Vehicle authentication method, device, electronic device and storage medium

Abstract

The present application provides a vehicle authentication method, device, electronic device and storage medium, the method comprising: obtaining a video image of a target vehicle matching a target start instruction, the target start instruction comprising target authentication information; determining the confidence of each frame of the target image under multiple vehicle perspectives based on the vehicle features corresponding to the target vehicle in each frame of the target image; determining the information to be authenticated corresponding to the video image based on the confidence of each frame of the target image under multiple vehicle perspectives; if the information to be authenticated matches the target authentication information, determining that the target vehicle is authenticated. The present application can ensure that the authenticated target vehicle is a real vehicle to the greatest extent, improve the reliability of the platform authentication results, and ensure the service security of the platform.

Description

Technical Field

[0001] The present application relates to the technical field of vehicle authentication, and in particular to a vehicle authentication method, device, electronic device and storage medium. Background Art

[0002] In the field of online ride-hailing, online ride-hailing platforms provide travel services to passengers based on the driver's registered vehicles, that is, they allocate service vehicles to passengers' travel requests so as to provide travel services to passengers based on the service vehicles.

[0003] In practice, there may be a situation where the actual vehicle booked by the passenger is inconsistent with the service vehicle displayed by the online ride-hailing platform, which will cause security issues for the online ride-hailing platform. Based on this, the online ride-hailing platform requires drivers to authenticate the corresponding service vehicle before providing travel services. At present, the authentication method of the online ride-hailing platform for service vehicles cannot ensure that the authenticated service vehicle is a real vehicle, which makes the reliability of the platform authentication result poor, and thus leads to low service security of the platform. Summary of the invention

[0004] In view of this, the purpose of the present application is to provide a vehicle authentication method, device, electronic device and storage medium, which can maximize the guarantee that the authenticated target vehicle is a real vehicle by extracting the information to be authenticated from the video image of the target vehicle that matches the target start command and authenticating the target vehicle, thereby improving the reliability of the platform authentication results and ensuring the service security of the platform.

[0005] In a first aspect, an embodiment of the present application provides a vehicle authentication method, the method comprising:

[0006] Acquire a video image of a target vehicle matching a target start instruction, wherein the target start instruction includes target authentication information; wherein the video image includes multiple frames of target images;

[0007] Based on the vehicle features corresponding to the target vehicle in each frame of the target image, determining the confidence of each frame of the target image under multiple vehicle perspectives;

[0008] Determining the information to be authenticated corresponding to the video image based on the confidence of each frame of the target image under multiple vehicle perspectives;

[0009] If the information to be authenticated matches the target authentication information, it is determined that the target vehicle has passed authentication.

[0010] In a possible implementation, determining the confidence of each frame of the target image under multiple vehicle perspectives based on the vehicle features corresponding to the target vehicle in each frame of the target image includes:

[0011] Extracting vehicle features corresponding to the target vehicle in each frame of the target image included in the video image;

[0012] For each frame of target image included in the video image, the vehicle features corresponding to the target vehicle in the frame of target image are processed based on the first processing parameters to obtain the confidence of the frame of target image under multiple vehicle viewing angles.

[0013] In a possible implementation manner, extracting the vehicle features corresponding to the target vehicle in each frame of the target image included in the video image includes:

[0014] For each frame of target image included in the video image, extracting image features corresponding to the frame of target image;

[0015] Based on the image features corresponding to the frame target image, the vehicle features corresponding to the target vehicle in the frame target image are determined.

[0016] In a possible implementation manner, determining the vehicle features corresponding to the target vehicle in the frame target image based on the image features corresponding to the frame target image includes:

[0017] Performing convolution processing on the image features corresponding to the target image of the frame based on the target convolution kernel to obtain the deep image features corresponding to the target image of the frame;

[0018] The deep image features corresponding to the frame target image are processed to determine the vehicle features corresponding to the target vehicle in the frame target image.

[0019] In a possible implementation manner, processing the deep image features corresponding to the frame target image to determine the vehicle features corresponding to the target vehicle in the frame target image includes:

[0020] Processing the deep image feature corresponding to the frame target image based on the second processing parameter, and determining the first vehicle attribute feature corresponding to the target vehicle in the frame target image based on the obtained processing result;

[0021] Processing the deep image feature corresponding to the frame target image based on the third processing parameter to obtain a second vehicle attribute feature corresponding to the target vehicle in the frame target image;

[0022] Based on the first vehicle attribute feature and the second vehicle attribute feature, a vehicle feature corresponding to the target vehicle in the frame target image is determined.

[0023] In a possible implementation, determining the first vehicle attribute feature corresponding to the target vehicle in the frame target image based on the obtained processing result includes:

[0024] Determining first attribute information of the vehicle frame based on a first position of a vehicle frame corresponding to the target vehicle in the frame target image;

[0025] Determine the third positions of the wheels in the vehicle frame based on the first position of the vehicle frame and the second positions of the wheels of the target vehicle in the target image frame.

[0026] The first vehicle attribute feature corresponding to the target vehicle in the target image of the frame is determined according to the first attribute information of the vehicle frame, the wheel types corresponding to the wheels in the target vehicle and the relative positions of the wheels in the vehicle frame.

[0027] In a possible implementation manner, the first processing parameter, the second processing parameter and the third processing parameter are determined in the following manner:

[0028] Acquire a sample video image corresponding to the target vehicle;

[0029] Taking the image feature of each frame of the sample image in the sample video image as the first input feature and taking the second attribute information of the target vehicle corresponding to each frame of the sample image in the sample video image as the output, training the second processing parameter;

[0030] The first processing parameter and the third processing parameter are trained by taking the image features included in each frame of the sample video image as the second input features, taking the vehicle features corresponding to the target vehicle in each frame of the sample image as the third input features corresponding to the third processing parameters, and taking the authentication information of the target vehicle corresponding to the sample video image as the output.

[0031] In a possible implementation, determining the information to be authenticated corresponding to the video image based on the confidence of each frame of the target image under multiple vehicle perspectives includes:

[0032] Determine a starting vehicle perspective corresponding to the first frame target image based on the confidence of the first frame target image under multiple vehicle perspectives respectively; wherein the multiple vehicle perspectives are arranged in a cyclic manner;

[0033] Acquire multiple paths corresponding to the video image, which start from the starting vehicle perspective and pass through other vehicle perspectives and then return to the starting vehicle perspective; wherein each path includes multiple frames of target images in the video image, and each frame of the target image in the path corresponds to a vehicle perspective;

[0034] Selecting a target path from the multiple paths based on the confidence of each frame of the target image under multiple vehicle perspectives;

[0035] According to the target path, information to be authenticated corresponding to the video image is determined.

[0036] In a possible implementation manner, taking the starting vehicle perspective as the starting point includes a time sequence of two vehicle perspectives passing through other vehicle perspectives and returning to the starting vehicle perspective again;

[0037] For any vehicle perspective sequence, the vehicle perspective corresponding to each frame of the target image is the first vehicle perspective corresponding to the previous frame of the target image or the next second vehicle perspective after the first vehicle perspective in the vehicle perspective sequence.

[0038] In a possible implementation, selecting a target path from the multiple paths based on the confidence of each frame of the target image under multiple vehicle perspectives includes:

[0039] Determining a total confidence corresponding to each of the multiple paths based on the confidence of each frame of the target image under multiple vehicle perspectives;

[0040] According to the total confidence corresponding to each path in the multiple paths, a target path with the maximum corresponding total confidence is selected from the multiple paths.

[0041] In a second aspect, an embodiment of the present application further provides a vehicle authentication device, the device comprising:

[0042] A first acquisition module is used to acquire a video image of a target vehicle matching a target start instruction, wherein the target start instruction includes target authentication information; wherein the video image includes multiple frames of target images;

[0043] A first determination module is used to determine the confidence of each frame of the target image under multiple vehicle perspectives based on the vehicle features corresponding to the target vehicle in each frame of the target image;

[0044] A second determination module is used to determine the information to be authenticated corresponding to the video image based on the confidence of each frame of the target image under multiple vehicle perspectives;

[0045] The third determination module is used to determine that the target vehicle is authenticated if the information to be authenticated matches the target authentication information.

