Method, device, equipment and storage medium for detecting image quality of illegally parked vehicles

Abstract

The present invention discloses a method, apparatus, device and storage medium for detecting the image quality of illegal vehicles. The method includes annotating an original image data set to obtain a training image data set, inputting the training image data set into a deep convolutional neural network for training to obtain a multi-dimensional classification model; when receiving detection image data, using the multi-dimensional classification model to classify the detection image data to obtain the first probability distribution information of the first feature and the second probability distribution information of at least one second feature; based on the first probability distribution information and the second probability distribution information, determining the shooting quality of the detection image data. By annotating the non-quantized first feature and the second features of different standards, training using a deep convolutional neural network, and finally classifying the detection image data using the obtained multi-dimensional classification model, the present invention improves the detection accuracy and detection efficiency for shooting images with non-quantized features and features of different standards.

Description

Technical Field

[0001] The present invention relates to the technical field of image detection, and particularly to a method, device, equipment and storage medium for detecting the quality of photographed images of illegal vehicles. Background Art

[0002] In practical applications such as the review of traffic violation evidence pictures, the necessary conditions for a valid violation are the integrity and clarity of the illegal vehicle. Among them, in the case of speeding violations, it is especially required that the illegal vehicle meet specific integrity and clarity conditions. In actual situations, the requirements for the clarity of illegal vehicles vary from place to place. Therefore, clarity needs to be given a quantitative rather than a qualitative judgment; different violations also have different definitions of vehicle integrity. For example, in the case of speeding, if the license plate of the vehicle is cut off or incomplete, the vehicle is considered incomplete, while in other violations, the vehicle may be considered incomplete only when both the front and rear of the vehicle are incomplete. Therefore, it is necessary to judge the shooting quality of the photographed illegal vehicle: quantitative clarity and integrity in different situations.

[0003] For such problems, traditional methods generally use traditional image processing algorithms, such as calculating traditional visual features of pictures, such as HOG, HARR, etc. to judge similarity, or comprehensively judge the shooting quality of the vehicle according to the conditions of each component of the vehicle body detected by the detector. Traditional methods using traditional visual features usually have low accuracy and poor robustness. For example, in the face of different lighting conditions, vehicle integrity situations, vehicle models, and vehicle postures, it may lead to incorrect judgments, and it is not easy to give a quantifiable clarity judgment, and it is impossible to well distinguish various definitions of integrity; while the method of judging according to the vehicle body components detected by the detector cannot well judge clarity, and it is easy to make incorrect judgments due to false detections, missed detections of the detector, or different component relationships caused by various occlusions, specific vehicle postures, and different vehicle models. Therefore, how to improve the detection accuracy and detection efficiency when detecting the clarity and integrity of photographed images of illegal vehicles is a technical problem that needs to be solved urgently.

[0004] The above content is only used to assist in understanding the technical solution of the present invention, and does not represent an admission that the above content is prior art. Summary of the Invention

[0005] The main purpose of the present invention is to provide a method, device, equipment and storage medium for detecting the quality of photographed images of illegal vehicles, aiming to solve the technical problem of low detection accuracy and detection efficiency when detecting the clarity and integrity of photographed images of illegal vehicles at present.

[0006] To achieve the above object, the present invention provides a method for detecting the quality of photographed images of illegal vehicles, the method comprising the following steps:

[0007] Perform a labeling action on the original image dataset to obtain a training image dataset; wherein, the labeling action includes labeling a first feature and at least one second feature;

[0008] Input the training image dataset into a deep convolutional neural network for training to obtain a multi-dimensional classification model;

[0009] When receiving detection image data, use the multi-dimensional classification model to classify the detection image data to obtain first probability distribution information of the first feature and second probability distribution information of at least one second feature;

[0010] Based on the first probability distribution information and the second probability distribution information, determine the shooting quality of the detection image data.

[0011] Optionally, the step of performing a labeling action on the original image dataset to obtain a training image dataset specifically includes:

[0012] Perform a first labeling action on the first feature of the original image dataset to obtain a first feature label, and perform a second labeling action on at least one second feature of the original image dataset to obtain at least one second feature label;

[0013] Generate a label group according to the first feature label and at least one second feature label, and use the label group to match the corresponding original image data to obtain a training image dataset.

