Vehicle authenticity authentication method, authentication device and readable storage medium

Abstract

The present application provides a method for authenticating the authenticity of a vehicle, an authentication device, and a readable storage medium, by acquiring multiple verification images of the vehicle to be authenticated; inputting the multiple verification images into a pre-trained vehicle abnormality probability prediction model to obtain the abnormal authentication probability of the vehicle to be authenticated; if it is detected that the abnormal authentication probability is greater than or equal to a preset threshold, the vehicle to be authenticated is determined to be an authentication abnormal vehicle. In this way, the authenticity of the vehicle to be authenticated can be quickly identified, with a high recognition accuracy, which helps to reduce the consumption of manual resources and time resources in the vehicle authentication process.

Description

Technical Field

[0001] The present application relates to the field of vehicle detection technology, and in particular to a vehicle authenticity authentication method, an authentication device, and a readable storage medium. Background Art

[0002] During the online car-hailing registration process, the driver is required to provide an image of the vehicle's appearance. Verifying the authenticity of the vehicle image provided by the driver is an important part of ensuring the authenticity of the driver's information. If the vehicle image complies with the rules, it will be reviewed and approved in a timely manner. If the vehicle image does not comply with the rules, the driver will be reminded in a timely manner until the driver provides a real vehicle image. This not only ensures the authenticity of the information provided by the driver, but also improves the driver's experience during the vehicle verification process. Therefore, how to accurately predict the authenticity of the vehicle through the vehicle image is an urgent problem to be solved. Summary of the invention

[0003] In view of this, the purpose of the present application is to provide a vehicle authenticity authentication method, an authentication device and a readable storage medium. By accumulating and identifying multiple collected vehicle images to be authenticated, the authenticity of the vehicle to be authenticated can be quickly identified with high recognition accuracy, which helps to reduce the consumption of manual and time resources in the vehicle authentication process.

[0004] According to a first aspect of the present application, a method for authenticating the authenticity of a vehicle is provided, the authentication method comprising:

[0005] Acquire multiple verification images of the vehicle to be authenticated;

[0006] Inputting the plurality of verification images into a pre-trained vehicle abnormality probability prediction model to obtain an abnormal authentication probability of the vehicle to be authenticated;

[0007] If it is detected that the abnormal authentication probability is greater than or equal to a preset threshold, the vehicle to be authenticated is determined to be an authentication abnormal vehicle.

[0008] In some embodiments of the present application, the step of inputting the plurality of verification images into a pre-trained vehicle abnormality probability prediction model to obtain the authentication abnormality probability of the vehicle to be authenticated includes:

[0009] Input each verification image into the vehicle abnormality probability prediction model in sequence according to the image acquisition order, and obtain the initial abnormality probability of each verification image in sequence according to the image input order;

[0010] Based on the initial abnormal probability of each verification image, the abnormal authentication probability of the vehicle to be authenticated is determined.

[0011] In some embodiments of the present application, determining the abnormal authentication probability of the vehicle to be authenticated based on the initial abnormal probability of each verification image includes:

[0012] In the process of obtaining the initial abnormality probability of each verification image in turn, when the initial abnormality probability of the Nth verification image is obtained, if it is detected that the sum of the initial abnormality probabilities of the first N verification images is greater than or equal to the preset threshold, the sum of the initial abnormality probabilities of the first N verification images is determined to be the abnormal authentication probability of the vehicle to be authenticated.

[0013] In some embodiments of the present application, determining the abnormal authentication probability of the vehicle to be authenticated based on the initial abnormal probability of each verification image includes:

[0014] In the process of obtaining the initial abnormal probability of each verification image in sequence, when the initial abnormal probability of the Mth verification image is obtained, it is detected that the authentication time from obtaining the initial abnormal probability of the first verification image to obtaining the initial abnormal probability of the Mth verification image is greater than or equal to the preset time threshold, and the sum of the initial abnormal probabilities of the first M verification images is determined as the abnormal authentication probability of the vehicle to be authenticated;

[0015] After inputting the plurality of verification images into a pre-trained vehicle abnormality probability prediction model to obtain the abnormality authentication probability of the vehicle to be authenticated, the authentication method includes:

[0016] If it is detected that the abnormal authentication probability is less than a preset threshold, it is determined that the vehicle to be authenticated is a successfully authenticated vehicle.

[0017] In some embodiments of the present application, the initial abnormality probability corresponding to each verification image is determined by the following steps:

[0018] Performing feature extraction processing on the verification image and determining a feature thermal image of the verification image;

[0019] Based on the characteristic thermal image, determining an image of a region of interest of the verification image from the verification image;

[0020] Based on the region of interest image, an initial abnormality probability of the verification image is determined.

[0021] In some embodiments of the present application, the step of performing feature extraction processing on the verification image and determining a feature thermal image of the verification image includes:

[0022] Extract feature layers under each preset feature dimension from the verification image;

[0023] Based on the extracted multiple feature layers and the preset weight of each preset feature dimension, determine the image category of the verification image and the initial thermal image of the verification image corresponding to the image category;

[0024] The initial thermal image is enlarged to be consistent with the image size of the verification image, and the enlarged initial thermal image is determined as the characteristic thermal image of the verification image.

[0025] In some embodiments of the present application, determining the region of interest image of the verification image from the verification image based on the characteristic thermal image includes:

[0026] Determine a regional characteristic value corresponding to each characteristic image region in the characteristic thermal image, and detect whether each regional characteristic value is greater than a preset extraction threshold;

[0027] Extracting the verification image region corresponding to each characteristic image region having a region characteristic value greater than a preset extraction threshold in the verification image;

[0028] Each verification image region is concatenated to generate a region of interest image of the verification image.

[0029] In some embodiments of the present application, determining the initial abnormality probability of the verification image based on the region of interest image includes:

[0030] Determining a classification category of the region of interest image based on a region value corresponding to each region of interest included in the region of interest image and a weight coefficient corresponding to each region value;

[0031] Based on the preset probability value corresponding to the classification category, an initial abnormal probability of the verification image is determined.

[0032] In some embodiments of the present application, after determining that the vehicle to be authenticated is an authentication abnormal vehicle if the abnormal authentication probability is detected to be greater than or equal to a preset threshold, the authentication method further includes:

[0033] The authentication result that the vehicle to be authenticated is an authentication abnormal vehicle is sent to the user, and the user is prompted to provide a verification image of a vehicle that matches the pre-registered vehicle information or a real vehicle.

[0034] In some embodiments of the present application, the vehicle abnormality probability prediction model is trained by the following steps:

[0035] Obtain multiple sample images of multiple sample vehicles and the actual abnormal authentication probability corresponding to each sample vehicle;

[0036] For each sample vehicle, multiple sample images of the sample vehicle and the actual abnormal authentication probability are input into the constructed deep learning model to obtain the predicted abnormal authentication probability corresponding to the sample vehicle;

[0037] For each sample vehicle, determining a deviation value between an actual abnormal authentication probability of the sample vehicle and the predicted abnormal authentication probability;

[0038] If there is a sample vehicle whose corresponding deviation value is greater than the preset deviation threshold, adjust the relevant parameters in the deep learning model until the deviation value corresponding to each sample vehicle is less than or equal to the preset deviation threshold, determine that the deep learning model training is completed, and determine the trained deep learning model as the trained vehicle abnormality probability prediction model.