[0046] In a possible implementation, the first determination module determines the confidence of each frame of the target image under multiple vehicle perspectives based on the vehicle features corresponding to the target vehicle in each frame of the target image, including:

[0047] Extracting vehicle features corresponding to the target vehicle in each frame of the target image included in the video image;

[0048] For each frame of target image included in the video image, the vehicle features corresponding to the target vehicle in the frame of target image are processed based on the first processing parameters to obtain the confidence of the frame of target image under multiple vehicle viewing angles.

[0049] In a possible implementation manner, the first determination module extracts vehicle features corresponding to the target vehicle in each frame of the target image included in the video image, including:

[0050] For each frame of target image included in the video image, extracting image features corresponding to the frame of target image;

[0051] Based on the image features corresponding to the frame target image, the vehicle features corresponding to the target vehicle in the frame target image are determined.

[0052] In a possible implementation manner, the first determining module determines the vehicle features corresponding to the target vehicle in the frame target image based on the image features corresponding to the frame target image, including:

[0053] Performing convolution processing on the image features corresponding to the target image of the frame based on the target convolution kernel to obtain the deep image features corresponding to the target image of the frame;

[0054] The deep image features corresponding to the frame target image are processed to determine the vehicle features corresponding to the target vehicle in the frame target image.

[0055] In a possible implementation manner, the first determining module processes the deep image features corresponding to the frame target image to determine the vehicle features corresponding to the target vehicle in the frame target image, including:

[0056] Processing the deep image feature corresponding to the frame target image based on the second processing parameter, and determining the first vehicle attribute feature corresponding to the target vehicle in the frame target image based on the obtained processing result;

[0057] Processing the deep image feature corresponding to the frame target image based on the third processing parameter to obtain a second vehicle attribute feature corresponding to the target vehicle in the frame target image;

[0058] Based on the first vehicle attribute feature and the second vehicle attribute feature, a vehicle feature corresponding to the target vehicle in the frame target image is determined.

[0059] In a possible implementation manner, the first determining module determines, based on the obtained processing result, a first vehicle attribute feature corresponding to the target vehicle in the frame target image, including:

[0060] Determining first attribute information of the vehicle frame based on a first position of a vehicle frame corresponding to the target vehicle in the frame target image;

[0061] Determine the third positions of the wheels in the vehicle frame based on the first position of the vehicle frame and the second positions of the wheels of the target vehicle in the target image frame.

[0062] The first vehicle attribute feature corresponding to the target vehicle in the target image of the frame is determined according to the first attribute information of the vehicle frame, the wheel types corresponding to the wheels in the target vehicle and the relative positions of the wheels in the vehicle frame.

[0063] In a possible implementation, the device further includes:

[0064] A second acquisition module is used to acquire a sample video image corresponding to the target vehicle;

[0065] A first training module is used to train the second processing parameter by taking the image feature of each frame of the sample image in the sample video image as the first input feature and taking the second attribute information of the target vehicle corresponding to each frame of the sample image in the sample video image as the output;

[0066] The second training module is used to train the first processing parameter and the third processing parameter using the image features included in each frame of the sample image in the sample video image as the second input feature, the vehicle features corresponding to the target vehicle in each frame of the sample image as the third input feature corresponding to the third processing parameter, and the authentication information of the target vehicle corresponding to the sample video image as the output.

[0067] In a possible implementation manner, the second determination module determines the information to be authenticated corresponding to the video image based on the confidence of each frame of the target image under multiple vehicle perspectives, including:

[0068] Determine a starting vehicle perspective corresponding to the first frame target image based on the confidence of the first frame target image under multiple vehicle perspectives respectively; wherein the multiple vehicle perspectives are arranged in a cyclic manner;

[0069] Acquire multiple paths corresponding to the video image, which start from the starting vehicle perspective and pass through other vehicle perspectives and then return to the starting vehicle perspective; wherein each path includes multiple frames of target images in the video image, and each frame of the target image in the path corresponds to a vehicle perspective;

[0070] Selecting a target path from the multiple paths based on the confidence of each frame of the target image under multiple vehicle perspectives;

[0071] According to the target path, information to be authenticated corresponding to the video image is determined.

[0072] In a possible implementation manner, taking the starting vehicle perspective as the starting point includes a time sequence of two vehicle perspectives passing through other vehicle perspectives and returning to the starting vehicle perspective again;

[0073] For any vehicle perspective sequence, the vehicle perspective corresponding to each frame of the target image is the first vehicle perspective corresponding to the previous frame of the target image or the next second vehicle perspective after the first vehicle perspective in the vehicle perspective sequence.

[0074] In a possible implementation manner, the second determination module selects a target path from the multiple paths based on the confidence of each frame of the target image under multiple vehicle perspectives, including:

[0075] Determining a total confidence corresponding to each of the multiple paths based on the confidence of each frame of the target image under multiple vehicle perspectives;

[0076] According to the total confidence corresponding to each path in the multiple paths, a target path with the maximum corresponding total confidence is selected from the multiple paths.

[0077] In a third aspect, an embodiment of the present application further provides an electronic device, comprising: 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 communicates with the storage medium through the bus, and the processor executes the machine-readable instructions to perform the steps of the vehicle authentication method as described in any one of the first aspects.

[0078] In a fourth aspect, an embodiment of the present application further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the vehicle authentication method described in the first aspect are executed.

[0079] The embodiment of the present application provides a vehicle authentication method, device, electronic device and storage medium, which obtains a video image of a target vehicle matching a target start instruction, determines the confidence of each frame of the target image under multiple vehicle perspectives based on the vehicle features corresponding to the target vehicle in each frame of the target image included in the video image, and then determines the information to be authenticated corresponding to the video image based on the confidence of each frame of the target image under multiple vehicle perspectives; finally, the target vehicle is authenticated based on the matching result between the information to be authenticated and the target authentication information included in the target start instruction. In this way, by extracting the information to be authenticated representing the real target vehicle (that is, the living target vehicle) from the video image of the target vehicle matching the target start instruction, and authenticating the target vehicle based on the information to be authenticated, it is possible to ensure that the authenticated target vehicle is a real vehicle to the greatest extent, improve the reliability of the platform authentication result, and ensure the service security of the platform. BRIEF DESCRIPTION OF THE DRAWINGS

[0080] 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.

[0081] Figure 1 A schematic diagram of the architecture of a vehicle authentication system provided by an embodiment of the present application is shown;

[0082] Figure 2a A flow chart of a vehicle authentication method provided by an embodiment of the present application is shown;

[0083] Figure 2b A schematic diagram of a vehicle perspective provided by an embodiment of the present application is shown;

[0084] Figure 3 A flow chart of another vehicle authentication method provided by an embodiment of the present application is shown;

[0085] Figure 4a A flow chart of another vehicle authentication method provided by an embodiment of the present application is shown;

[0086] Figure 4b A schematic diagram of the structure of a vehicle orientation recognition model in an embodiment of the present application is shown;

[0087] Figure 5 A flow chart of another vehicle authentication method provided by an embodiment of the present application is shown;

[0088] Figure 6A flow chart of another vehicle authentication method provided by an embodiment of the present application is shown;

[0089] Figure 7a A flow chart of another vehicle authentication method provided by an embodiment of the present application is shown;

[0090] Figure 7b A schematic diagram showing a confidence matrix corresponding to a video image is shown;

[0091] Figure 8 A schematic diagram of the structure of a vehicle authentication device provided in an embodiment of the present application is shown;

[0092] Fig. 9 A schematic structural diagram of an electronic device provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0093] 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. It should be understood that the drawings in the present application only serve the purpose of explanation and description and are not used to limit the scope of protection of the present application. In addition, it should be understood that the schematic drawings are not drawn in real proportion. The flowchart used in this application shows the operations implemented according to some embodiments of the present application. It should be understood that the operations of the flowchart can be implemented out of sequence, and the steps without logical context can be reversed in order or implemented simultaneously. In addition, those skilled in the art can add one or more other operations to the flowchart under the guidance of the content of the present application, or remove one or more operations from the flowchart.

[0094] In addition, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various 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, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.

[0095] In order to enable those skilled in the art to use the contents of this application, the following implementation methods are provided in conjunction with the specific application scenario "online car-hailing field". For those skilled in the art, the general principles defined herein can be applied to other embodiments and application scenarios without departing from the spirit and scope of this application. Although this application is mainly described around the field of online car-hailing, it should be understood that this is only an exemplary embodiment.