[0014] Optionally, the step of inputting the training image dataset into a deep convolutional neural network for training to obtain a multi-dimensional classification model specifically includes:

[0015] Input the label group corresponding to the original image data and the original image data into a deep convolutional neural network for training to obtain a multi-dimensional classification model.

[0016] Optionally, the first feature is a clarity feature, the second features include a front-end integrity feature, a rear-end integrity feature, and a license plate integrity feature, and the label group includes a clarity label, a front-end integrity label, a rear-end integrity label, and a license plate integrity label.

[0017] Optionally, the clarity feature corresponds to a first preset number of clarity levels, and the front-end integrity feature, the rear-end integrity feature, and the license plate integrity feature respectively correspond to a second preset number of integrity levels; the first labeling action is to label the clarity levels in the original image dataset, and the second labeling action is to label the front-end integrity levels, the rear-end integrity levels, and the license plate integrity levels in the original image dataset.

[0018] Optionally, the step of determining the shooting quality of the detected image data based on the first probability distribution information and the second probability distribution information specifically includes:

[0019] Based on the first probability distribution information and the second probability distribution information, determine the clarity score corresponding to the clarity feature and the integrity information corresponding to the integrity feature; wherein, the integrity information includes the front-end integrity information, the rear-end integrity information, and the license plate integrity information;

[0020] Based on a preset clarity threshold and a preset integrity requirement, determine the shooting quality of the detected image data.

[0021] Optionally, the step of determining the clarity score corresponding to the clarity feature and the integrity information corresponding to the integrity feature based on the first probability distribution information and the second probability distribution information specifically includes:

[0022] According to the first probability distribution information, assign a corresponding weight ratio to each clarity level, and based on the weight ratio, obtain the clarity score corresponding to the clarity feature;

[0023] According to the second probability distribution information, use the integrity corresponding to the level with the largest probability value in the integrity level as the integrity information.

[0024] In addition, to achieve the above object, the present invention also provides an illegal vehicle shooting image quality detection device, and the illegal vehicle shooting image quality detection device includes:

[0025] A labeling module for performing a labeling action on the original image data set to obtain a training image data set; wherein, the labeling action includes the labeling of the first feature and at least one second feature;

[0026] A training module for inputting the training image data set into a deep convolutional neural network for training to obtain a multi-dimensional classification model;

[0027] A classification module for, when receiving the detected image data, using the multi-dimensional classification model to classify the detected image data to obtain the first probability distribution information of the first feature and the second probability distribution information of at least one second feature;

[0028] A determination module for determining the shooting quality of the detected image data based on the first probability distribution information and the second probability distribution information.

[0029] In addition, to achieve the above object, the present invention further provides an illegal vehicle photographed image quality detection device, where the illegal vehicle photographed image quality detection device includes: a memory, a processor, and an illegal vehicle photographed image quality detection program stored on the memory and executable on the processor. When the illegal vehicle photographed image quality detection program is executed by the processor, the steps of the illegal vehicle photographed image quality detection method as described above are implemented.

[0030] In addition, to achieve the above object, the present invention further provides a storage medium, on which an illegal vehicle photographed image quality detection program is stored. When the illegal vehicle photographed image quality detection program is executed by a processor, the steps of the illegal vehicle photographed image quality detection method as described above are implemented.

[0031] An illegal vehicle photographed image quality detection method, device, equipment, and storage medium proposed in an embodiment of the present invention. The method includes performing a labeling operation on an original image data set to obtain a training image data set, inputting the training image data set into a deep convolutional neural network for training to obtain a multi-dimensional classification model; when receiving detection image data, using the multi-dimensional classification model to classify the detection image data to obtain first probability distribution information of a first feature and second probability distribution information of at least one second feature; based on the first probability distribution information and the second probability distribution information, determining the shooting quality of the detection image data. The present invention improves the detection accuracy and detection efficiency of shooting images with non-quantized features and different standard features by labeling non-quantized first features and second features with different standards, training using a deep convolutional neural network, and finally using the obtained multi-dimensional classification model to classify the detection image data. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a schematic structural diagram of an illegal vehicle photographed image quality detection device in an embodiment of the present invention;

[0033] Figure 2 It is a schematic flowchart of an illegal vehicle photographed image quality detection method in an embodiment of the present invention;

[0034] Figure 3 It is a structural block diagram of an illegal vehicle photographed image quality detection device in an embodiment of the present invention.