[0039] According to a second aspect of the present application, a vehicle authenticity authentication device is provided, the authentication device comprising:

[0040] An image acquisition module, used to acquire multiple verification images of the vehicle to be authenticated;

[0041] A probability determination module, used for inputting the plurality of verification images into a pre-trained vehicle abnormality probability prediction model to obtain an abnormal authentication probability of the vehicle to be authenticated;

[0042] The abnormal vehicle determination module is used to determine that the vehicle to be authenticated is an authentication abnormal vehicle if it is detected that the abnormal authentication probability is greater than or equal to a preset threshold.

[0043] In some embodiments of the present application, when the probability determination module is used to input the multiple verification images into a pre-trained vehicle abnormality probability prediction model to obtain the abnormal authentication probability of the vehicle to be authenticated, the probability determination module is used to:

[0044] Input each verification image into the vehicle abnormality probability prediction model in sequence according to the image acquisition order, and obtain the initial abnormality probability of each verification image in sequence according to the image input order;

[0045] Based on the initial abnormal probability of each verification image, the abnormal authentication probability of the vehicle to be authenticated is determined.

[0046] In some embodiments of the present application, when the probability determination module is used to determine the abnormal authentication probability of the vehicle to be authenticated based on the initial abnormal probability of each verification image, the probability determination module is used to:

[0047] In the process of obtaining the initial abnormality probability of each verification image in turn, when the initial abnormality probability of the Nth verification image is obtained, if it is detected that the sum of the initial abnormality probabilities of the first N verification images is greater than or equal to the preset threshold, the sum of the initial abnormality probabilities of the first N verification images is determined to be the abnormal authentication probability of the vehicle to be authenticated.

[0048] In some embodiments of the present application, when the probability determination module is used to determine the abnormal authentication probability of the vehicle to be authenticated based on the initial abnormal probability of each verification image, the probability determination module is used to:

[0049] In the process of obtaining the initial abnormal probability of each verification image in sequence, when the initial abnormal probability of the Mth verification image is obtained, it is detected that the authentication time from obtaining the initial abnormal probability of the first verification image to obtaining the initial abnormal probability of the Mth verification image is greater than or equal to the preset time threshold, and the sum of the initial abnormal probabilities of the first M verification images is determined as the abnormal authentication probability of the vehicle to be authenticated;

[0050] The authentication device further comprises an authentication success module, wherein the authentication success module is used to:

[0051] If it is detected that the abnormal authentication probability is less than a preset threshold, it is determined that the vehicle to be authenticated is a successfully authenticated vehicle.

[0052] In some embodiments of the present application, the probability determination module is used to determine the initial abnormality probability corresponding to each verification image through the following steps:

[0053] Performing feature extraction processing on the verification image and determining a feature thermal image of the verification image;

[0054] Based on the characteristic thermal image, determining an image of a region of interest of the verification image from the verification image;

[0055] Based on the region of interest image, an initial abnormality probability of the verification image is determined.

[0056] In some embodiments of the present application, when the probability determination module is used to perform feature extraction processing on the verification image and determine the characteristic thermal image of the verification image, the probability determination module is used to:

[0057] Extract feature layers under each preset feature dimension from the verification image;

[0058] Based on the extracted multiple feature layers and the preset weight of each preset feature dimension, determine the image category of the verification image and the initial thermal image of the verification image corresponding to the image category;

[0059] The initial thermal image is enlarged to be consistent with the image size of the verification image, and the enlarged initial thermal image is determined as the characteristic thermal image of the verification image.

[0060] In some embodiments of the present application, when the probability determination module is used to determine the region of interest image of the verification image from the verification image based on the characteristic thermal image, the probability determination module is used to:

[0061] Determine a regional characteristic value corresponding to each characteristic image region in the characteristic thermal image, and detect whether each regional characteristic value is greater than a preset extraction threshold;

[0062] Extracting the verification image region corresponding to each characteristic image region having a region characteristic value greater than a preset extraction threshold in the verification image;

[0063] Each verification image region is concatenated to generate a region of interest image of the verification image.

[0064] In some embodiments of the present application, the probability determination module is used to determine the initial abnormality probability of the verification image based on the region of interest image, and the probability determination module is used to:

[0065] Determining a classification category of the region of interest image based on a region value corresponding to each region of interest included in the region of interest image and a weight coefficient corresponding to each region value;

[0066] Based on the preset probability value corresponding to the classification category, an initial abnormal probability of the verification image is determined.

[0067] In some embodiments of the present application, the authentication device further includes a user prompt module, and the user prompt module is used to:

[0068] The authentication result that the vehicle to be authenticated is an authentication abnormal vehicle is sent to the user, and the user is prompted to provide a verification image of a vehicle that matches the pre-registered vehicle information or a real vehicle.

[0069] In some embodiments of the present application, the authentication device further includes a model training module, and the model training module is used to:

[0070] Obtain multiple sample images of multiple sample vehicles and the actual abnormal authentication probability corresponding to each sample vehicle;

[0071] For each sample vehicle, multiple sample images of the sample vehicle and the actual abnormal authentication probability are input into the constructed deep learning model to obtain the predicted abnormal authentication probability corresponding to the sample vehicle;

[0072] For each sample vehicle, determining a deviation value between an actual abnormal authentication probability of the sample vehicle and the predicted abnormal authentication probability;

[0073] If there is a sample vehicle whose corresponding deviation value is greater than the preset deviation threshold, adjust the relevant parameters in the deep learning model until the deviation value corresponding to each sample vehicle is less than or equal to the preset deviation threshold, determine that the deep learning model training is completed, and determine the trained deep learning model as the trained vehicle abnormality probability prediction model.

[0074] According to a third aspect of the present application, an electronic device is provided, comprising: a processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor and the memory communicate via the bus, and when the machine-readable instructions are executed by the processor, the steps of the vehicle authenticity authentication method as described above are performed.

[0075] According to a fourth aspect of the present application, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the vehicle authenticity authentication method as described above are executed.

[0076] The embodiments of the present application provide a method, an authentication device and a readable storage medium for authenticating the authenticity of a vehicle, which obtain multiple verification images of the vehicle to be authenticated; input the multiple verification images into a pre-trained vehicle abnormality probability prediction model to obtain the abnormal authentication probability of the vehicle to be authenticated; if it is detected that the abnormal authentication probability is greater than or equal to a preset threshold, the vehicle to be authenticated is determined to be an authentication abnormal vehicle.

[0077] In this way, by inputting multiple verification images of the vehicle to be authenticated into a pre-trained vehicle abnormality probability prediction model, the abnormal authentication probability of the vehicle to be authenticated output by the vehicle abnormality probability prediction model is compared with a preset threshold value. If the comparison result shows that the abnormal authentication probability is greater than or equal to the preset threshold value, the vehicle to be authenticated is determined to be an authenticated vehicle. The authenticity of the vehicle to be authenticated can be quickly identified with a high recognition accuracy, which helps to reduce the consumption of manual resources and time resources in the vehicle authentication process.

[0078] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, preferred embodiments are specifically cited below and described in detail with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0080] Figure 1 A schematic diagram of the architecture of an authentication system provided in an embodiment of the present application;

[0081] Figure 2 A flow chart of a vehicle authenticity authentication method provided in an embodiment of the present application;

[0082] Figure 3 A flow chart of a method for authenticating the authenticity of a vehicle provided in another embodiment of the present application;

[0083] Figure 4 This is one of the structural schematic diagrams of a vehicle authenticity authentication device provided in an embodiment of the present application;

[0084] Figure 5 A second structural diagram of a vehicle authenticity authentication device provided in an embodiment of the present application;

[0085] Figure 6 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0086] To make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. The components of the embodiments of the present application usually described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, each other embodiment obtained by those skilled in the art without making creative work belongs to the scope of protection of the present application.

