Method, device, equipment and storage medium for identifying vehicle painting methods

Abstract

The present invention relates to artificial intelligence, and provides a method, device, equipment and storage medium for identifying a vehicle painting method. The method obtains a vehicle image of a damaged vehicle, segments the vehicle image based on a component segmentation model and a rib line segmentation model to obtain a component contour and a rib line contour, locates a target contour according to the component contour and the rib line contour, detects the vehicle image based on a sheet metal part damage detection model to obtain a sheet metal part damage box, and if there is no intersection between the sheet metal part damage box and the target contour, a painting result can be accurately generated according to the sheet metal part damage box and the component contour. In addition, the present invention also relates to blockchain technology, and the painting result can be stored in the blockchain.

Description

Technical Field

[0001] The present invention relates to the technical field of artificial intelligence, and in particular, to a method, device, equipment and storage medium for identifying a vehicle painting method. Background Art

[0002] Currently, when determining the damage of a vehicle, the damaged parts and the type of damage are usually identified based on the images of the damaged vehicle uploaded by the insurance claimant. Then, the damaged vehicle is indemnified in combination with the damaged parts and the type of damage. However, for some damaged vehicles that need to be repainted, different painting methods will cause a large difference in the indemnity amount. Therefore, how to accurately identify the painting method of the damaged vehicle has become a technical problem that needs to be solved urgently. Summary of the Invention

[0003] In view of the above, it is necessary to provide a method, device, equipment and storage medium for identifying a vehicle painting method to solve the technical problem of how to accurately identify the painting method of the damaged vehicle.

[0004] On the one hand, the present invention provides a method for identifying a vehicle painting method, the vehicle painting method identification method comprising:

[0005] When a vehicle painting method identification request is received, a vehicle image of the damaged vehicle is obtained according to the vehicle painting method identification request;

[0006] Based on a pre-trained part segmentation model, the vehicle image is subjected to part segmentation processing to obtain multiple vehicle parts of the damaged vehicle and the part contour of each vehicle part;

[0007] Based on a pre-trained rib line segmentation model, the vehicle image is subjected to part segmentation processing to obtain multiple vehicle rib lines of the damaged vehicle and the rib line contour of each vehicle rib line;

[0008] According to the part contour of each vehicle part and the rib line contour of each vehicle rib line, the target contour of the target rib line among the multiple vehicle rib lines is located;

[0009] Based on a pre-trained sheet metal part damage detection model, the vehicle image is subjected to damage detection to obtain a sheet metal part damage box of the vehicle image;

[0010] It is detected whether there is an intersection between the sheet metal part damage box and the target contour;

[0011] If there is no intersection between the sheet metal part damage box and the target contour, a painting result of the damaged vehicle is generated according to the sheet metal part damage box and the part contour of each vehicle part.

[0012] According to a preferred embodiment of the present invention, the component segmentation model includes an encoding layer, a feature extraction layer, a standard convolutional layer, and a predicted contour output layer. The component segmentation of the vehicle image based on the pre-trained component segmentation model to obtain multiple vehicle components of the damaged vehicle and the component contour of each vehicle component includes:

[0013] Encoding the vehicle image based on the pixel vector mapping table in the encoding layer to obtain an image encoding vector;

[0014] Extracting first feature information of the image encoding vector based on the feature extraction layer;

[0015] Performing convolution processing on the first feature information based on the standard convolutional layer to obtain second feature information;

[0016] Fusing the first feature information and the second feature information to obtain fused feature information;

[0017] Performing mapping processing on the fused feature information based on the predicted contour output layer to obtain the component contour.

[0018] According to a preferred embodiment of the present invention, the positioning of the target contour of the target rib line among the multiple vehicle rib lines based on the component contour of each vehicle component and the rib line contour of each vehicle rib line includes:

[0019] Locating the component position of each vehicle component in the vehicle image according to the component contour;

[0020] Locating the rib line position of each vehicle rib line in the vehicle image according to the rib line contour;

[0021] Identifying the vehicle rib line within the component contour according to the rib line position and the component position as the target rib line;

[0022] Determining the rib line contour corresponding to the target rib line as the target contour.