[0035] The realization, functional features, and advantages of the object of the present invention will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0036] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0037] In practical applications such as the review of traffic violation evidence pictures, the necessary conditions for a valid violation are the integrity and clarity of the violating vehicle. Among them, in the case of speeding violations, it is especially required that the violating vehicle meet specific integrity and clarity conditions. In actual situations, the requirements for the clarity of violating vehicles vary from place to place. Therefore, clarity needs to be given a quantitative rather than a qualitative judgment. Different violations also have different definitions of vehicle integrity. For example, in the case of speeding, if the license plate of the vehicle is cut off or incomplete, the vehicle is considered incomplete, while in other violations, the vehicle may be considered incomplete only when both the front and rear of the vehicle are incomplete. Therefore, it is necessary to judge the shooting quality of the captured violating vehicle: quantitative clarity and integrity in different situations.

[0038] For such problems, traditional methods generally use traditional image processing algorithms, such as calculating traditional visual features of pictures like HOG, HARR, etc. to judge similarity, or comprehensively judging the shooting quality of the vehicle based on the conditions of each component of the vehicle body detected by the detector. Traditional methods using traditional visual features usually have low accuracy and poor robustness. For example, in the face of different lighting conditions, vehicle integrity situations, vehicle models, and vehicle postures, it may lead to incorrect judgments, and it is not easy to give a quantifiable clarity judgment, and it cannot well distinguish multiple definitions of integrity. The method of judging based on the vehicle body components detected by the detector cannot well judge clarity, and is prone to misdetection, missed detection by the detector, or incorrect judgments due to various occlusions, specific vehicle postures, and different component relationships caused by different vehicle models. Therefore, how to improve the detection accuracy and detection efficiency when detecting the clarity and integrity of the captured images of violating vehicles is a technical problem that urgently needs to be solved.

[0039] To solve this problem, various embodiments of the method for detecting the quality of captured images of violating vehicles according to the present invention are proposed. The method for detecting the quality of captured images of violating vehicles provided by the present invention improves the detection accuracy and detection efficiency of captured images with non-quantifiable features and different standard features by annotating non-quantifiable first features and second features with different standards, training using a deep convolutional neural network, and finally classifying the detected image data using the obtained multi-dimensional classification model.

[0040] Refer to Figure 1 , Figure 1 which is a schematic structural diagram of the device for detecting the quality of captured images of violating vehicles according to the embodiment solution of the present invention.

[0041] The device can be a user equipment (UE) such as a mobile phone, a smart phone, a laptop computer, a digital broadcast receiver, a personal digital assistant (PDA), a tablet computer (PAD), a handheld device, a vehicle-mounted device, a wearable device, a computing device or other processing devices connected to a wireless modem, a mobile station (MS), etc. The device may be referred to as a user terminal, a portable terminal, a desktop terminal, etc.

[0042] Generally, the device includes: at least one processor 301, a memory 302, and an illegal vehicle captured image quality detection program stored on the memory and executable on the processor. The illegal vehicle captured image quality detection program is configured to implement the steps of the illegal vehicle captured image quality detection method as described above.

[0043] The processor 301 may include one or more processing cores, such as a 4-core processor, an 8-core processor, etc. The processor 301 may be implemented in at least one hardware form of DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), or PLA (Programmable Logic Array). The processor 301 may also include a main processor and a coprocessor. The main processor is a processor for processing data in the wake state, also known as the CPU (Central Processing Unit); the coprocessor is a low-power processor for processing data in the standby state. In some embodiments, the processor 301 may be integrated with a GPU (Graphics Processing Unit), and the GPU is responsible for rendering and drawing the content to be displayed on the display screen. The processor 301 may also include an AI (Artificial Intelligence) processor, which is used to process operations related to the illegal vehicle captured image quality detection, enabling the illegal vehicle captured image quality detection model to autonomously train and learn, improving efficiency and accuracy.