[0087] In order to enable those skilled in the art to use the contents of this application, the following implementation method is given in conjunction with the specific application scenario of "authenticating the authenticity of the vehicle to be authenticated". For those skilled in the art, the general principles defined here can be applied to other embodiments and application scenarios without departing from the spirit and scope of this application. Although this application mainly describes inputting the verification image of the vehicle to be authenticated into a pre-trained vehicle abnormality probability prediction model, and verifying the authenticity of the vehicle to be authenticated based on the abnormal authentication probability output by the vehicle abnormality probability prediction model, it should be understood that this is only an exemplary embodiment.

[0088] It should be noted that the term "comprising" will be used in the embodiments of the present application to indicate the existence of the features declared thereafter, but does not exclude the addition of other features.

[0089] One aspect of the present application relates to an authentication system. The authentication system can input multiple verification images of the vehicle to be authenticated into a pre-trained vehicle abnormality probability prediction model, compare the abnormal authentication probability of the vehicle to be authenticated output by the vehicle abnormality probability prediction model with a preset threshold, and if the comparison result shows that the abnormal authentication probability is greater than or equal to the preset threshold, determine that the vehicle to be authenticated is an authenticated vehicle, and can quickly identify the authenticity of the vehicle to be authenticated, with high recognition accuracy, which helps to reduce the consumption of manual resources and time resources in the vehicle authentication process.

[0090] It is worth noting that before the present application was filed, the driver was required to provide an image of the vehicle's appearance during the online car-hailing registration process. Verifying the authenticity of the vehicle image provided by the driver is an important part of ensuring the authenticity of the driver's information. If the vehicle image complies with the rules, it will be reviewed and approved in a timely manner. If the vehicle image does not comply with the rules, the driver will be reminded in a timely manner until the driver provides a real vehicle image. This not only ensures the authenticity of the information provided by the driver, but also improves the driver's experience during the vehicle verification process. Therefore, how to accurately predict the authenticity of the vehicle through the vehicle image is an urgent problem to be solved. However, the purpose of this application is to provide a method for authenticating the authenticity of a vehicle, an authentication device, and a readable storage medium. The abnormal authentication probability of the vehicle to be authenticated is determined based on the verification image of the vehicle to be authenticated through a pre-trained vehicle abnormality probability prediction model, and the authenticity of the vehicle to be authenticated is determined according to the abnormal authentication probability. In the process of obtaining the verification image, the vehicle abnormality probability prediction model can quickly and accurately output the verification result of the authenticity of the vehicle to be authenticated, which helps to improve the efficiency and accuracy of vehicle authentication.

[0091] Figure 1 A schematic diagram of the architecture of an authentication system provided in an embodiment of the present application. The authentication system may be an online transportation service platform for transportation services such as taxis, chauffeur services, express trains, carpooling, bus services, driver rentals, or shuttle services, or any combination thereof. The authentication system may include one or more of a server 110, a network 120, a service provider 130, and a database 140.

[0092] In some embodiments, the server 110 may include a processor. The processor may process information and / or data related to the service request to perform one or more functions described in the present application. For example, the processor may wait to receive an authentication result of the authenticity of the vehicle to be authenticated based on a vehicle authentication request issued from the service provider 130 and multiple verification images of the vehicle to be authenticated. In some embodiments, the processor may include one or more processing cores (e.g., a single-core processor (S) or a multi-core processor (S)). By way of example only, the processor may include a central processing unit (CPU), an application specific integrated circuit (ASIC), an application specific instruction set processor (ASIP), a graphics processing unit (GPU), a physical processing unit (PPU), a digital signal processor (DSP), a field programmable gate array (FPGA), a programmable logic device (PLD), a controller, a microcontroller unit, a reduced instruction set computer (RISC), or a microprocessor, or any combination thereof.

[0093] In some embodiments, the device type corresponding to the service provider 130 can be a mobile device, such as a smart home device, a wearable device, a smart mobile device, a virtual reality device, or an augmented reality device, etc. It can also be a tablet computer, a laptop computer, or a built-in device in a motor vehicle, etc.

[0094] In some embodiments, the database 140 can be connected to the network 120 to communicate with one or more components in the route planning system (e.g., the server 110, the service provider 130, etc.). One or more components in the authentication system can access the data or instructions stored in the database 140 via the network 120. In some embodiments, the database 140 can be directly connected to one or more components in the route planning system, or the database 140 can also be part of the server 110.

[0095] The following is combined with the above Figure 1The contents described in the authentication system shown provide a detailed description of the vehicle authenticity authentication method provided in the embodiment of the present application.

[0096] See also Figure 2 , Figure 2 This is a flow chart of a vehicle authenticity authentication method provided in an embodiment of the present application. The method can be executed by a processor in an authentication system, and the specific execution process is:

[0097] S201. Acquire multiple verification images of the vehicle to be authenticated.

[0098] In this step, based on the image collection of the vehicle to be authenticated by the user of the vehicle to be authenticated, a plurality of verification images of the vehicle to be authenticated are acquired.

[0099] Here, the verification image of the vehicle to be authenticated may be a plurality of pictures captured by a user using a camera, and the pictures may be appearance images of the vehicle to be authenticated from multiple angles, interior images of the vehicle, etc.; the verification image may also be a video of the vehicle to be authenticated shot by a user using a camera over a period of time, and each frame in the video is captured, and the image corresponding to each frame is determined as the verification image of the vehicle to be authenticated.

[0100] When capturing the image frames, there may be cases where the image screenshot is unclear or a certain frame of image does not capture the image related to the vehicle to be authenticated. In this case, these unclear images and images that do not capture substantial content need to be deleted, and subsequent authentication of the vehicle to be authenticated will not be performed based on these images.

[0101] Here, a time range needs to be set for capturing verification images, because the verification image of the vehicle to be authenticated may not be received within the preset time range due to improper operation of the user when capturing the verification image. Since multiple pre-processed images may be captured to form a verification image in the process of capturing the verification image, the multiple pre-processed images can be used as multiple verification images of the vehicle to be authenticated. Similarly, for video capture of the vehicle to be authenticated, if the video of the vehicle to be authenticated is not captured within the preset time range, the video containing the vehicle to be authenticated taken after the camera is turned on can also be used as the captured video, and multiple verification images of the vehicle to be authenticated can be determined by capturing the captured video frame by frame.

[0102] S202: Input the plurality of verification images into a pre-trained vehicle abnormality probability prediction model to obtain an abnormal authentication probability of the vehicle to be authenticated.

[0103] In this step, the multiple verification images obtained in step S201 are input into a pre-trained vehicle abnormality probability prediction model, and the vehicle abnormality probability prediction model predicts the authentication abnormality probability that the vehicle to be authenticated has a problem.

[0104] Here, the input order of multiple verification images can be that after all the verification images are collected, all the verification images are input into the abnormal probability prediction model together, or they can be input into the abnormal probability prediction model in a certain order, wherein the input order can be the chronological order of the collection time of each verification image, or the same type of verification images can be input together according to the content of the image, for example, according to the content shown in the image, the appearance image of the vehicle to be authenticated is input together, the interior image of the vehicle to be authenticated is input together, etc.