[0096] 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.

[0097] The terms "passenger" and "service requester" in this application are used interchangeably to refer to a person, entity or tool that can request or order a service. The terms "driver", "user end" and "driver end" in this application are used interchangeably to refer to a person, entity or tool that can provide a service.

[0098] Figure 1 1 is a schematic diagram of the architecture of a vehicle authentication system 100 provided in an embodiment of the present application. For example, the vehicle authentication system 100 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 vehicle authentication system 100 may include one or more of a server 110 (i.e., an online car-hailing platform), a network 120, a user terminal 130, a target vehicle 140, and a database 150.

[0099] In some embodiments, the server 110 may include a processor. The processor may process information and / or data related to vehicle authentication to perform one or more functions described in the present application. For example, the processor determines the confidence of each frame of the target image under multiple vehicle perspectives based on the vehicle features corresponding to the target vehicle in each frame of the target image included in the acquired video image. In some embodiments, the processor may include one or more processing cores (e.g., a single-core processor (S) or a multi-core processor (M)). 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.

[0100] In some embodiments, the device type corresponding to the user terminal 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, or a tablet computer, a laptop computer, etc.

[0101] The target vehicle 140 is a service vehicle used by the driver to provide travel services to the service provider (ie, the passenger).

[0102] In some embodiments, the database 150 can be connected to the network 120 to communicate with one or more components in the vehicle authentication system 100 (e.g., the server 110, the user terminal 130, the target vehicle 140, etc.). One or more components in the vehicle authentication system 100 can access the data or instructions stored in the database 150 via the network 120. In some embodiments, the database 150 can be directly connected to one or more components in the vehicle authentication system 100, or the database 150 can also be part of the server 110.

[0103] Considering that after the online ride-hailing platform assigns a service vehicle to a passenger who requests a trip, there may be a situation where the actual vehicle reserved by the passenger is inconsistent with the service vehicle displayed on the online ride-hailing platform, which will cause security issues for the online ride-hailing platform. In order to ensure the security of the online ride-hailing platform and ensure that passengers have a better riding experience, the online ride-hailing platform needs to ensure that the service vehicle used by the service provider is consistent with the registered vehicle. Based on this, the online ride-hailing platform requires the service provider to authenticate the service vehicle used before providing travel services.

[0104] At present, the vehicle authentication process is usually based on the vehicle photo taken by the service provider. However, in this way, the vehicle photo may not be a real vehicle, but a screen-shot vehicle image or vehicle photo, etc. This makes the reliability of the authentication result low and cannot solve the security problem of the platform.

[0105] In order to avoid this situation, the embodiments of the present application provide a vehicle authentication method, device, electronic device and storage medium. By extracting the information to be authenticated from the video image of the target vehicle that matches the target start command and authenticating the target vehicle, it can maximize the assurance that the authenticated target vehicle is a real vehicle, thereby improving the reliability of the platform authentication results and ensuring the service security of the platform.

[0106] The following is combined with the above Figure 1 The contents described in the illustrated vehicle authentication system 100 provide a detailed description of the vehicle authentication method provided in the embodiment of the present application.

[0107] Reference Figure 2a FIG. 1 is a flow chart of a vehicle authentication method provided in an embodiment of the present application. The method can be executed by the server 110 in the vehicle authentication system 100. The specific execution process includes the following steps:

[0108] S201. Acquire a video image of a target vehicle matching a target start instruction, wherein the target start instruction includes target authentication information.

[0109] S202: Determine the confidence of each frame of the target image under multiple vehicle perspectives based on vehicle features corresponding to the target vehicle in each frame of the target image.

[0110] S203: Determine the information to be authenticated corresponding to the video image based on the confidence of each frame of the target image under multiple vehicle viewing angles.

[0111] S204: If the information to be authenticated matches the target authentication information, it is determined that the target vehicle has passed authentication.

[0112] A vehicle authentication method provided by an embodiment of the present application obtains a video image of a target vehicle matching a target start instruction, determines the confidence of each frame of the target image under multiple vehicle viewing angles based on the vehicle features corresponding to the target vehicle in each frame of the target image included in the video image, and then determines the information to be authenticated corresponding to the video image based on the confidence of each frame of the target image under multiple vehicle viewing angles; finally, the target vehicle is authenticated based on the matching result between the information to be authenticated and the target authentication information included in the target start instruction. In this way, by extracting the information to be authenticated representing the real target vehicle (i.e., the live target vehicle) from the video image of the target vehicle matching the target start instruction, and authenticating the target vehicle based on the information to be authenticated, it is possible to ensure that the authenticated target vehicle is a real vehicle to the greatest extent, thereby improving the reliability of the platform authentication result and ensuring the service security of the platform.

[0113] The following is a detailed description of the exemplary embodiments provided in the embodiments of the present application:

[0114] S201. Acquire a video image of a target vehicle matching a target start instruction, wherein the target start instruction includes target authentication information.

[0115] In the embodiment of the present application, the server determines multiple vehicle orientations (i.e., multiple vehicle viewing angles) of the vehicle according to different vehicle viewing angles. Optionally, the server determines 8 vehicle orientations, respectively corresponding to Figure 2b It corresponds to the 8 perspectives (i.e., 8 vehicle perspectives) in .

[0116] For the target vehicle, when the server detects that the target vehicle needs to be authenticated, it sends an authentication instruction to the user terminal corresponding to the target vehicle. The authentication instruction is used to notify the user terminal to take a video image of the target vehicle and upload it to the server, and the authentication instruction includes the target authentication information. Among them, the server can authenticate the target vehicle before the service provider uses the target vehicle each time; it can also authenticate the target vehicle regularly (i.e., every preset time), for example, once every 2 months. Here, the embodiment of the present application does not limit the specific time of vehicle authentication.

[0117] In one embodiment, the target authentication information includes the target vehicle perspective timing (i.e., the target vehicle orientation timing) corresponding to the video image to be captured, and the target vehicle perspective timing is the specific perspective circling direction of the target vehicle captured by the service provider. Each vehicle corresponds to two vehicle perspective timings, including a first perspective timing (e.g., a clockwise direction from vehicle perspective numbers from small to large) and a second perspective timing (e.g., a counterclockwise direction from vehicle perspective numbers from large to small), and the target vehicle perspective timing is randomly selected from one of the two vehicle perspective timings; for example, the target vehicle perspective timing is a clockwise direction from vehicle perspective numbers from small to large, and accordingly, the target authentication information is to capture the target vehicle in a clockwise direction from vehicle perspective numbers from small to large.

[0118] After receiving the authentication instruction, the user terminal captures the video image of the target vehicle according to the target vehicle viewing angle sequence in the authentication instruction, and sends the captured video image to the server.

[0119] In another embodiment, the target authentication information includes a target starting vehicle perspective and a target vehicle perspective timing corresponding to the video image to be captured; wherein the target starting vehicle perspective is a vehicle perspective randomly selected from multiple vehicle perspectives; for example, the target starting vehicle perspective is perspective 3, and the target vehicle perspective timing is a clockwise direction from vehicle perspective numbers from small to large. Accordingly, the target authentication information is to capture the target vehicle starting from perspective 3 and moving along the clockwise direction from vehicle perspective numbers from small to large.

[0120] After receiving the authentication instruction, the user terminal captures the video image of the target vehicle according to the target starting vehicle viewing angle and the target vehicle viewing angle sequence in the authentication instruction, and sends the captured video image to the server.

[0121] S202: Determine the confidence of each frame of the target image under multiple vehicle perspectives based on vehicle features corresponding to the target vehicle in each frame of the target image.

[0122] In the embodiment of the present application, after receiving the video image matching the target start instruction sent by the user terminal, the server extracts multiple frames of target images in the video image to obtain a picture sequence with a fixed length and arranged in a preset order. Here, the server extracts frames of the video image at equal time intervals, that is, obtains the target image in the video image at a preset time interval. For example, the server obtains a frame of the target image every 1 second to obtain a picture sequence including multiple frames of the target image.

[0123] After obtaining the image sequence, the server extracts the vehicle features corresponding to the target vehicle in each frame of the target image included in the image sequence, and determines the confidence of each frame of the target image under multiple vehicle perspectives based on the vehicle features corresponding to the target vehicle in each frame of the target image.

