信息存储入库，包括：
    在所述车辆特征信息包括所述车牌号的置信度的情况下，判断所述车牌号的置信度是否大于预设置信度；
    在所述车牌号的置信度大于所述预设置信度的情况下，将所述车辆特征信息、所述映射关系以及所述身份标识信息存储入库。
  7. 根据权利要求5所述的数据处理方法，其特征在于，所述设备信息包括标识信息；
    所述将所述车辆特征信息、所述映射关系以及所述身份标识信息存储入库，包括：
    在所述车辆特征信息包括所述拍摄时间和/或所述拍摄设备的设备信息的情况下，若所述拍摄时间符合第一预设格式，则将所述车辆特征信息、所述映射关系以及所述身份标识信息存储入库；
    或者，若所述标识信息符合第二预设格式，则将所述车辆特征信息、所述映射关系以及所述身份标识信息存储入库。
  8. 根据权利要求7所述的数据处理方法，其特征在于，所述确定所述目标车辆图像的车辆特征信息，包括：
    接收客户端发送的所述入库请求；所述入库请求包括所述待处理图像的所述拍摄时间和/或所述拍摄设备的设备信息；
    对所述入库请求进行解析，得到所述拍摄时间和/或所述拍摄设备的设备信息。
  9. 根据权利要求7所述的数据处理方法，其特征在于，所述设备信息包括以下至少一项：
    所述拍摄设备的位置信息；所述拍摄设备的标识信息。
  10. 根据权利要求1所述的数据处理方法，其特征在于，所述响应于对待处理图像的入库请求，对所述待处理图像进行识别，包括：
    接收客户端发送的所述入库请求，其中，所述入库请求包括所述待处理图像；
    获取所述入库请求中携带的所述待处理图像；
    在所述待处理图像符合第二预设条件的情况下，对所述待处理图像进行识别。
  11. 根据权利要求2所述的数据处理方法，其特征在于，所述基于所述目标车辆图像在所述待处理图像上外扩处理得到的图像，确定所述目标车辆图像的车辆特征信息，包括：
    将所述目标车辆图像在所述待处理图像上外扩处理得到的图像作为车辆增强图像；
    基于所述车辆增强图像，确定车牌图像；
    基于所述车辆增强图像和车辆特征检测模型，确定所述车辆增强图像的车辆特征子信息；
    基于所述车牌图像和车牌特征检测模型，确定所述车牌图像的车牌特征子信息；
    基于所述车辆特征子信息和所述车牌特征子信息，确定所述目标车辆图像的车辆特征信息。
  12. 一种数据处理装置，其特征在于，包括：
    图像识别模块，用于响应于对待处理图像的入库请求，对所述待处理图像进行识别；
    第一确定模块，用于在识别所述待处理图像中包含车辆的情况下，确定至少一个车辆图像；其中，每个车辆图像分别对应一个车辆；
    第二确定模块，用于确定所述车辆图像的图像质量信息；
    图像筛选模块，用于从所述车辆图像中筛选所述图像质量信息达到第一预设条件的车辆图像，作为目标车辆图像；
    第三确定模块，用于基于所述目标车辆图像和所述待处理图像，确定所述目标车辆图像的车辆特征信息；
    特征存储模块，用于将所述目标车辆图像的车辆特征信息存储入库。
  13. 一种计算机设备，其特征在于，包括：处理器、存储器和总线，所述存储器存储有所述处理器可执行的机器可读指令，当计算机设备运行时，所述处理器与所述存储器之间通过总线通信，所述机器可读指令被所述处理器执行时执行如权利要求1至11任一项所述的数据处理方法的步骤。
  14. 一种计算机可读存储介质，其特征在于，该计算机可读存储介质上存储有计算机程序，该计算机程序被处理器运行时执行如权利要求1至11任一项所述的数据处理方法的步骤。
  15. 一种计算机程序产品，包括计算机可读代码，或者承载有计算机可读代码的非易失性计算机可读存储介质，当所述计算机可读代码在电子设备的处理器中运行时，所述电子设备中的处理器执行用于实现权利要求1-11中的任一权利要求所述的方法。
PCT/CN2021/121759 2021-06-28 2021-09-29 一种数据处理方法、装置、计算机设备和存储介质 WO2023272991A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202110718804.5A CN113435333A (zh) 2021-06-28 2021-06-28 一种数据处理方法、装置、计算机设备和存储介质
CN202110718804.5 2021-06-28

Publications (1)

Publication Number Publication Date
WO2023272991A1 true WO2023272991A1 (zh) 2023-01-05

Family

ID=77754905

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/121759 WO2023272991A1 (zh) 2021-06-28 2021-09-29 一种数据处理方法、装置、计算机设备和存储介质

Country Status (2)

Country Link
CN (1) CN113435333A (zh)
WO (1) WO2023272991A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115830079A (zh) * 2023-02-15 2023-03-21 天翼交通科技有限公司 交通参与者的轨迹追踪方法、装置及介质

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113435333A (zh) * 2021-06-28 2021-09-24 深圳市商汤科技有限公司 一种数据处理方法、装置、计算机设备和存储介质