[0023] According to a preferred embodiment of the present invention, the damage detection model includes a plurality of detection convolutional layers. The damage detection of the vehicle image based on the pre-trained sheet metal part damage detection model to obtain the sheet metal part damage box of the vehicle image includes:

[0024] Extracting features of the vehicle image based on the plurality of detection convolutional layers to obtain damage detection features of each detection convolutional layer;

[0025] Performing weighted sum operation on the damage detection features based on the network layer weights of the plurality of detection convolutional layers to obtain damage feature information of the vehicle image;

[0026] Identify a detection box corresponding to the damage feature information in the vehicle image as a damage detection box;

[0027] Locate the detection box position of the damage detection box in the vehicle image;

[0028] Obtain a preset sheet metal part, and obtain the sheet metal part position corresponding to the preset sheet metal part from the part positions;

[0029] According to the sheet metal part position and the detection box position, determine a damage detection box that intersects with the preset sheet metal part as the sheet metal part damage box.

[0030] According to a preferred embodiment of the present invention, the part contour corresponds to the part position, and generating the painting result of the damaged vehicle according to the sheet metal part damage box and the part contour of each vehicle part includes:

[0031] Obtain the sheet metal part damage position of the sheet metal part damage box from the detection box positions;

[0032] According to the sheet metal part damage position and the part position, count the number of parts of the vehicle parts included in the sheet metal part damage box;

[0033] If the number of parts is less than or equal to a preset number, determine that the painting result is a semi-spray method.

[0034] According to a preferred embodiment of the present invention, the vehicle painting method identification method further includes:

[0035] If the number of parts is greater than the preset number, or the sheet metal part damage box intersects with the target contour, determine that the painting result is a full-spray method.

[0036] According to a preferred embodiment of the present invention, detecting whether the sheet metal part damage box intersects with the target contour includes:

[0037] Obtain the first pixel points in the sheet metal part damage box from the vehicle image;

[0038] Obtain the second pixel points in the target contour from the vehicle image;

[0039] Detect whether there are pixel points in the first pixel points that are at the same pixel position as the second pixel points;

[0040] If there are pixel points in the first pixel points that are at the same pixel position as the second pixel points, determine that the sheet metal part damage box intersects with the target contour; or

[0041] If there is no pixel point in the first pixel point that is at the same pixel position as the second pixel point, it is determined that there is no intersection between the sheet metal part damage frame and the target contour.

[0042] On the other hand, the present invention also proposes a vehicle painting method recognition device, and the vehicle painting method recognition device includes:

[0043] An acquisition unit, configured to obtain a vehicle image of a damaged vehicle according to the vehicle painting method recognition request when receiving the vehicle painting method recognition request;

[0044] A segmentation unit, configured to perform component segmentation processing on the vehicle image based on a pre-trained component segmentation model to obtain multiple vehicle components of the damaged vehicle and the component contour of each vehicle component;

[0045] The segmentation unit is further configured to perform component segmentation processing on the vehicle image based on a pre-trained rib line segmentation model to obtain multiple vehicle rib lines of the damaged vehicle and the rib line contour of each vehicle rib line;

[0046] A positioning unit, configured to locate the target contour of the target rib line among the multiple vehicle rib lines according to the component contour of each vehicle component and the rib line contour of each vehicle rib line;

[0047] A detection unit, configured to perform damage detection on the vehicle image based on a pre-trained sheet metal part damage detection model to obtain a sheet metal part damage frame of the vehicle image;

[0048] The detection unit is further configured to detect whether there is an intersection between the sheet metal part damage frame and the target contour;

[0049] A generation unit, configured to generate a painting result of the damaged vehicle according to the sheet metal part damage frame and the component contour of each vehicle component if there is no intersection between the sheet metal part damage frame and the target contour.

[0050] On the other hand, the present invention also proposes an electronic device, and the electronic device includes:

[0051] A memory, storing computer-readable instructions; and

[0052] A processor, executing the computer-readable instructions stored in the memory to implement the vehicle painting method recognition method.

[0053] On the other hand, the present invention also proposes a computer-readable storage medium, and computer-readable instructions are stored in the computer-readable storage medium, and the computer-readable instructions are executed by a processor in an electronic device to implement the vehicle painting method recognition method.

[0054] As can be seen from the above technical solutions, the present application can accurately locate the target contour of the target rib line by combining the component contour of each vehicle component and the rib line contour of each vehicle rib line. Since the target rib line has a certain number of rib lines reduced compared to the multiple vehicle rib lines, it is not necessary to detect all vehicle rib lines, which can improve the intersection detection efficiency between the sheet metal part damage box and the target contour, thereby improving the generation efficiency of the painting result. By further combining the component contour of each vehicle component to detect the sheet metal part damage box when there is no intersection between the sheet metal part damage box and the target contour, the generation accuracy of the painting result can be improved, and thus the accuracy of vehicle insurance claims can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 is a flowchart of a preferred embodiment of the method for identifying vehicle painting methods of the present invention.