[0105] Here, the abnormal authentication probability output by the vehicle abnormality probability prediction model can be a value less than 1 or any value greater than 1.

[0106] Here, a deep learning model can be selected as the basic model of the vehicle abnormality probability prediction model.

[0107] S203: If it is detected that the abnormal authentication probability is greater than or equal to a preset threshold, determine that the vehicle to be authenticated is an authentication abnormal vehicle.

[0108] In this step, if it is detected that the abnormal authentication probability output by the vehicle abnormal probability prediction model in step S202 is greater than or equal to a preset threshold, then the vehicle to be detected is considered to be an authentication abnormal vehicle.

[0109] Among them, the preset threshold can be determined based on the correspondence between the abnormal probability in the historical vehicle verification process and the authenticated abnormal vehicles. For example, in the previous multiple vehicle authenticity authentication processes, there is a probability value X. When the abnormal authentication probability of the vehicle to be authenticated is greater than X, a considerable portion of the vehicles are abnormal vehicles. At this time, this probability value X can be determined as the preset threshold.

[0110] The vehicle authenticity authentication method provided in the embodiment of the present application obtains multiple verification images of the vehicle to be authenticated; the multiple verification images are input into a pre-trained vehicle abnormality probability prediction model to obtain the abnormal authentication probability of the vehicle to be authenticated; if it is detected that the abnormal authentication probability is greater than or equal to a preset threshold, the vehicle to be authenticated is determined to be an authentication abnormal vehicle.

[0111] In this way, by inputting multiple verification images of the vehicle to be authenticated into a pre-trained vehicle abnormality probability prediction model, the abnormal authentication probability of the vehicle to be authenticated output by the vehicle abnormality probability prediction model is compared with a preset threshold value. If the comparison result shows that the abnormal authentication probability is greater than or equal to the preset threshold value, the vehicle to be authenticated is determined to be an authenticated vehicle. The authenticity of the vehicle to be authenticated can be quickly identified with a high recognition accuracy, which helps to reduce the consumption of manual resources and time resources in the vehicle authentication process.

[0112] See also Figure 3 , Figure 3 This is a flow chart of a method for authenticating the authenticity of a vehicle provided in another embodiment of the present application. The method can be executed by a processor in an authentication system, and the specific execution process is:

[0113] S301, obtaining multiple verification images of the vehicle to be authenticated.

[0114] S302: Input the plurality of verification images into a pre-trained vehicle abnormality probability prediction model to obtain an abnormal authentication probability of the vehicle to be authenticated.

[0115] S303: If it is detected that the abnormal authentication probability is greater than or equal to a preset threshold, the vehicle to be authenticated is determined to be an authentication abnormal vehicle.

[0116] S304: sending a verification result indicating that the vehicle to be verified is a verification abnormal vehicle to the user, and prompting the user to provide a verification image of a vehicle that matches the pre-registered vehicle information or a real vehicle.

[0117] In this step, after the authentication process of step S301 to step S303, it is determined that the vehicle to be authenticated is an authentication abnormal vehicle, the authentication result that the vehicle to be authenticated is an authentication abnormal vehicle is sent to the user, and the user is prompted to provide information that is the same as the vehicle information of the pre-registered vehicle or a verification image of the real vehicle, and the vehicle to be authenticated is verified again.

[0118] Here, vehicle authentication anomalies may include the following two situations: one is that the vehicle information is inconsistent with the vehicle information registered by the user when registering. The vehicle information for comparison may be information such as vehicle color, vehicle model, and license plate number. For example, the user's registered vehicle is a white seven-seater car, but the vehicle identified according to the verification image provided by the user is a red five-seater car. In this case, it can be determined that the vehicle to be authenticated is abnormal; the other is that the verification image provided by the user is not a real-time image, but may be a vehicle photo downloaded by the user from the Internet at will. The vehicle to be authenticated does not exist in real life. In this case, the vehicle to be authenticated can also be considered abnormal.

[0119] Here, the authentication result can be sent through the terminal where the user logs in to request authentication, and when the authentication failure result is sent, the reason for the authentication abnormality can also be sent accordingly, so that the user can provide the verification image again in a targeted manner.

[0120] For example, after the authentication process, it is determined that the vehicle to be authenticated is a model car that does not really exist. When sending the authentication result, the message that can be sent is "The vehicle is an authentication abnormal vehicle, please provide a verification image of the real vehicle."

[0121] Among them, the description of S301 to S303 can refer to the description of S201 to S203, and can achieve the same technical effect, which is not repeated here.

[0122] Furthermore, step S302 includes: inputting each verification image into the vehicle abnormality probability prediction model in sequence according to the image acquisition order, and obtaining the initial abnormality probability of each verification image in sequence according to the image input order; based on the initial abnormality probability of each verification image, determining the abnormal authentication probability of the vehicle to be authenticated.

[0123] In this step, each verification image is input into a pre-trained vehicle abnormality probability prediction model in sequence according to the order in which each verification image is collected. After each verification image is input into the vehicle abnormality probability prediction model, the vehicle abnormality probability prediction model will output an initial abnormality probability corresponding to the verification image. After the initial abnormality probabilities of the multiple verification images acquired are determined, the abnormal authentication probability of the vehicle to be authenticated is determined based on the initial abnormality probability corresponding to each verification image.

[0124] Here, the input of the verification image can be input into the vehicle abnormality probability prediction model as soon as a verification image is obtained, or it can be input into the vehicle abnormality probability prediction model in sequence according to the acquisition time of each verification image within a period of time after the verification image is acquired.

[0125] Furthermore, the abnormal authentication probability of the vehicle to be authenticated is determined based on the initial abnormal probability of each verification image, including: in the process of obtaining the initial abnormal probability of each verification image in turn, when the initial abnormal probability of the Nth verification image is obtained, if it is detected that the sum of the initial abnormal probabilities of the first N verification images is greater than or equal to the preset threshold, determining the sum of the initial abnormal probabilities of the first N verification images as the abnormal authentication probability of the vehicle to be authenticated.

[0126] In this step, in the process of obtaining the initial abnormality probability of each verification image in turn, after obtaining the initial abnormality probability of the Nth verification image, the sum of the N-1 initial abnormality probabilities corresponding to the N-1 verification images before the Nth verification image is added to the initial abnormality probability of the Nth verification image to determine the sum of the initial abnormality probabilities of the first N verification images, and detect whether the sum of the initial abnormality probabilities of the first N verification images is greater than a preset threshold. When it is detected that the sum of the initial abnormality probabilities of the first N verification images is greater than the preset threshold, the verification image is no longer input into the vehicle abnormality probability prediction model, and the sum of the initial abnormality probabilities of the first N verification images is used as the abnormal authentication probability of the vehicle to be authenticated.

[0127] Wherein, N is a positive integer greater than or equal to 1 and less than the total number of verification images.

[0128] Furthermore, the method of determining the abnormal authentication probability of the vehicle to be authenticated based on the initial abnormal probability of each verification image includes: in the process of obtaining the initial abnormal probability of each verification image in turn, when the initial abnormal probability of the Mth verification image is obtained, detecting that the authentication time from obtaining the initial abnormal probability of the first verification image to obtaining the initial abnormal probability of the Mth verification image is greater than or equal to a preset time threshold, and determining the sum of the initial abnormal probabilities of the first M verification images as the abnormal authentication probability of the vehicle to be authenticated; after inputting the multiple verification images into a pre-trained vehicle abnormal probability prediction model to obtain the abnormal authentication probability of the vehicle to be authenticated, the authentication method includes: if it is detected that the abnormal authentication probability is less than the preset threshold, determining that the vehicle to be authenticated is a successfully authenticated vehicle.