[0044] The memory 302 may include one or more computer-readable storage media, which may be non-transitory. The memory 302 may also include high-speed random access memory and non-volatile memory, such as one or more magnetic disk storage devices and flash storage devices. In some embodiments, the non-transitory computer-readable storage media in the memory 302 is used to store at least one instruction for being executed by the processor 301 to implement the method for detecting the image quality of illegally parked vehicle captured images provided in the method embodiments of the present application.

[0045] In some embodiments, the terminal may further optionally include: a communication interface 303 and at least one peripheral device. The processor 301, the memory 302, and the communication interface 303 may be connected through a bus or signal lines. Each peripheral device may be connected to the communication interface 303 through a bus, signal lines, or a circuit board. Specifically, the peripheral device includes at least one of a radio frequency circuit 304, a display screen 305, and a power supply 306.

[0046] The communication interface 303 may be used to connect at least one peripheral device related to I / O (Input / Output) to the processor 301 and the memory 302. The communication interface 303 is used to receive the movement trajectories and other data of multiple mobile terminals uploaded by the user through the peripheral device. In some embodiments, the processor 301, the memory 302, and the communication interface 303 are integrated on the same chip or circuit board; in some other embodiments, any one or two of the processor 301, the memory 302, and the communication interface 303 may be implemented on a separate chip or circuit board, and this embodiment does not limit this.

[0047] The radio frequency circuit 304 is used to receive and transmit RF (Radio Frequency) signals, also known as electromagnetic signals. The radio frequency circuit 304 communicates with the communication network and other communication devices through electromagnetic signals, so as to obtain the movement trajectories and other data of multiple mobile terminals. The radio frequency circuit 304 converts electrical signals into electromagnetic signals for transmission, or converts the received electromagnetic signals into electrical signals. Optionally, the radio frequency circuit 304 includes: an antenna system, an RF transceiver, one or more amplifiers, a tuner, an oscillator, a digital signal processor, a codec chipset, a user identity module card, and so on. The radio frequency circuit 304 can communicate with other terminals through at least one wireless communication protocol. The wireless communication protocol includes but is not limited to: metropolitan area network, various generations of mobile communication networks (2G, 3G, 4G, and 5G), wireless local area network, and / or WiFi (Wireless Fidelity) network. In some embodiments, the radio frequency circuit 304 may further include a circuit related to NFC (Near Field Communication), which is not limited in this application.

[0048] The display screen 305 is used to display the UI (User Interface). The UI may include graphics, text, icons, videos, and any combination thereof. When the display screen 305 is a touch display screen, the display screen 305 also has the ability to collect touch signals on or above the surface of the display screen 305. The touch signals can be input to the processor 301 as control signals for processing. At this time, the display screen 305 can also be used to provide virtual buttons and / or a virtual keyboard, also known as soft buttons and / or a soft keyboard. In some embodiments, the display screen 305 can be one, the front panel of the electronic device; in other embodiments, the display screen 305 can be at least two, respectively arranged on different surfaces of the electronic device or in a folding design; in still other embodiments, the display screen 305 can be a flexible display screen, arranged on the curved surface or folding surface of the electronic device. Even, the display screen 305 can be set to an irregular non-rectangular shape, that is, a special-shaped screen. The display screen 305 can be prepared from materials such as LCD (Liquid Crystal Display) and OLED (Organic Light-Emitting Diode).

[0049] The power supply 306 is used to supply power to each component in the electronic device. The power supply 306 can be alternating current, direct current, a disposable battery, or a rechargeable battery. When the power supply 306 includes a rechargeable battery, the rechargeable battery can support wired charging or wireless charging. The rechargeable battery can also be used to support fast charging technology.

[0050] Those skilled in the art can understand that Figure 1 the structure shown in Figure 1 does not constitute a limitation on the illegal vehicle photograph image quality detection device, and may include more or fewer components than shown in the figure, or combine some components, or have different component arrangements.