[0124] In the embodiment of the present application, a trained vehicle orientation recognition model is pre-stored in the server, and the vehicle orientation recognition model includes a plurality of different tasks. In the embodiment of the present application, the model has three tasks. Among them, the first task is to output the first vehicle feature corresponding to each frame of the target image (that is, the vehicle feature corresponding to the video image of the target vehicle) based on the multiple frames of the target image in the picture sequence; the second task is to output the second vehicle feature corresponding to each frame of the target image (that is, the second vehicle feature corresponding to the video image of the target vehicle) based on the multiple frames of the target image in the picture sequence; the third task is to output the confidence of each frame of the target image included in the video image under multiple vehicle perspectives based on the vehicle features corresponding to the target vehicle (including the first vehicle feature and the second vehicle feature).

[0125] The execution order of the first task and the second task in the model is not limited; the third task is executed after the first task and the second task, and is processed based on the output results of the first task and the second task, and outputs corresponding results.

[0126] Correspondingly, the server inputs multiple frames of target images in the image sequence into the above-mentioned pre-trained vehicle orientation recognition model. The first task of the vehicle orientation recognition model outputs the first vehicle feature corresponding to each frame of the target image, the second task outputs the second vehicle feature corresponding to each frame of the target image, and the third task outputs the confidence of each frame of the target image under multiple vehicle perspectives.

[0127] S203: Determine the information to be authenticated corresponding to the video image based on the confidence of each frame of the target image under multiple vehicle viewing angles.

[0128] In an embodiment of the present application, the multiple frames of target images in the picture sequence are arranged in a preset order. The server determines the starting vehicle perspective based on the first frame of the target image, and uses the starting vehicle perspective to be authenticated as the starting point and the end point, and obtains multiple paths corresponding to the video image (specifically, the picture sequence) in a manner such that each path includes multiple frames of target images in the video image (specifically, the picture sequence) and each frame of the target image in each path corresponds to a vehicle perspective.

[0129] Furthermore, based on the confidence of each frame of the target image under multiple vehicle perspectives, a target path representing the information to be authenticated corresponding to the target video is selected from multiple paths corresponding to the video image, and based on the target path, it is determined that the information to be authenticated corresponds to the image.

[0130] Optionally, the information to be authenticated includes a timing sequence of a vehicle viewing angle to be authenticated; or, the information to be authenticated includes a starting vehicle viewing angle to be authenticated and a timing sequence of a vehicle viewing angle to be authenticated.

[0131] S204: If the information to be authenticated matches the target authentication information, it is determined that the target vehicle has passed authentication.

[0132] In the embodiment of the present application, the server matches the information to be authenticated with the target authentication information in the target start instruction. If the two match, it is determined that the target vehicle has been authenticated.

[0133] In one embodiment, the target authentication information includes the target vehicle perspective timing, and the information to be authenticated includes the vehicle perspective timing to be authenticated; accordingly, the server compares the vehicle perspective timing to be authenticated in the information to be authenticated and the target vehicle perspective timing in the target authentication information. If the vehicle perspective timing to be authenticated and the target vehicle perspective timing are the same, it is determined that the target vehicle authentication is successful.

[0134] In another embodiment, the target authentication information includes the target starting vehicle viewing angle and the target vehicle viewing angle timing; the information to be authenticated includes the target starting vehicle viewing angle and the target vehicle viewing angle timing.

[0135] The server compares the perspective of the starting vehicle to be authenticated in the information to be authenticated and the perspective of the target starting vehicle in the target authentication information, and compares the perspective timing of the vehicle to be authenticated in the information to be authenticated and the perspective timing of the target vehicle in the target authentication information. If the perspective of the starting vehicle to be authenticated and the perspective of the target starting vehicle are the same, and the perspective timing of the vehicle to be authenticated and the perspective timing of the target vehicle are the same, then it is determined that the target vehicle authentication is successful.

[0136] A vehicle authentication method provided in an embodiment of the present application can maximize the assurance that the authenticated target vehicle is a real vehicle by extracting information to be authenticated that represents a real target vehicle (i.e., a living target vehicle) from a video image of a target vehicle that matches a target start instruction, and authenticating the target vehicle based on the information to be authenticated, thereby improving the reliability of the platform authentication results and ensuring the service security of the platform.

[0137] Further, such as Figure 3 As shown, in the vehicle authentication method provided in the embodiment of the present application, the confidence of each frame of the target image under multiple vehicle perspectives is determined based on the vehicle features corresponding to the target vehicle in each frame of the target image, including:

[0138] S301, extracting vehicle features corresponding to the target vehicle in each frame of the target image included in the video image.

[0139] Wherein, for each frame of target image included in the video image, image features corresponding to the frame of target image are extracted; based on the image features corresponding to the frame of target image, vehicle features corresponding to the target vehicle in the frame of target image are determined.

[0140] In an embodiment of the present application, a trained CNN-based vehicle orientation recognition model is pre-stored in the server. The first task of the vehicle orientation recognition model is to output the first vehicle feature corresponding to each frame of the target image (that is, the vehicle feature corresponding to the video image of the target vehicle) based on multiple frames of target images in the picture sequence; the second task is to output the second vehicle feature corresponding to each frame of the target image (that is, the second vehicle feature corresponding to the video image of the target vehicle) based on multiple frames of target images in the picture sequence. Accordingly, for each frame of the target image, after extracting the image features corresponding to the frame of the target image, the image features corresponding to the frame of the target image are input into the vehicle orientation recognition model. The first task of the vehicle orientation recognition model outputs the first vehicle feature corresponding to each frame of the target image, and the second task outputs the second vehicle feature corresponding to each frame of the target image. Based on the first vehicle feature and the second vehicle feature, the vehicle feature corresponding to the target vehicle in the frame of the target image is obtained.

[0141] S302: for each frame of target image included in the video image, process the vehicle features corresponding to the target vehicle in the frame of target image based on the first processing parameter to obtain the confidence of the frame of target image under multiple vehicle viewing angles.

[0142] In an embodiment of the present application, the third task of the vehicle orientation recognition model is to output the confidence of each frame of the target image included in the video image under multiple vehicle perspectives based on the vehicle features (including the first vehicle features and the second vehicle features) corresponding to the target vehicle.

[0143] Correspondingly, after obtaining the vehicle features corresponding to the target vehicle in the frame target image based on the first vehicle features and the second vehicle features, the server inputs the vehicle features corresponding to the target vehicle into the third task, and the third task outputs the confidence of each frame target image included in the video image under multiple vehicle perspectives.

[0144] In one embodiment, the vehicle orientation recognition model includes a fully connected layer, and the vehicle features corresponding to the target vehicle in the frame target image are input into the fully connected layer to obtain the confidence of each frame target image included in the video image under multiple vehicle perspectives.

[0145] Further, such as Figure 4a As shown, in the vehicle authentication method provided in the embodiment of the present application, the determining of the vehicle features corresponding to the target vehicle in the frame target image based on the image features corresponding to the frame target image includes:

[0146] S401, performing convolution processing on image features corresponding to the target image of the frame based on the target convolution kernel to obtain deep image features corresponding to the target image of the frame.

[0147] In the embodiment of the present application, the vehicle orientation recognition model includes multiple convolution layers, each of which corresponds to a corresponding target convolution kernel, and the multiple convolution layers are sequentially connected and process the input image features through the corresponding target convolution kernels. Figure 4b As shown, there are two convolution layers, corresponding to the first convolution layer and the second convolution layer respectively; for each frame of target image included in the video image, after extracting the image features corresponding to the frame of target image, the image features corresponding to the frame of target image are input into the first convolution layer of the vehicle orientation recognition model, the first convolution layer performs convolution processing on the image features based on the first target convolution kernel to obtain the processed image features, and the processed image features are input into the second convolution layer, the second convolution layer performs convolution processing on the processed image features based on the second target convolution kernel to obtain the deep image features corresponding to the frame of target image.

[0148] S402: Process deep image features corresponding to the frame of target image to determine vehicle features corresponding to the target vehicle in the frame of target image.

[0149] In an embodiment of the present application, the vehicle orientation recognition model also includes two task layers, namely the first task layer and the second task layer (i.e., the pooling layer). The deep image features corresponding to the target image of the frame obtained after convolution processing are input into the two task layers respectively. The two task layers process the deep image features corresponding to the target image of the frame respectively, and obtain the vehicle attribute features corresponding to the target vehicle according to the output results of the two task layers.