[0129] In this step, in the process of obtaining the initial abnormal probability of each verification image in turn, when the initial probability of the Mth verification image is obtained, it is determined that the time from obtaining the initial abnormal probability of the first verification image to the initial abnormal probability of the Mth verification image is greater than or equal to the preset time threshold. At this time, the sum of the initial abnormal probabilities of the first M verification images is determined to be the abnormal authentication probability of the vehicle to be authenticated, and it is detected whether the abnormal authentication probability determined at this time is less than the preset threshold. If it is detected that the abnormal authentication probability is less than the preset threshold, the vehicle to be authenticated is determined to be a successfully authenticated vehicle.

[0130] Here, the preset time threshold may be set according to a historical prediction process or according to the total number of verification images. After determining that the number of input verification images reaches a certain proportion of the total number and the authentication probability is representative, the preset time threshold may be set according to the processing time of each verification image. For example, there are 50 verification images in total. According to historical experience, the authentication probabilities determined by 80% of the verification images are representative, that is, the time after processing 40 verification images can be used as the preset time threshold. Assuming that the processing time for each verification image is 0.08 seconds, the preset time threshold can be approximately set to 3 seconds.

[0131] Furthermore, the initial abnormality probability corresponding to each verification image is determined by the following steps: performing feature extraction processing on the verification image and determining a characteristic thermal image of the verification image; based on the characteristic thermal image, determining a region of interest image of the verification image from the verification image; based on the region of interest image, determining the initial abnormality probability of the verification image.

[0132] In this step, feature extraction is performed on the verification image, and a characteristic thermal image of the verification image is determined based on at least one feature layer extracted from the verification image; based on the determined characteristic thermal image, an image of a region of interest in the verification image is determined from the verification image; and based on the region value obtained from the image of the region of interest and each preset, an initial abnormal probability of the verification image is determined.

[0133] Here, the characteristic thermal image may refer to a heatmap. Heatmap can use color changes to represent data information, and it can intuitively represent the size of the data value with a defined color depth. Data is often clustered based on abundance similarity between species or samples as needed, and the clustered data is represented on a heatmap. High-abundance and low-abundance species can be clustered in blocks, and the similarities and differences in community composition of multiple samples at each classification level can be reflected through color gradients and similarity.

[0134] Here, the region of interest image determined from the verification image can effectively propose areas that have no effect on the initial abnormality probability prediction and highlight areas that are helpful for the initial abnormality probability prediction, which can effectively reduce the amount of calculation for the initial abnormality probability prediction and help improve the efficiency of determining the initial abnormality probability.

[0135] Furthermore, the feature extraction process is performed on the verification image, and a characteristic thermal image of the verification image is determined, including: extracting a feature layer under each preset feature dimension from the verification image; determining an image category of the verification image and an initial thermal image of the verification image corresponding to the image category based on the extracted multiple feature layers and the preset proportion weight of each preset feature dimension; enlarging the initial thermal image to be consistent with the image size of the verification image, and determining the enlarged initial thermal image as the characteristic thermal image of the verification image.

[0136] In this step, feature extraction is performed on the verification image under each preset feature dimension, and multiple feature layers are extracted from the verification image, wherein one feature layer or multiple feature layers can be determined under each preset feature dimension, and the image category of the verification image and the initial thermal image corresponding to the image category are determined according to the preset proportion weight of each preset feature dimension. During the feature extraction process of the verification image, the image size of the verification image may change, that is, the size of the initial thermal image may be inconsistent with the size of the verification image. At this time, it is necessary to adjust the size of the initial thermal image, and determine the adjusted initial thermal image as the verification image.

[0137] The preset feature dimensions include but are not limited to the chromaticity of the image, the specific content of the image, etc.

[0138] Here, when determining the image category of the verification image, classification can be performed through a softmax classifier, and multiple feature layers can be input into the softmax classifier, and according to the pre-proportion weights corresponding to each preset feature dimension, the category probability corresponding to each feature layer after passing through the softmax classifier is obtained, wherein the sum of the category probabilities corresponding to each feature layer is 1; from the multiple category probabilities obtained, the category corresponding to the category probability with the largest probability value is determined to be the image category of the verification image.

[0139] Here, each image category corresponds to an initial thermal image. According to each image category, each initial thermal map presented is different. The difference lies in that, under different image categories, the weight coefficient corresponding to each preset feature dimension is different. For example, the weight coefficient of feature dimension A under image category 1 is 0.25, while the weight coefficient of feature dimension A under image category 2 is 0.35. After determining the image category corresponding to the verification image, each feature layer is weighted and summed according to the weight coefficient corresponding to each preset feature dimension under the category to obtain the initial thermal image of the corresponding verification image.

[0140] Here, when restoring the size of the initial thermal image, the size of the initial thermal image may be restored to be consistent with the image size of the verification image through a resize function.

[0141] Furthermore, based on the characteristic thermal image, determining the region of interest image of the verification image from the verification image includes: determining the regional characteristic value corresponding to each characteristic image region in the characteristic thermal image, and detecting whether each regional characteristic value is greater than a preset extraction threshold; extracting the verification image region corresponding to each characteristic image region whose regional characteristic value is greater than the preset extraction threshold in the verification image; and splicing each verification image region to generate the region of interest image of the verification image.

[0142] In this step, after obtaining the characteristic thermal image, the characteristic thermal image will be divided into multiple characteristic image areas according to the direction in which the image needs to be judged, and each characteristic image area corresponds to a corresponding regional characteristic value. It is detected whether each regional characteristic value is greater than a preset extraction threshold, and the characteristic image area corresponding to the regional characteristic value greater than the preset extraction threshold is determined. In the verification image, multiple verification image areas are determined according to each characteristic image area, each verification image area is extracted from the verification image, and each verification image area is spliced ​​to generate an image of the area of ​​interest.

[0143] Here, the setting of the preset extraction threshold can be set according to the data of the historical image extraction process. In the embodiment of the present application, the vehicle authentication abnormality probability is extracted, and the preset extraction threshold is set according to the area value of the abnormal feature area. The setting of the preset extraction threshold is trained according to the authentication requirements during the training process of the vehicle abnormality probability prediction model.

[0144] Here, for the correspondence between the characteristic image area and the verification image area, each characteristic image area can be numbered in the process of forming the characteristic image area, and the corresponding position of the verification image can be numbered at the same time. When determining the correspondence between the characteristic image area and the verification image area, a one-to-one correspondence can be made according to the number.

[0145] Furthermore, the initial abnormal probability of the verification image is determined based on the area of ​​interest image, including: determining the classification category of the area of ​​interest image based on the area value corresponding to each area of ​​interest included in the area of ​​interest image and the weight coefficient corresponding to each area value; and determining the initial abnormal probability of the verification image based on a preset probability value corresponding to the classification category.

[0146] In this step, the area of ​​interest image is classified according to the area value corresponding to each area of ​​interest included in the area of ​​interest image and the weight coefficient corresponding to each area value, the classification category of the area of ​​interest image is determined, and the preset probability value corresponding to the classification category is used to determine the initial abnormal probability of the verification image.

[0147] Here, the classification category may be an abnormality level of the authentication abnormality probability corresponding to the image of the region of interest, each abnormality level corresponds to an initial abnormality probability, and a higher abnormality level corresponds to a greater initial abnormality probability.