[0051] An embodiment of the present invention provides a method for detecting the quality of an illegal vehicle photograph image. Referring to Figure 2 , Figure 2 which is a schematic flowchart of the first embodiment of the method for detecting the quality of an illegal vehicle photograph image of the present invention.

[0052] In this embodiment, the method for detecting the quality of an illegal vehicle photograph image includes the following steps:

[0053] Step S100, perform a labeling action on the original image dataset to obtain a training image dataset; wherein, the labeling action includes labeling a first feature and at least one second feature.

[0054] Specifically, in order to obtain a training image dataset for inputting into a deep convolutional neural network, a multi-dimensional classification model is constructed to detect photograph images with non-quantized features and different standard features. When performing the labeling action, a first labeling action is performed on the first feature of the original image dataset to obtain a first feature label, and a second labeling action is performed on at least one second feature of the original image dataset to obtain at least one second feature label; according to the first feature label and at least one second feature label, a label group is generated, and the original image data corresponding to the label group is matched to obtain a training image dataset.

[0055] Further, the first feature is a clarity feature, and the second features include a front-end integrity feature, a rear-end integrity feature, and a license plate integrity feature, and the label group includes a clarity label, a front-end integrity label, a rear-end integrity label, and a license plate integrity label.

[0056] In some embodiments, the clarity feature corresponds to a first preset number of clarity levels, and the front-end integrity feature, the rear-end integrity feature, and the license plate integrity feature respectively correspond to a second preset number of integrity levels; the first labeling action is to label the clarity level in the original image dataset, and the second labeling action is to label the front-end integrity level, the rear-end integrity level, and the license plate integrity level in the original image dataset.

[0057] It is easy to understand that when performing labeling, for non-quantized features such as the clarity feature, the original images of illegal vehicles taken can be classified according to clarity by manual labeling or a labeling tool to obtain a first preset number of clarity levels. In this embodiment, the original images are divided into 12 levels according to clarity.

[0058] In addition, for different standard features such as license plate integrity features, the original images of the illegal vehicles captured can be classified according to the front of the vehicle, the rear of the vehicle, and the integrity of the license plate by manual annotation or annotation tools, obtaining a second preset number of integrity levels. In this embodiment, the front of the vehicle, the rear of the vehicle, and the license plate are respectively divided into 4 levels according to integrity.

[0059] Based on the above annotation process, a 4D label [a1, b1, b2, b3] can be obtained, where a1 has 12 labels representing clarity classification, b1 has 4 labels representing front vehicle integrity classification, b2 has 4 labels representing rear vehicle integrity classification, and b3 has 4 labels representing license plate integrity classification.

[0060] Step S200: Input the training image dataset into a deep convolutional neural network for training to obtain a multi-dimensional classification model.

[0061] Specifically, after obtaining the label group and the original image data, input the label group corresponding to the original image data and the original image data into a deep convolutional neural network for training to obtain a multi-dimensional classification model.

[0062] It is easy to understand that during the training process, each illegal vehicle picture in the original image dataset and the corresponding 4D label [a1, b1, b2, b3] are input into a deep convolutional neural network for multi-dimensional classification model training, and the input and training process continues until the training is completed.

[0063] Step S300: When receiving the detection image data, use the multi-dimensional classification model to classify the detection image data to obtain the first probability distribution information of the first feature and the second probability distribution information of at least one second feature.

[0064] Specifically, after the training is completed and the multi-dimensional classification model is obtained, use the multi-dimensional classification model to classify the received detection image data to obtain the first probability distribution information of clarity and the second probability distribution information of the integrity of the front of the vehicle, the rear of the vehicle, and the license plate.

[0065] It is easy to understand that in practical applications, the trained multi-dimensional classification model is used to classify the illegal vehicles detected by the detector to be determined, obtaining a probability distribution of 24 lengths; among them, it includes a clarity probability distribution of 12 lengths, a front vehicle integrity probability distribution of 4 lengths, a rear vehicle integrity probability distribution of 4 lengths, and a license plate integrity probability distribution of 4 lengths.