[0150] like Figure 5 As shown, in the embodiment of the present application, the process of processing the deep image features corresponding to the frame target image based on the above two task layers is as follows:

[0151] S501, processing the deep image features corresponding to the frame target image based on the second processing parameters, and determining the first vehicle attribute features corresponding to the target vehicle in the frame target image based on the obtained processing results.

[0152] In the embodiment of the present application, the first task layer in the vehicle orientation recognition model has a second processing parameter. For each frame of target image included in the video image, after extracting the image features corresponding to the frame of target image, the image features corresponding to the frame of target image are input into the first task layer, and the first task layer processes the deep image features corresponding to the frame of target image based on the second processing parameters to obtain the processing result corresponding to the frame of target image. After obtaining the processing result corresponding to the frame of target image, the server determines the first vehicle attribute feature corresponding to the target vehicle in the frame of target image based on the processing result.

[0153] Here, the processing result corresponding to the frame target image includes the second attribute information of the target vehicle corresponding to the target vehicle in the frame target image. The second attribute information of the target vehicle includes: the first position of the vehicle frame corresponding to the target vehicle, the second positions corresponding to each wheel included in the target vehicle, and the wheel types corresponding to each wheel in the target vehicle.

[0154] S502: Process the deep image features corresponding to the frame target image based on a third processing parameter to obtain a second vehicle attribute feature corresponding to the target vehicle in the frame target image.

[0155] In the embodiment of the present application, the second task layer (i.e., the pooling layer) in the vehicle orientation recognition model has a third processing parameter. For each frame of target image included in the video image, after extracting the image features corresponding to the frame of target image, the image features corresponding to the frame of target image are input into the second task layer, and the second task layer processes the deep image features corresponding to the frame of target image based on the third processing parameters to obtain the second vehicle attribute features corresponding to the target vehicle in the frame of target image.

[0156] S503: Determine a vehicle feature corresponding to the target vehicle in the target image frame based on the first vehicle attribute feature and the second vehicle attribute feature.

[0157] In the embodiment of the present application, the first vehicle attribute feature and the second vehicle attribute feature are spliced ​​to obtain the vehicle feature corresponding to the target vehicle in the frame target image.

[0158] In the embodiment of the present application, after the first task layer outputs the processing result of the deep image feature corresponding to the frame target image, the processing result needs to be further processed to obtain the first vehicle attribute feature corresponding to the target vehicle in the frame target image. Further, Figure 6 As shown, the vehicle authentication method provided in the embodiment of the present application, based on the obtained processing result, determines the first vehicle attribute feature corresponding to the target vehicle in the frame target image, including:

[0159] S601. Determine first attribute information of the vehicle frame based on a first position of the vehicle frame corresponding to the target vehicle in the target image of the frame.

[0160] In the embodiment of the present application, the second attribute information of the target vehicle includes the first position of the vehicle frame corresponding to the target vehicle, the second position corresponding to each wheel included in the target vehicle, and the wheel type corresponding to each wheel in the target vehicle. The first attribute information obtained based on the first position includes the length, width, and aspect ratio of the vehicle frame.

[0161] For example, the first position of the vehicle frame is (x1, y1), (x2, y2), and accordingly, the length L of the vehicle frame is: the absolute value of (x2-x1); the width W of the vehicle frame is: the absolute value of (y2-y1); the aspect ratio is L / W, that is, (x2-x1) / (y2-y1).

[0162] S602: Determine the third positions of the wheels in the vehicle frame based on the first position of the vehicle frame and the second positions of the wheels of the target vehicle in the target image frame.

[0163] In an embodiment of the present application, the processing result (i.e., the second attribute information of the target vehicle) includes the second position corresponding to each wheel of the target vehicle, which is the position of the pixel point of each vehicle in the target image of the frame. Here, for each wheel, the relative position of the wheel relative to the vehicle frame is calculated based on the second position corresponding to the vehicle and the first position corresponding to the vehicle frame.

[0164] S603, determining a first vehicle attribute feature corresponding to the target vehicle in the target image of the frame according to the first attribute information of the vehicle frame, the wheel type corresponding to each wheel in the target vehicle, and the relative position of each wheel in the vehicle frame.

[0165] In an embodiment of the present application, the first attribute information of the vehicle frame corresponds to three elements, the target vehicle corresponds to four wheels, the relative position of each wheel in the vehicle frame corresponds to two elements, and accordingly, the four vehicles correspond to eight elements respectively; among which, if the corresponding wheel is not detected, the element value of the corresponding position is 0.

[0166] The above multiple elements (i.e., 11 elements) are spliced ​​together to obtain the first vehicle attribute feature corresponding to the target vehicle. Among them, the wheel type of each wheel of the target vehicle is identified by the position of each element (i.e., the arrangement order). For example, the first vehicle attribute feature includes 11 elements, the first element position represents the length of the vehicle frame, the second element position represents the width of the vehicle frame, the third element position represents the aspect ratio of the vehicle frame, the fourth and fifth element positions represent the left front wheel, the sixth and seventh element positions represent the left rear wheel, the eighth and ninth element positions represent the right front wheel, and the tenth and eleventh element positions represent the right rear wheel.

[0167] Correspondingly, the first vehicle attribute feature and the second vehicle attribute feature are concatenated to obtain the vehicle feature corresponding to the target vehicle.

[0168] In the embodiment of the present application, there is no limitation on the order of splicing the first vehicle attribute feature and the second vehicle attribute feature.

[0169] In the embodiment of the present application, the first processing parameter, the second processing parameter and the third processing parameter in the vehicle orientation recognition model need to be pre-trained, wherein the second processing parameter is firstly trained, and when the second processing parameter is trained, the first processing parameter and the third processing parameter are trained again based on the second processing parameter. The specific training process includes the following steps:

[0170] 1. Obtain a sample video image corresponding to the target vehicle.

[0171] 2. Using the image features included in each frame of the sample video image as the first input feature and the second attribute information of the target vehicle corresponding to each frame of the sample video image as the output, the second processing parameter is trained.

[0172] 3. Using the image features included in each frame of the sample video image as the second input feature, using the vehicle features corresponding to the target vehicle in each frame of the sample image as the third input feature corresponding to the third processing parameter, using the authentication information of the target vehicle corresponding to the sample video image as the output, training the first processing parameter and the third processing parameter. Here, the second input feature may be the same as or different from the first input feature.

[0173] In an embodiment of the present application, the image features included in each frame of the sample image in the sample video image are used as the input features of the first convolutional layer in the vehicle orientation recognition model, the vehicle features corresponding to the target vehicle in each frame of the sample image are used as the third input features of the second task layer (i.e., the pooling layer) in the vehicle orientation recognition model, and the authentication information of the target vehicle corresponding to the sample video image is used as the output, and the first processing parameters and the third processing parameters are trained at the same time.

[0174] Further, such as Figure 7a As shown, the vehicle authentication method provided in the embodiment of the present application, the information to be authenticated corresponding to the video image is determined based on the confidence of each frame of the target image under multiple vehicle viewing angles, including:

[0175] S701. Determine a starting vehicle perspective corresponding to the first frame target image based on confidences of the first frame target image under multiple vehicle perspectives; wherein the multiple vehicle perspectives are arranged in a circular manner.

[0176] In the embodiment of the present application, the server includes multiple vehicle perspectives, which are arranged in a circular manner. Figure 2b As shown, the server includes 8 vehicle perspectives, which are arranged in a ring. Based on the confidence of the first frame target image under multiple vehicle perspectives, the server selects the vehicle perspective with the highest confidence as the starting vehicle perspective corresponding to the first frame target image. For example, vehicle perspective 3 is determined as the starting vehicle perspective of the first frame target image.

[0177] S702, obtaining multiple paths corresponding to the video image, which take the starting vehicle perspective as the starting point and pass through other vehicle perspectives and then return to the starting vehicle perspective; wherein each path includes multiple frames of target images in the video image, and each frame of the target image in the path corresponds to a vehicle perspective.