[0148] Furthermore, the vehicle abnormality probability prediction model is trained by the following steps:

[0149] (1) Obtain multiple sample images of multiple sample vehicles and the actual abnormal authentication probability corresponding to each sample vehicle.

[0150] In this step, multiple sample vehicles and multiple sample images corresponding to each sample vehicle are obtained, and the actual abnormal authentication probability corresponding to the sample vehicle when being authenticated needs to be obtained.

[0151] Here, all the sample vehicles are vehicles that have been authenticated in the historical vehicle authentication process, and the sample vehicles may include normal vehicles that have passed the authentication, and may also include abnormal vehicles that have been authenticated abnormally.

[0152] (2) For each sample vehicle, multiple sample images of the sample vehicle and the actual abnormal authentication probability are input into the constructed deep learning model to obtain the predicted abnormal authentication probability corresponding to the sample vehicle.

[0153] In this step, for each sample vehicle, multiple sample images of the sample vehicle are input into the constructed deep learning model in the order of historical acquisition. At the same time, the actual abnormal authentication probability corresponding to the sample vehicle is also input into the constructed deep learning model. Based on the input sample information, the predicted abnormal authentication probability of the sample vehicle output by the deep learning model is obtained.

[0154] (3) For each sample vehicle, determine the deviation value between the actual abnormal authentication probability of the sample vehicle and the predicted abnormal authentication probability.

[0155] In this step, according to step (2), multiple sample images corresponding to all sample vehicles and the actual abnormal authentication probability are input into the deep learning model, and the predicted abnormal authentication probability corresponding to each sample vehicle is obtained, and the deviation value between the corresponding actual abnormal authentication probability and the predicted abnormal authentication probability is calculated for each sample vehicle.

[0156] (4) If there is a sample vehicle whose corresponding deviation value is greater than a preset deviation threshold, adjust the relevant parameters in the deep learning model until the deviation value corresponding to each sample vehicle is less than or equal to the preset deviation threshold, determine that the deep learning model is trained, and determine the trained deep learning model as the trained vehicle abnormality probability prediction model.

[0157] In this step, it is detected whether the deviation value of each sample vehicle is greater than the preset deviation threshold. As long as it is determined that the deviation value corresponding to a sample vehicle is greater than the preset deviation threshold, it means that there is still an error in the prediction of the vehicle by the deep learning model and adjustment is required. The loss function is used to adjust all relevant parameters in the deep learning model, and after each adjustment of the relevant parameters is completed, steps (1) to (3) are repeated until the deviation value corresponding to each sample vehicle is less than or equal to the preset deviation threshold. It is determined that the deep learning model has been trained, and the trained deep learning model is determined as the trained vehicle abnormality probability prediction model.

[0158] Here, taking the example of constructing the deep learning model, where the deep learning model includes a heatmap extraction layer, a ROI extraction layer, and a ROI discrimination layer, when adjusting the relevant parameters in the deep learning model, it is necessary to uniformly adjust the parameters of each layer included in the deep learning model. The adjustable parameters include but are not limited to the weight coefficient of each preset feature dimension in the heatmap extraction layer, the preset extraction parameters of the ROI extraction layer, etc.

[0159] The method for authenticating the authenticity of a vehicle provided in an embodiment of the present application obtains multiple verification images of a vehicle to be authenticated; inputs the multiple verification images into a pre-trained vehicle abnormality probability prediction model to obtain an abnormal authentication probability of the vehicle to be authenticated; if it is detected that the abnormal authentication probability is greater than or equal to a preset threshold, determines that the vehicle to be authenticated is an authentication abnormal vehicle; sends a authentication result that the vehicle to be authenticated is an authentication abnormal vehicle to a user, prompting the user to provide a verification image of a vehicle that matches the pre-registered vehicle information or a real existing vehicle.

[0160] In this way, by inputting multiple verification images of the vehicle to be authenticated into a pre-trained vehicle abnormality probability prediction model, the abnormal authentication probability of the vehicle to be authenticated output by the vehicle abnormality probability prediction model is compared with a preset threshold value. If the comparison result shows that the abnormal authentication probability is greater than or equal to the preset threshold value, the vehicle to be authenticated is determined to be an authenticated vehicle. The authenticity of the vehicle to be authenticated can be quickly identified with a high recognition accuracy, which helps to reduce the consumption of manual resources and time resources in the vehicle authentication process.

[0161] See also Figure 4 , Figure 5 , Figure 4 This is one of the structural schematic diagrams of a vehicle authenticity authentication device provided in an embodiment of the present application. Figure 5 This is a second structural diagram of a vehicle authenticity authentication device provided in an embodiment of the present application. Figure 4 As shown in , the authentication device 400 includes:

[0162] The image acquisition module 410 is used to acquire multiple verification images of the vehicle to be authenticated.

[0163] The probability determination module 420 is used to input the multiple verification images into a pre-trained vehicle abnormality probability prediction model to obtain the abnormal authentication probability of the vehicle to be authenticated.

[0164] The abnormal vehicle determination module 430 is configured to determine that the vehicle to be authenticated is an authentication abnormal vehicle if it is detected that the abnormal authentication probability is greater than or equal to a preset threshold.

[0165] Further, such as Figure 5 As shown, the authentication device 400 further includes a user prompt module 440, and the user prompt module 440 is used to:

[0166] The authentication result that the vehicle to be authenticated is an authentication abnormal vehicle is sent to the user, and the user is prompted to provide a verification image of a vehicle that matches the pre-registered vehicle information or a real vehicle.

[0167] Further, such as Figure 5 As shown, the authentication device 400 further includes a model training module 450, and the model training module 450 is used to:

[0168] Obtain multiple sample images of multiple sample vehicles and the actual abnormal authentication probability corresponding to each sample vehicle;

[0169] For each sample vehicle, multiple sample images of the sample vehicle and the actual abnormal authentication probability are input into the constructed deep learning model to obtain the predicted abnormal authentication probability corresponding to the sample vehicle;

[0170] For each sample vehicle, determining a deviation value between an actual abnormal authentication probability of the sample vehicle and the predicted abnormal authentication probability;

[0171] If there is a sample vehicle whose corresponding deviation value is greater than the preset deviation threshold, adjust the relevant parameters in the deep learning model until the deviation value corresponding to each sample vehicle is less than or equal to the preset deviation threshold, determine that the deep learning model training is completed, and determine the trained deep learning model as the trained vehicle abnormality probability prediction model.

[0172] Furthermore, when the probability determination module 420 is used to input the plurality of verification images into a pre-trained vehicle abnormality probability prediction model to obtain the abnormal authentication probability of the vehicle to be authenticated, the probability determination module 420 is used to:

[0173] Input each verification image into the vehicle abnormality probability prediction model in sequence according to the image acquisition order, and obtain the initial abnormality probability of each verification image in sequence according to the image input order;

[0174] Based on the initial abnormal probability of each verification image, the abnormal authentication probability of the vehicle to be authenticated is determined.

[0175] Furthermore, when the probability determination module 420 is used to determine the abnormal authentication probability of the vehicle to be authenticated based on the initial abnormal probability of each verification image, the probability determination module 420 is used to:

[0176] In the process of obtaining the initial abnormality probability of each verification image in turn, when the initial abnormality probability of the Nth verification image is obtained, if it is detected that the sum of the initial abnormality probabilities of the first N verification images is greater than or equal to the preset threshold, the sum of the initial abnormality probabilities of the first N verification images is determined to be the abnormal authentication probability of the vehicle to be authenticated.