[0056] Figure 2 is a functional module diagram of a preferred embodiment of the device for identifying vehicle painting methods of the present invention.

[0057] Figure 3 is a schematic structural diagram of an electronic device of a preferred embodiment for implementing the method for identifying vehicle painting methods of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0058] In order to make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0059] As Figure 1 shown, it is a flowchart of a preferred embodiment of the method for identifying vehicle painting methods of the present invention. According to different requirements, the order of steps in this flowchart can be changed, and some steps can be omitted.

[0060] The method for identifying vehicle painting methods can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is to use a digital computer or a machine controlled by a digital computer to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use knowledge to obtain the best results of theory, method, technology, and application system.

[0061] Artificial intelligence basic technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, and mechatronics. Artificial intelligence software technologies mainly include several major directions such as computer vision technology, robotics, biometric technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0062] The vehicle painting method recognition method is applied to one or more electronic devices. The electronic device is a device capable of automatically performing numerical calculations and / or information processing according to pre-set or stored computer-readable instructions. Its hardware includes, but is not limited to, a microprocessor, an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA), a digital signal processor (DSP), an embedded device, etc.

[0063] The electronic device can be any electronic product that can perform human-computer interaction with the user. For example, a personal computer, a tablet computer, a smart phone, a personal digital assistant (PDA), a game console, an Internet Protocol Television (IPTV), a smart wearable device, etc.

[0064] The electronic device may include a network device and / or a user device. Among them, the network device includes, but is not limited to, a single network electronic device, a group of electronic devices composed of multiple network electronic devices, or a cloud composed of a large number of hosts or network electronic devices based on cloud computing (Cloud Computing).

[0065] The network where the electronic device is located includes, but is not limited to: the Internet, a wide area network, a metropolitan area network, a local area network, a virtual private network (VPN), etc.

[0066] S10. When a vehicle painting method recognition request is received, obtain a vehicle image of the damaged vehicle according to the vehicle painting method recognition request.

[0067] In at least one embodiment of the present invention, the vehicle painting method recognition request may be triggered and generated by a staff member responsible for insurance claims, or may be triggered and generated by a staff member repairing the damaged vehicle, or may be triggered and generated by the holding user of the damaged vehicle. The vehicle painting method recognition request may also be triggered and generated after the holding user uploads the vehicle image.

[0068] The damaged vehicle refers to a vehicle that needs to be repaired or claimed. The vehicle image refers to an image obtained by photographing the damaged vehicle in a damaged state.

[0069] In at least one embodiment of the present invention, the electronic device obtains a vehicle image of the damaged vehicle according to the vehicle painting method recognition request, including:

[0070] Extract the request time, vehicle identification code, and image storage path from the vehicle painting method recognition request;

[0071] Obtain the image corresponding to the request time and the vehicle identification code simultaneously from the image storage path as the vehicle image.

[0072] Among them, the request time refers to the time point when the vehicle painting method recognition request is triggered. The request time is the same as the time point when the vehicle image is uploaded to the image library.

[0073] By combining the request time and the vehicle identification code, the vehicle image can be accurately obtained. At the same time, since the vehicle image can be directly obtained from the image storage path, the acquisition efficiency of the vehicle image can be improved.

[0074] S11, perform component segmentation processing on the vehicle image based on a pre-trained component segmentation model to obtain multiple vehicle components of the damaged vehicle and the component contour of each vehicle component.

[0075] In at least one embodiment of the present invention, the component segmentation model includes an encoding layer, a feature extraction layer, a standard convolutional layer, and a predicted contour output layer. The component segmentation model is used to segment the component contour corresponding to the vehicle components in the vehicle image.

[0076] The multiple vehicle components include fenders, trunk lids, vehicle lights, etc. The component contour refers to the line forming the outer edge of the corresponding vehicle component.

[0077] In at least one embodiment of the present invention, the electronic device performs component segmentation processing on the vehicle image based on a pre-trained component segmentation model to obtain multiple vehicle components of the damaged vehicle and the component contour of each vehicle component, including:

[0078] Perform encoding processing on the vehicle image based on the pixel vector mapping table in the encoding layer to obtain an image encoding vector;

[0079] Extract the first feature information of the image encoding vector based on the feature extraction layer;

[0080] Perform convolutional processing on the first feature information based on the standard convolutional layer to obtain second feature information;

[0081] Fuse the first feature information and the second feature information to obtain fused feature information;

[0082] Perform mapping processing on the fused feature information based on the predicted contour output layer to obtain the component contour.