[0066] Step S400: Based on the first probability distribution information and the second probability distribution information, determine the shooting quality of the detection image data.

[0067] Specifically, based on the first probability distribution information and the second probability distribution information, determine the clarity score corresponding to the clarity feature and the integrity information corresponding to the integrity feature; wherein, the integrity information includes the front-end integrity information, the rear-end integrity information, and the license plate integrity information; based on a preset clarity threshold and a preset integrity requirement, determine the shooting quality of the detected image data.

[0068] Further, according to the first probability distribution information, assign a corresponding weight ratio to each clarity level, and based on the weight ratio, obtain the clarity score corresponding to the clarity feature; according to the second probability distribution information, use the integrity corresponding to the level with the largest probability value in the integrity level as the integrity information.

[0069] In practical applications, when performing quality detection on clarity parameters, use the probability distribution of 12 digits corresponding to clarity. Each score corresponding to clarity is given a corresponding weight ratio score, and the product of the weight ratio score and the corresponding probability value is used as the final clarity score.

[0070] In addition, when performing quality detection on integrity parameters, use the probability distribution of 4 digits corresponding to 3 integrity levels respectively, and obtain the corresponding integrity classification result according to the largest score.

[0071] In the final quality detection of pictures of illegal vehicles, according to the definition of clarity in each region, those higher than the given clarity threshold are classified as clear, and those lower than the clarity threshold are classified as unclear; according to specific illegal requirements, judge the integrity classification of the corresponding front-end, rear-end, and license plate, and give the final integrity judgment. Finally, based on the integrity and clarity requirements, obtain the quality detection result of the pictures of illegal vehicles.

[0072] In this embodiment, the method for detecting the shooting quality of illegal vehicles based on deep learning is different from the traditional image feature method and the detection component logic method. It directly judges based on a deep convolutional neural network, and makes distinctions for the quantifiable requirements of clarity and different definitions of integrity, and can obtain more accurate judgments and give more flexible results according to the requirements of each region. By labeling the non-quantifiable first feature and the second feature with different standards, and using a deep convolutional neural network for training, and finally using the obtained multi-dimensional classification model to classify the detected image data, the detection accuracy and detection efficiency of the shooting images with non-quantifiable features and different standard features are improved.

[0073] Refer to Figure 3 , Figure 3 which is the structural block diagram of the embodiment of the device for detecting the quality of pictures of illegal vehicles according to the present invention.

[0074] As Figure 3As shown in the figure, the illegal vehicle photograph image quality detection device proposed in the embodiment of the present invention includes:

[0075] A labeling module 10, configured to perform a labeling action on the original image dataset to obtain a training image dataset; wherein, the labeling action includes labeling of a first feature and at least one second feature;

[0076] A training module 20, configured to input the training image dataset into a deep convolutional neural network for training to obtain a multi-dimensional classification model;

[0077] A classification module 30, configured to, when receiving detection image data, classify the detection image data by using the multi-dimensional classification model to obtain first probability distribution information of the first feature and second probability distribution information of at least one second feature;

[0078] A determination module 40, configured to determine the shooting quality of the detection image data based on the first probability distribution information and the second probability distribution information.

[0079] As an implementation manner, the labeling module 10 is further configured to perform a first labeling action on the first feature of the original image dataset to obtain a first feature label, and perform a second labeling action on at least one second feature of the original image dataset to obtain at least one second feature label; generate a label group according to the first feature label and the at least one second feature label, and use the label group to match the corresponding original image data to obtain a training image dataset.

[0080] As an implementation manner, the training module 20 is further configured to input the label group corresponding to the original image data and the original image data into a deep convolutional neural network for training to obtain a multi-dimensional classification model.

[0081] As an implementation manner, in the labeling module 10, the first feature is a clarity feature, and the second features include a front-end integrity feature, a rear-end integrity feature, and a license plate integrity feature, and the label group includes a clarity label, a front-end integrity label, a rear-end integrity label, and a license plate integrity label.