[0177] Furthermore, when the probability determination module 420 is used to determine the abnormal authentication probability of the vehicle to be authenticated based on the initial abnormal probability of each verification image, the probability determination module 420 is used to:

[0178] In the process of obtaining the initial abnormal probability of each verification image in sequence, when the initial abnormal probability of the Mth verification image is obtained, it is detected that the authentication time from obtaining the initial abnormal probability of the first verification image to obtaining the initial abnormal probability of the Mth verification image is greater than or equal to the preset time threshold, and the sum of the initial abnormal probabilities of the first M verification images is determined as the abnormal authentication probability of the vehicle to be authenticated;

[0179] Further, such as Figure 5 As shown, the authentication device 400 further includes an authentication success module 460, and the authentication success module 460 is used to:

[0180] If it is detected that the abnormal authentication probability is less than a preset threshold, it is determined that the vehicle to be authenticated is a successfully authenticated vehicle.

[0181] Furthermore, the probability determination module 420 is used to determine the initial abnormality probability corresponding to each verification image through the following steps:

[0182] Performing feature extraction processing on the verification image and determining a feature thermal image of the verification image;

[0183] Based on the characteristic thermal image, determining an image of a region of interest of the verification image from the verification image;

[0184] Based on the region of interest image, an initial abnormality probability of the verification image is determined.

[0185] Furthermore, when the probability determination module 420 is used to perform feature extraction processing on the verification image and determine the characteristic thermal image of the verification image, the probability determination module 420 is used to:

[0186] Extract feature layers under each preset feature dimension from the verification image;

[0187] Based on the extracted multiple feature layers and the preset weight of each preset feature dimension, determine the image category of the verification image and the initial thermal image of the verification image corresponding to the image category;

[0188] The initial thermal image is enlarged to be consistent with the image size of the verification image, and the enlarged initial thermal image is determined as the characteristic thermal image of the verification image.

[0189] Further, when the probability determination module 420 is used to determine the region of interest image of the verification image from the verification image based on the characteristic thermal image, the probability determination module 420 is used to:

[0190] Determine a regional characteristic value corresponding to each characteristic image region in the characteristic thermal image, and detect whether each regional characteristic value is greater than a preset extraction threshold;

[0191] Extracting the verification image region corresponding to each characteristic image region having a region characteristic value greater than a preset extraction threshold in the verification image;

[0192] Each verification image region is concatenated to generate a region of interest image of the verification image.

[0193] Furthermore, the probability determination module 420 is used to determine the initial abnormal probability of the verification image based on the region of interest image. The probability determination module 420 is used to:

[0194] Determining a classification category of the region of interest image based on a region value corresponding to each region of interest included in the region of interest image and a weight coefficient corresponding to each region value;

[0195] Based on the preset probability value corresponding to the classification category, an initial abnormal probability of the verification image is determined.

[0196] The vehicle authenticity authentication device provided in the embodiment of the present application obtains multiple verification images of the vehicle to be authenticated; inputs the multiple verification images into a pre-trained vehicle abnormality probability prediction model to obtain the abnormal authentication probability of the vehicle to be authenticated; if it is detected that the abnormal authentication probability is greater than or equal to a preset threshold, the vehicle to be authenticated is determined to be an authentication abnormal vehicle.

[0197] In this way, by inputting multiple verification images of the vehicle to be authenticated into a pre-trained vehicle abnormality probability prediction model, the abnormal authentication probability of the vehicle to be authenticated output by the vehicle abnormality probability prediction model is compared with a preset threshold value. If the comparison result shows that the abnormal authentication probability is greater than or equal to the preset threshold value, the vehicle to be authenticated is determined to be an authenticated vehicle. The authenticity of the vehicle to be authenticated can be quickly identified with a high recognition accuracy, which helps to reduce the consumption of manual resources and time resources in the vehicle authentication process.

[0198] See also Figure 6 , Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Figure 6 As shown in , the electronic device 600 includes a processor 610 , a memory 620 and a bus 630 .

[0199] The memory 620 stores machine-readable instructions executable by the processor 610. When the electronic device 600 is running, the processor 610 communicates with the memory 620 via the bus 630. When the machine-readable instructions are executed by the processor 610, the above-mentioned Figure 2 as well as Figure 3 The specific implementation of the steps of the vehicle authenticity authentication method in the method embodiment shown can be found in the method embodiment, and will not be repeated here.

[0200] The present application also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the computer program can execute the above-mentioned Figure 2 as well as Figure 3 The specific implementation of the steps of the vehicle authenticity authentication method in the method embodiment shown can be found in the method embodiment, and will not be repeated here.

[0201] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0202] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of devices or units can be electrical, mechanical or other forms.

[0203] 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 on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0204] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0205] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium that is executable by a processor. Based on this understanding, the technical solution of the present application can essentially be embodied in the form of a software product, or in other words, the part that contributes to the prior art or the part of the technical solution. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0206] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present application, which are used to illustrate the technical solutions of the present application, rather than to limit them. The protection scope of the present application is not limited thereto. Although the present application is described in detail with reference to the above-mentioned embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-mentioned embodiments within the technical scope disclosed in the present application, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A method for authenticating the authenticity of a vehicle, characterized in that: The authentication method comprises: Acquire multiple verification images of the vehicle to be authenticated; According to the image acquisition order, each verification image is input into the vehicle abnormality probability prediction model in turn, and the initial abnormality probability of each verification image is obtained in turn according to the image input order; Determining the abnormal authentication probability of the vehicle to be authenticated based on the initial abnormal probability of each verification image; If it is detected that the abnormal authentication probability is greater than or equal to a preset threshold, the vehicle to be authenticated is determined to be an authentication abnormal vehicle, wherein the abnormal authentication probability of the vehicle to be authenticated based on the initial abnormal probability of each verification image is determined, including: In the process of obtaining the initial abnormality probability of each verification image in sequence, when the initial abnormality probability of the Nth verification image is obtained, if it is detected that the sum of the initial abnormality probabilities of the first N verification images is greater than or equal to the preset threshold, the sum of the initial abnormality probabilities of the first N verification images is determined as the abnormal authentication probability of the vehicle to be authenticated; or In the process of obtaining the initial abnormal probability of each verification image in sequence, when the initial abnormal probability of the Mth verification image is obtained, it is detected that the authentication time from obtaining the initial abnormal probability of the first verification image to obtaining the initial abnormal probability of the Mth verification image is greater than or equal to the preset time threshold, and the sum of the initial abnormal probabilities of the first M verification images is determined as the abnormal authentication probability of the vehicle to be authenticated; If it is detected that the abnormal authentication probability is less than the preset threshold, it is determined that the vehicle to be authenticated is a successfully authenticated vehicle.

2. The authentication method according to claim 1, characterized in that: The initial abnormality probability corresponding to each verification image is determined by the following steps: Performing feature extraction processing on the verification image and determining a feature thermal image of the verification image; Based on the characteristic thermal image, determining a region of interest image of the verification image from the verification image; Based on the region of interest image, an initial abnormality probability of the verification image is determined.

3. The authentication method according to claim 2, characterized in that: The step of performing feature extraction processing on the verification image and determining a feature thermal image of the verification image includes: Extracting a feature layer under each preset feature dimension from the verification image; Based on the extracted multiple feature layers and the preset proportion weight of each preset feature dimension, determine the image category of the verification image and the initial thermal image of the verification image corresponding to the image category; The initial thermal image is enlarged to be consistent with the image size of the verification image, and the enlarged initial thermal image is determined as a characteristic thermal image of the verification image.