[0083] Among them, the pixel vector mapping table stores the mapping relationship between pixel values and vectors.

[0084] Through the pixel vector mapping table, the vehicle image can be accurately encoded, improving the representation ability of the image encoding vector for the vehicle image. Based on the feature extraction layer and the standard convolutional layer, the feature information of the vehicle image can be respectively extracted. Further, by combining the first feature information and the second feature information, the fused feature information of the vehicle image can be accurately generated, thereby improving the segmentation accuracy of the component contour.

[0085] Specifically, the electronic device fuses the first feature information and the second feature information to obtain the fused feature information, including:

[0086] Performing dimensionality reduction processing on the first feature information to obtain low-dimensional features;

[0087] Performing upsampling processing on the second feature information to obtain sampled features, and the dimension of the sampled features is equal to the dimension of the low-dimensional features;

[0088] Calculating the average value of the low-dimensional features and the sampled features in each vector dimension to obtain the fused feature information.

[0089] By performing dimensionality reduction processing on the first feature information and upsampling processing on the second feature information, it can ensure that the dimension of the sampled features is equal to the dimension of the low-dimensional features, which is beneficial to the generation of the fused feature information.

[0090] S12. Based on the pre-trained tendon line segmentation model, perform component segmentation processing on the vehicle image to obtain multiple vehicle tendon lines of the damaged vehicle and the tendon line contour of each vehicle tendon line.

[0091] In at least one embodiment of the present invention, the tendon line segmentation model is used to segment the tendon line contour corresponding to the vehicle tendon line in the vehicle image.

[0092] In this embodiment, since the network structure of the tendon line segmentation model is similar to the network structure of the component segmentation model, the specific manner in which the electronic device performs component segmentation processing on the vehicle image based on the tendon line segmentation model will not be elaborated in this application.

[0093] S13. Locate the target contour of the target tendon line among the multiple vehicle tendon lines according to the component contour of each vehicle component and the tendon line contour of each vehicle tendon line.

[0094] In at least one embodiment of the present invention, the target tendon line refers to the vehicle tendon line within the component contour.

[0095] In at least one embodiment of the present invention, the electronic device locates the target contour of the target rib line among the multiple vehicle rib lines according to the part contour of each vehicle part and the rib line contour of each vehicle rib line, including:

[0096] Locate the part position of each vehicle part in the vehicle image according to the part contour;

[0097] Locate the rib line position of each vehicle rib line in the vehicle image according to the rib line contour;

[0098] Identify the vehicle rib line within the part contour as the target rib line according to the rib line position and the part position;

[0099] Determine the rib line contour corresponding to the target rib line as the target contour.

[0100] By combining the rib line position and the part position, the vehicle rib line within the part contour can be accurately identified, thereby improving the screening accuracy of the target rib line and thus improving the accuracy of the target contour.

[0101] S14. Perform damage detection on the vehicle image based on a pre-trained sheet metal part damage detection model to obtain a sheet metal part damage box of the vehicle image.

[0102] In at least one embodiment of the present invention, the damage detection model includes multiple detection convolutional layers.

[0103] The sheet metal part damage box refers to the damage detection box corresponding to the sheet metal part in the damaged vehicle. The damaged sheet metal part is included in the sheet metal part damage box.

[0104] In at least one embodiment of the present invention, the electronic device performs damage detection on the vehicle image based on a pre-trained sheet metal part damage detection model to obtain a sheet metal part damage box of the vehicle image, including:

[0105] Extract features from the vehicle image based on the multiple detection convolutional layers to obtain damage detection features of each detection convolutional layer;

[0106] Perform a weighted sum operation on the damage detection features based on the network layer weights of the multiple detection convolutional layers to obtain damage feature information of the vehicle image;

[0107] Identify a detection box corresponding to the damage feature information from the vehicle image as a damage detection box;

[0108] Locate the detection box position of the damage detection box in the vehicle image;

[0109] Obtain a preset sheet metal part, and obtain the sheet metal part position corresponding to the preset sheet metal part from the part positions;

[0110] According to the sheet metal part position and the detection frame position, determine the damage detection frame that intersects with the preset sheet metal part as the sheet metal part damage frame.