[0178] In an embodiment of the present application, the server uses the starting vehicle perspective as the starting point and end point of the path, so that each path includes multiple frames of target images in the video image (i.e., picture sequence), and each frame of target image in each path corresponds to a vehicle perspective to obtain multiple paths corresponding to the video image (i.e., picture sequence). Among them, the starting point includes two vehicle perspective time sequences (including clockwise and counterclockwise directions) passing through other vehicle perspectives and returning to the starting vehicle perspective again. For any vehicle perspective time sequence, the vehicle perspective corresponding to each frame of target image in the path is the first vehicle perspective corresponding to the previous frame of target image of the frame of target image or the next second vehicle perspective after the first vehicle perspective in the vehicle perspective time sequence.

[0179] In the embodiment of the present application, for multiple frames of target images extracted from video images, a confidence matrix is ​​constructed based on the confidence distribution of each frame of the target image under multiple vehicle perspectives; the rows in the confidence matrix represent the target images, and the columns in the confidence matrix represent the vehicle perspectives. Figure 7b As shown, it is a confidence matrix corresponding to the video image constructed by the server, where Vi represents the i-th viewing angle, i is a positive integer, and accordingly, V 1-V 8 represent the vehicle viewing angle; Fi represents the i-th frame target image extracted from the video image, and F1-F16 represent multiple frames of target images. In the confidence matrix, the element at the (Fi, Vj) position of the matrix represents the confidence of the i-th frame target image at the vehicle viewing angle j.

[0180] In the embodiment of the present application, the starting vehicle perspective corresponding to the target vehicle is V3, and the server obtains multiple paths from (F1, V3) to (F16, V3). Each path includes the target image F1 to the target image F16, and each frame of the target image in the path corresponds to a vehicle perspective (that is, each row of the confidence matrix only takes one element); for any vehicle perspective sequence, the vehicle perspective corresponding to each frame of the target image in each path is the first vehicle perspective corresponding to the previous frame of the target image or the next second vehicle perspective after the first vehicle perspective in the vehicle perspective sequence, and the selection of the target image in the path cannot go back in the row direction.

[0181] S703 . Select a target path from the multiple paths based on the confidence of each frame of the target image under multiple vehicle perspectives.

[0182] In an embodiment of the present application, based on the confidence of each frame of the target image under multiple vehicle perspectives, the total confidence corresponding to each of the multiple paths is determined; according to the total confidence corresponding to each of the multiple paths, the target path with the largest corresponding total confidence is selected from the multiple paths.

[0183] In an optional embodiment, for each of the multiple paths selected from the confidence matrix, the target vehicle perspective corresponding to each frame of the target image in the path is obtained, and based on the confidence of each frame of the target image under the target vehicle perspective corresponding to the path, the total confidence corresponding to the path is determined, that is, the confidence corresponding to each frame of the target image included in the path is added to obtain the total confidence corresponding to the path; based on the total confidence corresponding to each path, the target path with the largest total confidence is selected from the multiple paths.

[0184] After determining the target path, the server further determines whether the total confidence corresponding to the target path is greater than a preset threshold. If so, the target path is determined to be legal; otherwise, the target path is invalid.

[0185] S704: Determine the information to be authenticated corresponding to the video image according to the target path.

[0186] In one embodiment, the information to be authenticated includes the vehicle perspective timing to be authenticated; accordingly, the server obtains the specific vehicle perspective timing corresponding to the target path, and determines the specific vehicle perspective timing as the vehicle perspective timing to be authenticated.

[0187] In another embodiment, the information to be authenticated includes the starting vehicle perspective to be authenticated and the vehicle perspective timing to be authenticated; accordingly, the server obtains the starting vehicle perspective and the specific vehicle perspective timing included in the target path, and determines the starting vehicle perspective as the starting vehicle perspective to be authenticated, and determines the specific vehicle perspective timing as the vehicle perspective timing to be authenticated.

[0188] Correspondingly, the server compares the information to be authenticated with the target authentication information. If the information to be authenticated matches the target authentication information, it is determined that the target vehicle is alive and the authentication of the target vehicle is confirmed to be successful.

[0189] The following is an overall description of a vehicle authentication method provided in an embodiment of the present application:

[0190] 1. When the server detects that the target vehicle needs to be authenticated, it generates an authentication instruction including the target authentication information and sends it to the user end. The target authentication information is randomly selected from the existing authentication information.

[0191] 2. The user takes a video image corresponding to the target vehicle and uploads it to the server.

[0192] 3. The server extracts frames from the video images uploaded by the user at equal time intervals to obtain a sequence of images arranged in a preset order and having a fixed length.

[0193] 4. The server uses the vehicle orientation recognition model to identify the confidence distribution of each frame of the target image in the image sequence in 3 at each vehicle perspective to form a confidence matrix.

[0194] 5. The server confirms the starting vehicle perspective corresponding to the first frame of the target image based on the confidence distribution of each frame of the target image at each vehicle perspective, and calculates the target path with the highest total score (i.e., total confidence) in the clockwise and counterclockwise directions corresponding to the perspective at the corresponding position on the confidence matrix in 4. When the total confidence (i.e., total score) corresponding to the target path is greater than the preset threshold, the target vehicle perspective timing corresponding to the target path is determined to be the vehicle perspective timing to be authenticated.

[0195] 6. The server compares the starting vehicle perspective and the perspective timing of the vehicle to be authenticated in the target road with the corresponding information in the target authentication information. If the comparison results are consistent, it is determined that the vehicle is alive, and the target vehicle authentication is confirmed to be successful.

[0196] A vehicle authentication method provided in an embodiment of the present application can maximize the assurance that the authenticated target vehicle is a real vehicle by extracting information to be authenticated that represents a real target vehicle (i.e., a living target vehicle) from a video image of a target vehicle that matches a target start instruction, and authenticating the target vehicle based on the information to be authenticated, thereby improving the reliability of the platform authentication results and ensuring the service security of the platform.

[0197] Based on the same inventive concept, a vehicle authentication device corresponding to the vehicle authentication method is also provided in the embodiment of the present application. Since the principle of solving the problem by the device in the embodiment of the present application is similar to the above-mentioned vehicle authentication method in the embodiment of the present application, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be repeated.

[0198] Reference Figure 8 FIG. 1 is a schematic diagram of a vehicle authentication device provided in an embodiment of the present application, wherein the device comprises:

[0199] A first acquisition module 801 is used to acquire a video image of a target vehicle matching a target start instruction, wherein the target start instruction includes target authentication information;

[0200] A first determination module 802 is used to determine the confidence of each frame of the target image under multiple vehicle perspectives based on the vehicle features corresponding to the target vehicle in each frame of the target image;

[0201] A second determination module 803 is used to determine the information to be authenticated corresponding to the video image based on the confidence of each frame of the target image under multiple vehicle perspectives;

[0202] The third determination module 804 is configured to determine that the target vehicle has been authenticated if the information to be authenticated matches the target authentication information.

[0203] In a possible implementation, the first determination module 802 determines the confidence of each frame of the target image under multiple vehicle perspectives based on the vehicle features corresponding to the target vehicle in each frame of the target image, including:

[0204] Extracting vehicle features corresponding to the target vehicle in each frame of the target image included in the video image;

[0205] For each frame of target image included in the video image, the vehicle features corresponding to the target vehicle in the frame of target image are processed based on the first processing parameters to obtain the confidence of the frame of target image under multiple vehicle viewing angles.

[0206] In a possible implementation manner, the first determination module 802 extracts vehicle features corresponding to the target vehicle in each frame of the target image included in the video image, including:

[0207] For each frame of target image included in the video image, extracting image features corresponding to the frame of target image;

[0208] Based on the image features corresponding to the frame target image, the vehicle features corresponding to the target vehicle in the frame target image are determined.

[0209] In a possible implementation manner, the first determination module 802 determines the vehicle features corresponding to the target vehicle in the frame target image based on the image features corresponding to the frame target image, including:

[0210] Performing convolution processing on the image features corresponding to the target image of the frame based on the target convolution kernel to obtain the deep image features corresponding to the target image of the frame;

[0211] The deep image features corresponding to the frame target image are processed to determine the vehicle features corresponding to the target vehicle in the frame target image.