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111242119A (zh) * 2020-01-02 2020-06-05 腾讯科技（深圳）有限公司 车辆图像的处理方法、装置、电子设备和计算机可读介质
CN112949584A (zh) * 2021-03-30 2021-06-11 深圳市商汤科技有限公司 一种图像筛选方法及装置、电子设备和存储介质
US20210192214A1 (en) * 2020-06-30 2021-06-24 Beijing Baidu Netcom Science Technology Co., Ltd. Vehicle re-identification method, apparatus, device and storage medium
CN113435333A (zh) * 2021-06-28 2021-09-24 深圳市商汤科技有限公司 一种数据处理方法、装置、计算机设备和存储介质

Family Cites Families (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN100565555C (zh) * 2007-12-05 2009-12-02 浙江工业大学 基于计算机视觉的违章停车检测装置
CN103530366B (zh) * 2013-10-12 2016-11-23 湖北微模式科技发展有限公司 一种基于自定义特征的车辆搜索方法与系统
CN104361359A (zh) * 2014-11-25 2015-02-18 深圳市哈工大交通电子技术有限公司 基于图像检测的车辆识别方法
CN104778444B (zh) * 2015-03-10 2018-01-16 公安部交通管理科学研究所 道路场景下车辆图像的表观特征分析方法
CN105046196B (zh) * 2015-06-11 2018-04-17 西安电子科技大学 基于级联卷积神经网络的前车车辆信息结构化输出方法
CN109325420A (zh) * 2018-08-27 2019-02-12 广州烽火众智数字技术有限公司 一种基于大数据的车辆特征检索方法和系统
CN109344805A (zh) * 2018-10-29 2019-02-15 北京深瞐科技有限公司 交通违法图片智能预审方法及装置
CN109885718B (zh) * 2019-02-28 2021-05-28 江南大学 一种基于深度车贴检测的嫌疑车辆检索方法
CN110765954A (zh) * 2019-10-24 2020-02-07 浙江大华技术股份有限公司 一种车辆重识别方法、设备及存储装置
CN111914648A (zh) * 2020-07-01 2020-11-10 珠海大横琴科技发展有限公司 车辆检测和识别的方法及装置、电子设备、存储介质
CN111783724B (zh) * 2020-07-14 2024-03-26 上海依图网络科技有限公司 一种目标对象识别方法和装置

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111242119A (zh) * 2020-01-02 2020-06-05 腾讯科技（深圳）有限公司 车辆图像的处理方法、装置、电子设备和计算机可读介质
US20210192214A1 (en) * 2020-06-30 2021-06-24 Beijing Baidu Netcom Science Technology Co., Ltd. Vehicle re-identification method, apparatus, device and storage medium
CN112949584A (zh) * 2021-03-30 2021-06-11 深圳市商汤科技有限公司 一种图像筛选方法及装置、电子设备和存储介质
CN113435333A (zh) * 2021-06-28 2021-09-24 深圳市商汤科技有限公司 一种数据处理方法、装置、计算机设备和存储介质

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115830079A (zh) * 2023-02-15 2023-03-21 天翼交通科技有限公司 交通参与者的轨迹追踪方法、装置及介质

Also Published As

Publication number Publication date
CN113435333A (zh) 2021-09-24

Similar Documents

Publication Publication Date Title
US11087447B2 (en) Systems and methods for quality assurance of image recognition model
CN112560999B (zh) 一种目标检测模型训练方法、装置、电子设备及存储介质
CN109284733B (zh) 一种基于yolo和多任务卷积神经网络的导购消极行为监控方法
US20220101644A1 (en) Pedestrian re-identification method, device, electronic device and computer-readable storage medium
WO2023272991A1 (zh) 一种数据处理方法、装置、计算机设备和存储介质
CN112381775B (zh) 一种图像篡改检测方法、终端设备及存储介质
CN112861648B (zh) 文字识别方法、装置、电子设备及存储介质
US10402640B1 (en) Method and system for schematizing fields in documents
CN110245714B (zh) 图像识别方法、装置及电子设备
US20200218772A1 (en) Method and apparatus for dynamically identifying a user of an account for posting images
WO2016139964A1 (ja) 注目領域抽出装置および注目領域抽出方法
WO2020056968A1 (zh) 数据降噪方法、装置、计算机设备和存储介质
CN112633255B (zh) 目标检测方法、装置及设备
CN113011144A (zh) 表单信息的获取方法、装置和服务器
EP3274919B1 (en) Establishment anchoring with geolocated imagery
CN111368289A (zh) 一种恶意软件检测方法和装置
WO2023071180A1 (zh) 真伪识别方法、装置、电子设备以及存储介质
CN112764839B (zh) 一种用于管理服务平台的大数据配置方法及系统
CN116109864A (zh) 服装的检测识别方法、装置、终端及计算机可读存储介质
CN111737371B (zh) 可动态预测的数据流量检测分类方法及装置
CN113591657A (zh) Ocr版面识别的方法、装置、电子设备及介质
CN112333182A (zh) 档案处理方法、装置、服务器及存储介质
Huang et al. Machine Learning-Based Online Source Identification for Image Forensics
CN115174275B (zh) 基于云端的远程控制方法及装置
CN115205555B (zh) 确定相似图像的方法、训练方法、信息确定方法及设备

Legal Events

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

Ref country code: DE