[0111] Wherein, the network layer weight refers to the weight corresponding to each detection convolutional layer in the sheet metal part damage detection model, and the network layer weight can be determined according to the detection ability or detection loss value of each detection convolutional layer in the sheet metal part damage detection model.

[0112] The preset sheet metal part generally includes vehicle parts such as fenders and trunk lids.

[0113] By combining the multiple detection convolutional layers and the network layer weights, the damage feature information of the vehicle image can be accurately extracted, thereby improving the recognition accuracy of the damage detection frame. Further, through the sheet metal part position and the detection frame position, the damage detection frame that intersects with the preset sheet metal part can be directly determined, improving the determination efficiency of the sheet metal part damage frame.

[0114] S15, detect whether there is an intersection between the sheet metal part damage frame and the target contour.

[0115] In at least one embodiment of the present invention, the electronic device detecting whether there is an intersection between the sheet metal part damage frame and the target contour includes:

[0116] Obtain the first pixel points in the sheet metal part damage frame from the vehicle image;

[0117] Obtain the second pixel points in the target contour from the vehicle image;

[0118] Detect whether there are pixel points in the first pixel points that are at the same pixel position as the second pixel points;

[0119] If there are pixel points in the first pixel points that are at the same pixel position as the second pixel points, determine that there is an intersection between the sheet metal part damage frame and the target contour; or

[0120] If there are no pixel points in the first pixel points that are at the same pixel position as the second pixel points, determine that there is no intersection between the sheet metal part damage frame and the target contour.

[0121] Wherein, the first pixel points refer to the pixel points that form the sheet metal part damage frame in the vehicle image, and the second pixel points refer to the pixel points that form the target contour in the vehicle image.

[0122] Through the above embodiments, since the rib lines are relatively small features compared to the vehicle, the detection accuracy can be improved by detecting whether there is an intersection between the damaged frame of the sheet metal part and the target contour based on pixel points.

[0123] S16. If there is no intersection between the damaged frame of the sheet metal part and the target contour, generate the painting result of the damaged vehicle according to the damaged frame of the sheet metal part and the part contours of each vehicle part.

[0124] It should be emphasized that to further ensure the privacy and security of the above painting result, the above painting result can also be stored in a node of a blockchain.

[0125] In at least one embodiment of the present invention, the painting result includes a partial painting method and a full painting method. Among them, the partial painting method means performing local painting on the damaged vehicle, and the full painting method means performing overall painting on the damaged vehicle.

[0126] In at least one embodiment of the present invention, the part contour corresponds to the part position, and the electronic device generates the painting result of the damaged vehicle according to the damaged frame of the sheet metal part and the part contours of each vehicle part, including:

[0127] Obtain the damaged position of the sheet metal part of the damaged frame of the sheet metal part from the position of the detection frame;

[0128] According to the damaged position of the sheet metal part and the part position, count the number of vehicle parts included in the damaged frame of the sheet metal part;

[0129] If the number of parts is less than or equal to a preset number, determine that the painting result is the partial painting method.

[0130] Among them, the preset number is usually set to 1.

[0131] When there is no intersection between the damaged frame of the sheet metal part and the target contour, it indicates that there is no cross-part damage to the damaged vehicle. Further, by comparing the relationship between the number of parts and the preset number, it is possible to avoid inaccurate painting results caused by multiple non-adjacent part damages in the damaged vehicle.

[0132] In at least one embodiment of the present invention, the vehicle painting method identification method further includes:

[0133] If the number of parts is greater than the preset number, or there is an intersection between the damaged frame of the sheet metal part and the target contour, determine that the painting result is the full painting method.

[0134] By the above embodiments, directly determining the painting result as the full painting method can not only avoid color difference in the damaged parts after repair caused by too large damaged area or crossing the rib lines on the vehicle parts, but also avoid inaccurate vehicle insurance claims settlement results.

[0135] As can be seen from the above technical solutions, the present application can accurately locate the target contour of the target rib line by combining the part contour of each vehicle part and the rib line contour of each vehicle rib line. Since the target rib line has a certain number of rib lines less than the multiple vehicle rib lines, it is not necessary to detect all vehicle rib lines, which can improve the intersection detection efficiency between the sheet metal part damage frame and the target contour, thereby improving the generation efficiency of the painting result. When there is no intersection between the sheet metal part damage frame and the target contour, further detecting the sheet metal part damage frame by combining the part contour of each vehicle part can improve the generation accuracy of the painting result, and thus can improve the accuracy of vehicle insurance claims settlement.