[0212] In a possible implementation, the first determination module 802 processes the deep image features corresponding to the frame target image to determine the vehicle features corresponding to the target vehicle in the frame target image, including:

[0213] Processing the deep image feature corresponding to the frame target image based on the second processing parameter, and determining the first vehicle attribute feature corresponding to the target vehicle in the frame target image based on the obtained processing result;

[0214] Processing the deep image feature corresponding to the frame target image based on the third processing parameter to obtain a second vehicle attribute feature corresponding to the target vehicle in the frame target image;

[0215] Based on the first vehicle attribute feature and the second vehicle attribute feature, a vehicle feature corresponding to the target vehicle in the frame target image is determined.

[0216] In a possible implementation, the first determination module 802 determines the first vehicle attribute feature corresponding to the target vehicle in the frame target image based on the obtained processing result, including:

[0217] Determining first attribute information of the vehicle frame based on a first position of a vehicle frame corresponding to the target vehicle in the frame target image;

[0218] Determine the third positions of the wheels in the vehicle frame based on the first position of the vehicle frame and the second positions of the wheels of the target vehicle in the target image frame.

[0219] The first vehicle attribute feature corresponding to the target vehicle in the target image of the frame is determined according to the first attribute information of the vehicle frame, the wheel types corresponding to the wheels in the target vehicle and the relative positions of the wheels in the vehicle frame.

[0220] In a possible implementation, the device further includes:

[0221] A second acquisition module is used to acquire a sample video image corresponding to the target vehicle;

[0222] A first training module is used to train the second processing parameter by taking the image feature of each frame of the sample image in the sample video image as the first input feature and taking the second attribute information of the target vehicle corresponding to each frame of the sample image in the sample video image as the output;

[0223] The second training module is used to train the first processing parameter and the third processing parameter using the image features included in each frame of the sample image in the sample video image as the second input feature, the vehicle features corresponding to the target vehicle in each frame of the sample image as the third input feature corresponding to the third processing parameter, and the authentication information of the target vehicle corresponding to the sample video image as the output.

[0224] In a possible implementation, the second determination module 803 determines the information to be authenticated corresponding to the video image based on the confidence of each frame of the target image under multiple vehicle perspectives, including:

[0225] Determine a starting vehicle perspective corresponding to the first frame target image based on the confidence of the first frame target image under multiple vehicle perspectives respectively; wherein the multiple vehicle perspectives are arranged in a cyclic manner;

[0226] Acquire multiple paths corresponding to the video image, which start from the starting vehicle perspective and pass through other vehicle perspectives and then return to the starting vehicle perspective; wherein each path includes multiple frames of target images in the video image, and each frame of the target image in the path corresponds to a vehicle perspective;

[0227] Selecting a target path from the multiple paths based on the confidence of each frame of the target image under multiple vehicle perspectives;

[0228] According to the target path, information to be authenticated corresponding to the video image is determined.

[0229] In a possible implementation manner, the starting vehicle perspective is taken as the starting point, including two vehicle perspectives in sequence, passing through other vehicle perspectives and returning to the starting vehicle perspective again;

[0230] For any vehicle perspective sequence, the vehicle perspective corresponding to each frame of the target image is the first vehicle perspective corresponding to the previous frame of the target image or the next second vehicle perspective after the first vehicle perspective in the vehicle perspective sequence.

[0231] In a possible implementation manner, the second determination module 803 selects a target path from the multiple paths based on the confidence of each frame of the target image under multiple vehicle perspectives, including:

[0232] Determining a total confidence corresponding to each of the multiple paths based on the confidence of each frame of the target image under multiple vehicle perspectives;

[0233] According to the total confidence corresponding to each path in the multiple paths, a target path with the maximum corresponding total confidence is selected from the multiple paths.

[0234] A vehicle authentication device provided in an embodiment of the present application can maximize the assurance that the authenticated target vehicle is a real vehicle by extracting information to be authenticated that represents a real target vehicle (i.e., a living target vehicle) from a video image of a target vehicle that matches a target start instruction, and authenticating the target vehicle based on the information to be authenticated, thereby improving the reliability of the platform authentication results and ensuring the service security of the platform.

[0235] like Fig. 9 As shown, an embodiment of the present application further provides an electronic device 900, which includes: a processor 901, a memory 902 and a bus, wherein the memory 902 stores machine-readable instructions executable by the processor 901, and when the electronic device is running, the processor 901 communicates with the memory 902 through the bus, and the processor 901 executes the machine-readable instructions to perform the steps of the vehicle authentication method provided in the above embodiment.

[0236] Specifically, the above-mentioned memory 902 and processor 901 can be general-purpose memories and processors, which are not specifically limited here. When the processor 901 runs the computer program stored in the memory 902, the vehicle authentication method provided in the above-mentioned embodiment can be executed.

[0237] Corresponding to the above-mentioned vehicle authentication method, an embodiment of the present application further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the vehicle authentication method provided in the above-mentioned embodiment are executed.

[0238] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, the specific working process of the system and device described above can refer to the corresponding process in the method embodiment, and will not be repeated in this application. In the several embodiments provided in this application, it should be understood that the disclosed system, device and method can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the modules is only a logical function division. There may be other division methods in actual implementation. For example, multiple modules 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, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0239] The modules described as separate components may or may not be physically separated, and the components shown as modules 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.

[0240] 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.

[0241] 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, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, server, or 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 USB flash drives, mobile hard drives, ROM, RAM, magnetic disks, or optical disks.

[0242] The above are only specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A vehicle authentication method, characterized in that: The method comprises: Acquire a video image of a target vehicle matching a target start instruction, wherein the target start instruction includes target authentication information; wherein the video image includes multiple frames of target images, and the target authentication information includes a target vehicle viewing angle timing; Based on the vehicle features corresponding to the target vehicle in each frame of the target image, determining the confidence of each frame of the target image under multiple vehicle perspectives; Based on the confidence of each frame of the target image under multiple vehicle perspectives, determining the information to be authenticated corresponding to the video image, wherein the information to be authenticated includes a time sequence of the vehicle perspective to be authenticated; If the information to be authenticated matches the target authentication information, it is determined that the target vehicle has passed authentication.

2. The vehicle authentication method according to claim 1, characterized in that: The step of determining the confidence of each frame of the target image under multiple vehicle perspectives based on the vehicle features corresponding to the target vehicle in each frame of the target image comprises: Extracting vehicle features corresponding to the target vehicle in each frame of the target image included in the video image; For each frame of target image included in the video image, the vehicle features corresponding to the target vehicle in the frame of target image are processed based on the first processing parameters to obtain the confidence of the frame of target image under multiple vehicle viewing angles.

3. The vehicle authentication method according to claim 2, characterized in that: The extracting the vehicle features corresponding to the target vehicle in each frame of the target image included in the video image includes: For each frame of target image included in the video image, extracting image features corresponding to the frame of target image; Based on the image features corresponding to the frame target image, the vehicle features corresponding to the target vehicle in the frame target image are determined.

4. The vehicle authentication method according to claim 3, characterized in that: The determining, based on the image features corresponding to the frame target image, the vehicle features corresponding to the target vehicle in the frame target image includes: Performing convolution processing on the image features corresponding to the target image of the frame based on the target convolution kernel to obtain the deep image features corresponding to the target image of the frame; The deep image features corresponding to the frame target image are processed to determine the vehicle features corresponding to the target vehicle in the frame target image.

5. The vehicle authentication method according to claim 4, characterized in that: The processing of the deep image features corresponding to the frame target image to determine the vehicle features corresponding to the target vehicle in the frame target image includes: Processing the deep image feature corresponding to the frame target image based on the second processing parameter, and determining the first vehicle attribute feature corresponding to the target vehicle in the frame target image based on the obtained processing result; Processing the deep image feature corresponding to the frame target image based on the third processing parameter to obtain a second vehicle attribute feature corresponding to the target vehicle in the frame target image; Based on the first vehicle attribute feature and the second vehicle attribute feature, a vehicle feature corresponding to the target vehicle in the frame target image is determined.

6. The vehicle authentication method according to claim 5, characterized in that: The determining, based on the obtained processing result, a first vehicle attribute feature corresponding to the target vehicle in the target image of the frame includes: Determining first attribute information of the vehicle frame based on a first position of a vehicle frame corresponding to the target vehicle in the frame target image; Determine the third positions of the wheels in the vehicle frame based on the first position of the vehicle frame and the second positions of the wheels of the target vehicle in the target image frame. The first vehicle attribute feature corresponding to the target vehicle in the target image of the frame is determined according to the first attribute information of the vehicle frame, the wheel types corresponding to the wheels in the target vehicle and the relative positions of the wheels in the vehicle frame.