4. The authentication method according to claim 2, characterized in that: The step of determining the region of interest image of the verification image from the verification image based on the characteristic thermal image comprises: Determine a regional characteristic value corresponding to each characteristic image region in the characteristic thermal image, and detect whether each regional characteristic value is greater than a preset extraction threshold; Extracting the verification image region corresponding to each characteristic image region having a region characteristic value greater than the preset extraction threshold in the verification image; Each verification image region is stitched together to generate an interest region image of the verification image.

5. The authentication method according to claim 2, characterized in that: The determining, based on the region of interest image, an initial abnormality probability of the verification image includes: Determining a classification category of the region of interest image based on a region value corresponding to each region of interest included in the region of interest image and a weight coefficient corresponding to each region value; Based on the preset probability value corresponding to the classification category, an initial abnormal probability of the verification image is determined.

6. The authentication method according to claim 1, characterized in that: After the abnormal authentication probability is detected to be greater than or equal to a preset threshold and the vehicle to be authenticated is determined to be an authentication abnormal vehicle, the authentication method further includes: The authentication result that the vehicle to be authenticated is an authentication abnormal vehicle is sent to the user, and the user is prompted to provide a verification image of a vehicle that matches the pre-registered vehicle information or a real vehicle.

7. The authentication method according to claim 1, characterized in that: The vehicle abnormality probability prediction model is trained by the following steps: Obtain multiple sample images of multiple sample vehicles and the actual abnormal authentication probability corresponding to each sample vehicle; For each sample vehicle, multiple sample images of the sample vehicle and the actual abnormal authentication probability are input into the constructed deep learning model to obtain the predicted abnormal authentication probability corresponding to the sample vehicle; For each of the sample vehicles, determining a deviation value between the actual abnormal authentication probability and the predicted abnormal authentication probability of the sample vehicle; If there is a sample vehicle whose corresponding deviation value is greater than the preset deviation threshold, adjust the relevant parameters in the deep learning model until the deviation value corresponding to each sample vehicle is less than or equal to the preset deviation threshold, determine that the deep learning model training is completed, and determine the trained deep learning model as the trained vehicle abnormality probability prediction model.

8. A vehicle authenticity authentication device, characterized in that: The authentication device comprises: An image acquisition module, used to acquire multiple verification images of the vehicle to be authenticated; A probability determination module, for sequentially inputting each verification image into the vehicle abnormality probability prediction model according to the image acquisition order, and sequentially obtaining the initial abnormality probability of each verification image according to the image input order; and determining the abnormal authentication probability of the vehicle to be authenticated based on the initial abnormality probability of each verification image; The abnormal vehicle determination module is used to determine that the vehicle to be authenticated is an authentication abnormal vehicle if it is detected that the abnormal authentication probability is greater than or equal to a preset threshold, wherein when the probability determination module is used to determine the abnormal authentication probability of the vehicle to be authenticated based on the initial abnormal probability of each verification image, the probability determination module is also used to: In the process of obtaining the initial abnormality probability of each verification image in sequence, when the initial abnormality probability of the Nth verification image is obtained, if it is detected that the sum of the initial abnormality probabilities of the first N verification images is greater than or equal to the preset threshold, the sum of the initial abnormality probabilities of the first N verification images is determined as the abnormal authentication probability of the vehicle to be authenticated; or In the process of obtaining the initial abnormal probability of each verification image in sequence, when the initial abnormal probability of the Mth verification image is obtained, it is detected that the authentication time from obtaining the initial abnormal probability of the first verification image to obtaining the initial abnormal probability of the Mth verification image is greater than or equal to the preset time threshold, and the sum of the initial abnormal probabilities of the first M verification images is determined as the abnormal authentication probability of the vehicle to be authenticated; The authentication device further comprises an authentication success module, wherein the authentication success module is used to: If it is detected that the abnormal authentication probability is less than the preset threshold, it is determined that the vehicle to be authenticated is a successfully authenticated vehicle.

9. The authentication device according to claim 8, characterized in that: The probability determination module is used to determine the initial abnormality probability corresponding to each verification image through the following steps: Performing feature extraction processing on the verification image and determining a feature thermal image of the verification image; Based on the characteristic thermal image, determining a region of interest image of the verification image from the verification image; Based on the region of interest image, an initial abnormality probability of the verification image is determined.

10. The authentication device according to claim 9, characterized in that: When the probability determination module is used to perform feature extraction processing on the verification image and determine the characteristic thermal image of the verification image, the probability determination module is used to: Extracting a feature layer under each preset feature dimension from the verification image; Based on the extracted multiple feature layers and the preset proportion weight of each preset feature dimension, determine the image category of the verification image and the initial thermal image of the verification image corresponding to the image category; The initial thermal image is enlarged to be consistent with the image size of the verification image, and the enlarged initial thermal image is determined as a characteristic thermal image of the verification image.

11. The authentication device according to claim 9, characterized in that: When the probability determination module is used to determine the region of interest image of the verification image from the verification image based on the characteristic thermal image, the probability determination module is used to: Determine a regional characteristic value corresponding to each characteristic image region in the characteristic thermal image, and detect whether each regional characteristic value is greater than a preset extraction threshold; Extracting the verification image region corresponding to each characteristic image region having a region characteristic value greater than the preset extraction threshold in the verification image; Each verification image region is stitched together to generate an interest region image of the verification image.

12. The authentication device according to claim 9, characterized in that: The probability determination module is used to determine the initial abnormal probability of the verification image based on the region of interest image, and the probability determination module is used to: Determining a classification category of the region of interest image based on a region value corresponding to each region of interest included in the region of interest image and a weight coefficient corresponding to each region value; Based on the preset probability value corresponding to the classification category, an initial abnormal probability of the verification image is determined.

13. The authentication device according to claim 8, characterized in that: The authentication device further comprises a user prompting module, wherein the user prompting module is used for: The authentication result that the vehicle to be authenticated is an authentication abnormal vehicle is sent to the user, and the user is prompted to provide a verification image of a vehicle that matches the pre-registered vehicle information or a real vehicle.

14. The authentication device according to claim 8, characterized in that: The authentication device further comprises a model training module, wherein the model training module is used to: Obtain multiple sample images of multiple sample vehicles and the actual abnormal authentication probability corresponding to each sample vehicle; For each sample vehicle, multiple sample images of the sample vehicle and the actual abnormal authentication probability are input into the constructed deep learning model to obtain the predicted abnormal authentication probability corresponding to the sample vehicle; For each of the sample vehicles, determining a deviation value between the actual abnormal authentication probability and the predicted abnormal authentication probability of the sample vehicle; If there is a sample vehicle whose corresponding deviation value is greater than the preset deviation threshold, adjust the relevant parameters in the deep learning model until the deviation value corresponding to each sample vehicle is less than or equal to the preset deviation threshold, determine that the deep learning model training is completed, and determine the trained deep learning model as the trained vehicle abnormality probability prediction model.

15. An electronic device, characterized in that: include: A processor, a storage medium and a bus, wherein the storage medium stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor and the storage medium communicate via the bus, and the processor executes the machine-readable instructions to perform the steps of the vehicle authenticity authentication method as described in any one of claims 1 to 7.

16. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the vehicle authenticity authentication method as claimed in any one of claims 1 to 7 are executed.

Images