[0082] As an implementation manner, in the labeling module 10, the clarity feature corresponds to a first preset number of clarity levels, and the front-end integrity feature, the rear-end integrity feature, and the license plate integrity feature respectively correspond to a second preset number of integrity levels; the first labeling action is to label the clarity levels in the original image dataset, and the second labeling action is to label the front-end integrity levels, the rear-end integrity levels, and the license plate integrity levels in the original image dataset.

[0083] As an implementation manner, the determination module 40 is further configured to determine a sharpness score corresponding to the sharpness feature and integrity information corresponding to the integrity feature based on the first probability distribution information and the second probability distribution information; wherein the integrity information includes front-end integrity information, parking-space integrity information, and license-plate integrity information; and determine the shooting quality of the detected image data based on a preset sharpness threshold and a preset integrity requirement.

[0084] As an implementation manner, the determination module 40 is further configured to assign a corresponding weight ratio to each sharpness level according to the first probability distribution information, and obtain a sharpness score corresponding to the sharpness feature based on the weight ratio; and use the integrity corresponding to the level with the largest probability value in the integrity level as the integrity information according to the second probability distribution information.

[0085] The illegal vehicle shooting image quality detection device provided in this embodiment improves the detection accuracy and detection efficiency of shooting images with non-quantized features and different standard features by labeling non-quantized first features and second features with different standards, training using a deep convolutional neural network, and finally classifying the detected image data using the obtained multi-dimensional classification model.

[0086] For other embodiments or specific implementation manners of the illegal vehicle shooting image quality detection device of the present invention, reference may be made to the above method embodiments, and details will not be described herein again.

[0087] In addition, an embodiment of the present invention further provides a storage medium, on which an illegal vehicle shooting image quality detection program is stored. When the illegal vehicle shooting image quality detection program is executed by a processor, the steps of the illegal vehicle shooting image quality detection method as described above are implemented. Therefore, details will not be described herein again. In addition, the beneficial effects of using the same method will not be described again. For the technical details not disclosed in the embodiment of the computer-readable storage medium involved in this application, reference may be made to the description of the method embodiment of this application. By way of example, the program instructions may be deployed to be executed on one computing device, or on multiple computing devices located at one location, or on multiple computing devices distributed at multiple locations and interconnected by a communication network.

[0088] Those of ordinary skill in the art can understand that all or part of the processes of implementing the methods in the above embodiments can be completed by instructing relevant hardware through a computer program. The above program can be stored in a computer-readable storage medium. When the program is executed, it may include the processes of the above method embodiments. Among them, the above storage medium may be a magnetic disk, an optical disc, a read-only memory (ROM), or a random access memory (RAM), etc.

[0089] In addition, it should be noted that the device embodiments described above are merely illustrative. 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 distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. In addition, in the attached drawings of the device embodiments provided by the present invention, the connection relationships between the modules indicate that there is a communication connection between them, which can be specifically implemented as one or more communication buses or signal lines. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0090] Through the description of the above embodiments, those skilled in the art can clearly understand that the present invention can be implemented by means of software plus necessary general hardware, and of course, it can also be implemented by dedicated hardware including application-specific integrated circuits, dedicated CPUs, dedicated memories, dedicated components, etc. Generally, functions completed by computer programs can be easily implemented by corresponding hardware, and the specific hardware structures for implementing the same function can also be various, such as analog circuits, digital circuits or dedicated circuits. However, for the present invention, in more cases, software program implementation is a better implementation manner. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art can be embodied in the form of a software product. This computer software product is stored in a readable storage medium, such as a floppy disk, USB flash drive, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disk or optical disc of a computer, etc., and includes several instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

Claims

1. A method for detecting the image quality of illegal vehicle photographs, characterized in that, The method includes the following steps: Performing a labeling action on the original image dataset to obtain a training image dataset; wherein, the labeling action includes labeling of a first feature and at least one second feature; Specifically including: Performing a first labeling action on the first feature of the original image dataset to obtain a first feature label, and performing a second labeling action on at least one second feature of the original image dataset to obtain at least one second feature label; Generating a label group according to the first feature label and the at least one second feature label, and using the label group to match the corresponding original image data to obtain a training image dataset; The first feature is a clarity feature, the second features include a front-end integrity feature, a rear-end integrity feature, and a license plate integrity feature, and the label group includes a clarity label, a front-end integrity label, a rear-end integrity label, and a license plate integrity label; Inputting the training image dataset into a deep convolutional neural network for training to obtain a multi-dimensional classification model; When receiving detection image data, using the multi-dimensional classification model to classify the detection image data to obtain first probability distribution information of the first feature and second probability distribution information of the at least one second feature; Based on the first probability distribution information and the second probability distribution information, determining the shooting quality of the detection image data.