7. The vehicle authentication method according to claim 5, characterized in that: The first processing parameter, the second processing parameter and the third processing parameter are determined by: Acquire a sample video image corresponding to the target vehicle; Taking the image feature of each frame of the sample image in the sample video image as the first input feature and taking the second attribute information of the target vehicle corresponding to each frame of the sample image in the sample video image as the output, training the second processing parameter; The first processing parameter and the third processing parameter are trained by taking the image features included in each frame of the sample video image as the second input features, taking the vehicle features corresponding to the target vehicle in each frame of the sample image as the third input features corresponding to the third processing parameters, and taking the authentication information of the target vehicle corresponding to the sample video image as the output.

8. The vehicle authentication method according to any one of claims 1 to 7, characterized in that: The determining the information to be authenticated corresponding to the video image based on the confidence of each frame of the target image under multiple vehicle viewing angles includes: Determine a starting vehicle perspective corresponding to the first frame target image based on the confidence of the first frame target image under multiple vehicle perspectives respectively; wherein the multiple vehicle perspectives are arranged in a cyclic manner; Acquire multiple paths corresponding to the video image, which start from the starting vehicle perspective and pass through other vehicle perspectives and then return to the starting vehicle perspective; wherein each path includes multiple frames of target images in the video image, and each frame of the target image in the path corresponds to a vehicle perspective; Selecting a target path from the multiple paths based on the confidence of each frame of the target image under multiple vehicle perspectives; According to the target path, information to be authenticated corresponding to the video image is determined.

9. The vehicle authentication method according to claim 8, characterized in that: in, Taking the starting vehicle perspective as the starting point, a time sequence of two vehicle perspectives passes through other vehicle perspectives and then returns to the starting vehicle perspective; For any vehicle perspective sequence, the vehicle perspective corresponding to each frame of the target image is the first vehicle perspective corresponding to the previous frame of the target image or the next second vehicle perspective after the first vehicle perspective in the vehicle perspective sequence.

10. The vehicle authentication method according to claim 8, characterized in that: The selecting a target path from the multiple paths based on the confidence of each frame of the target image under multiple vehicle perspectives includes: Determining a total confidence corresponding to each of the multiple paths based on the confidence of each frame of the target image under multiple vehicle perspectives; According to the total confidence corresponding to each path in the multiple paths, a target path with the maximum corresponding total confidence is selected from the multiple paths.

11. A vehicle authentication device, characterized in that: The device comprises: A first acquisition module is used to acquire a video image of a target vehicle matching a target start instruction, wherein the target start instruction includes target authentication information; wherein the video image includes multiple frames of target images, and the target authentication information includes a target vehicle viewing angle timing; A first determination module is used to determine the confidence of each frame of the target image under multiple vehicle perspectives based on the vehicle features corresponding to the target vehicle in each frame of the target image; A second determination module is used to determine the information to be authenticated corresponding to the video image based on the confidence of each frame of the target image under multiple vehicle perspectives, wherein the information to be authenticated includes a time sequence of the vehicle perspective to be authenticated; The third determination module is used to determine that the target vehicle is authenticated if the information to be authenticated matches the target authentication information.

12. The vehicle authentication device according to claim 11, characterized in that: The first determination module determines the confidence of each frame of the target image under multiple vehicle perspectives based on the vehicle features corresponding to the target vehicle in each frame of the target image, including: Extracting vehicle features corresponding to the target vehicle in each frame of the target image included in the video image; For each frame of target image included in the video image, the vehicle features corresponding to the target vehicle in the frame of target image are processed based on the first processing parameters to obtain the confidence of the frame of target image under multiple vehicle viewing angles.

13. The vehicle authentication device according to claim 12, characterized in that: The first determination module extracts the vehicle features corresponding to the target vehicle in each frame of the target image included in the video image, including: For each frame of target image included in the video image, extracting image features corresponding to the frame of target image; Based on the image features corresponding to the frame target image, the vehicle features corresponding to the target vehicle in the frame target image are determined.

14. The vehicle authentication device according to claim 13, characterized in that: The first determination module determines the vehicle features corresponding to the target vehicle in the frame target image based on the image features corresponding to the frame target image, including: Performing convolution processing on the image features corresponding to the target image of the frame based on the target convolution kernel to obtain the deep image features corresponding to the target image of the frame; The deep image features corresponding to the frame target image are processed to determine the vehicle features corresponding to the target vehicle in the frame target image.

15. The vehicle authentication device according to claim 14, characterized in that: The first determination module processes the deep image features corresponding to the frame target image to determine the vehicle features corresponding to the target vehicle in the frame target image, including: Processing the deep image feature corresponding to the frame target image based on the second processing parameter, and determining the first vehicle attribute feature corresponding to the target vehicle in the frame target image based on the obtained processing result; Processing the deep image feature corresponding to the frame target image based on the third processing parameter to obtain a second vehicle attribute feature corresponding to the target vehicle in the frame target image; Based on the first vehicle attribute feature and the second vehicle attribute feature, a vehicle feature corresponding to the target vehicle in the frame target image is determined.

16. The vehicle authentication device according to claim 15, characterized in that: The first determination module determines the first vehicle attribute feature corresponding to the target vehicle in the frame target image based on the obtained processing result, including: Determining first attribute information of the vehicle frame based on a first position of a vehicle frame corresponding to the target vehicle in the frame target image; Determine the third positions of the wheels in the vehicle frame based on the first position of the vehicle frame and the second positions of the wheels of the target vehicle in the target image frame. The first vehicle attribute feature corresponding to the target vehicle in the target image of the frame is determined according to the first attribute information of the vehicle frame, the wheel types corresponding to the wheels in the target vehicle and the relative positions of the wheels in the vehicle frame.

17. The vehicle authentication device according to claim 15, characterized in that: The device also includes: A second acquisition module is used to acquire a sample video image corresponding to the target vehicle; A first training module is used to train the second processing parameter by taking the image feature of each frame of the sample image in the sample video image as the first input feature and taking the second attribute information of the target vehicle corresponding to each frame of the sample image in the sample video image as the output; The second training module is used to train the first processing parameter and the third processing parameter using the image features included in each frame of the sample image in the sample video image as the second input feature, the vehicle features corresponding to the target vehicle in each frame of the sample image as the third input feature corresponding to the third processing parameter, and the authentication information of the target vehicle corresponding to the sample video image as the output.

18. The vehicle authentication device according to any one of claims 11 to 17, characterized in that: The second determination module determines the information to be authenticated corresponding to the video image based on the confidence of each frame of the target image under multiple vehicle perspectives, including: Determine a starting vehicle perspective corresponding to the first frame target image based on the confidence of the first frame target image under multiple vehicle perspectives respectively; wherein the multiple vehicle perspectives are arranged in a cyclic manner; Acquire multiple paths corresponding to the video image, which start from the starting vehicle perspective and pass through other vehicle perspectives and then return to the starting vehicle perspective; wherein each path includes multiple frames of target images in the video image, and each frame of the target image in the path corresponds to a vehicle perspective; Selecting a target path from the multiple paths based on the confidence of each frame of the target image under multiple vehicle perspectives; According to the target path, information to be authenticated corresponding to the video image is determined.

19. The vehicle authentication device according to claim 18, characterized in that: in, Taking the starting vehicle perspective as the starting point, a time sequence of two vehicle perspectives passes through other vehicle perspectives and then returns to the starting vehicle perspective; For any vehicle perspective sequence, the vehicle perspective corresponding to each frame of the target image is the first vehicle perspective corresponding to the previous frame of the target image or the next second vehicle perspective after the first vehicle perspective in the vehicle perspective sequence.

20. The vehicle authentication device according to claim 18, characterized in that: The second determination module selects a target path from the multiple paths based on the confidence of each frame of the target image under multiple vehicle perspectives, including: Determining a total confidence corresponding to each of the multiple paths based on the confidence of each frame of the target image under multiple vehicle perspectives; According to the total confidence corresponding to each path in the multiple paths, a target path with the maximum corresponding total confidence is selected from the multiple paths.

21. 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 authentication method as described in any one of claims 1 to 10.

22. 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 authentication method according to any one of claims 1 to 10 are executed.

Images