2. The method for detecting the image quality of illegally parked vehicles according to claim 1, wherein The step of inputting the training image dataset into a deep convolutional neural network for training to obtain a multi-dimensional classification model specifically includes: Inputting the label group corresponding to the original image data and the original image data into a deep convolutional neural network for training to obtain a multi-dimensional classification model.

3. The method for detecting the image quality of illegal vehicle shooting according to claim 1, characterized in that, The clarity feature corresponds to a first preset number of clarity levels, and the front-end integrity feature, the rear-end integrity feature, and the license plate integrity feature respectively correspond to a second preset number of integrity levels; the first labeling action is to label the clarity levels in the original image dataset, and the second labeling action is to label the front-end integrity levels, the rear-end integrity levels, and the license plate integrity levels in the original image dataset.

4. The method for detecting the image quality of illegal vehicle shooting according to claim 3, characterized in that, The step of determining the shooting quality of the detection image data based on the first probability distribution information and the second probability distribution information specifically includes: Based on the first probability distribution information and the second probability distribution information, determining a clarity score corresponding to the clarity feature and integrity information corresponding to the integrity feature; wherein, the integrity information includes front-end integrity information, rear-end integrity information, and license plate integrity information; Based on a preset clarity threshold and a preset integrity requirement, determining the shooting quality of the detection image data.

5. The method for detecting the image quality of illegally parked vehicles according to claim 4, characterized in that The step of determining a clarity score corresponding to the clarity feature and integrity information corresponding to the integrity feature based on the first probability distribution information and the second probability distribution information specifically includes: According to the first probability distribution information, assigning a corresponding weight ratio to each clarity level, and based on the weight ratio, obtaining the clarity score corresponding to the clarity feature; According to the second probability distribution information, taking the integrity corresponding to the level with the largest probability value in the integrity levels as the integrity information.

6. An illegal vehicle photograph image quality detection device, characterized in that, The illegal vehicle shooting image quality detection device includes: An annotation module is used to perform annotation actions on the original image dataset to obtain a training image dataset. Among them, the annotation actions include the annotation of the first feature and at least one second feature; Specifically, it includes: Performing a first annotation action on the first feature of the original image dataset to obtain a first feature label, and performing a second annotation action on at least one second feature of the original image dataset to obtain at least one second feature label; Generating a label group according to the first feature label and at least one second feature label, and using the label group to match the corresponding original image data to obtain a training image dataset; The first feature is a clarity feature, the second features include a front-end integrity feature, a rear-end integrity feature, and a license plate integrity feature, and the label group includes a clarity label, a front-end integrity label, a rear-end integrity label, and a license plate integrity label; A training module is used to input the training image dataset into a deep convolutional neural network for training to obtain a multi-dimensional classification model; A classification module is used to, when receiving detection image data, use the multi-dimensional classification model to classify the detection image data to obtain first probability distribution information of the first feature and second probability distribution information of at least one second feature; A determination module is used to determine the shooting quality of the detection image data based on the first probability distribution information and the second probability distribution information.

7. An illegal vehicle photographed image quality detection device, characterized in that The illegal vehicle shooting image quality detection device includes: a memory, a processor, and an illegal vehicle shooting image quality detection program stored on the memory and executable on the processor. When the illegal vehicle shooting image quality detection program is executed by the processor, it implements the steps of the illegal vehicle shooting image quality detection method according to any one of claims 1 to 5.

8. A storage medium, characterized in that, An illegal vehicle shooting image quality detection program is stored on the storage medium. When the illegal vehicle shooting image quality detection program is executed by the processor, it implements the steps of the illegal vehicle shooting image quality detection method according to any one of claims 1 to 5.

Images