[0136] As Figure 2 shown, it is a functional module diagram of a preferred embodiment of the vehicle painting method recognition device of the present invention. The vehicle painting method recognition device 11 includes an acquisition unit 110, a segmentation unit 111, a positioning unit 112, a detection unit 113 and a generation unit 114. The module / unit referred to in the present invention means a series of computer-readable instruction segments that can be acquired by a processor 13 and can complete fixed functions, and are stored in a memory 12. In this embodiment, the functions of each module / unit will be described in detail in subsequent embodiments.

[0137] When receiving a vehicle painting method recognition request, the acquisition unit 110 acquires a vehicle image of the damaged vehicle according to the vehicle painting method recognition request.

[0138] In at least one embodiment of the present invention, the vehicle painting method recognition request can be triggered and generated by a staff member responsible for insurance claims settlement, or can be triggered and generated by a staff member repairing the damaged vehicle, or can be triggered and generated by the holding user of the damaged vehicle. The vehicle painting method recognition request can also be triggered and generated after the holding user uploads the vehicle image.

[0139] The damaged vehicle refers to a vehicle that needs to be repaired or claimed, and the vehicle image refers to an image obtained by photographing the damaged vehicle in a damaged state.

[0140] In at least one embodiment of the present invention, the acquisition unit 110 acquiring a vehicle image of the damaged vehicle according to the vehicle painting method recognition request includes:

[0141] Extract the request time, vehicle identification code, and image storage path from the vehicle paint spraying method recognition request;

[0142] Obtain the image corresponding to the request time and the vehicle identification code simultaneously from the image storage path as the vehicle image.

[0143] Wherein, the request time refers to the time point when the vehicle paint spraying method recognition request is triggered. The request time is the same as the time point when the vehicle image is uploaded to the image library.

[0144] By combining the request time and the vehicle identification code, the vehicle image can be accurately obtained. At the same time, since the vehicle image can be directly obtained from the image storage path, the acquisition efficiency of the vehicle image can be improved.

[0145] The segmentation unit 111 performs component segmentation processing on the vehicle image based on a pre-trained component segmentation model to obtain multiple vehicle components of the damaged vehicle and the component contour of each vehicle component.

[0146] In at least one embodiment of the present invention, the component segmentation model includes an encoding layer, a feature extraction layer, a standard convolution layer, and a predicted contour output layer. The component segmentation model is used to segment the component contour corresponding to the vehicle components in the vehicle image.

[0147] The multiple vehicle components include fenders, trunk lids, headlights, etc. The component contour refers to the line forming the outer edge of the corresponding vehicle component.

[0148] In at least one embodiment of the present invention, the segmentation unit 111 performing component segmentation processing on the vehicle image based on a pre-trained component segmentation model to obtain multiple vehicle components of the damaged vehicle and the component contour of each vehicle component includes:

[0149] Perform encoding processing on the vehicle image based on the pixel vector mapping table in the encoding layer to obtain an image encoding vector;

[0150] Extract the first feature information of the image encoding vector based on the feature extraction layer;

[0151] Perform convolution processing on the first feature information based on the standard convolution layer to obtain second feature information;

[0152] Fuse the first feature information and the second feature information to obtain fused feature information;

[0153] Perform mapping processing on the fused feature information based on the predicted contour output layer to obtain the component contour.

[0154] Among them, the pixel vector mapping table stores the mapping relationship between pixel values and vectors.

[0155] Through the pixel vector mapping table, the vehicle image can be accurately encoded, improving the representation ability of the image encoding vector for the vehicle image. Based on the feature extraction layer and the standard convolutional layer, the feature information of the vehicle image can be respectively extracted. Further, by combining the first feature information and the second feature information, the fused feature information of the vehicle image can be accurately generated, thereby improving the segmentation accuracy of the component contour.

[0156] Specifically, the segmentation unit 111 fuses the first feature information and the second feature information, and the obtained fused feature information includes:

[0157] Perform dimensionality reduction processing on the first feature information to obtain low-dimensional features;

[0158] Perform upsampling processing on the second feature information to obtain sampled features, and the dimension of the sampled features is equal to the dimension of the low-dimensional features;

[0159] Calculate the average value of the low-dimensional features and the sampled features in each vector dimension to obtain the fused feature information.

[0160] By performing dimensionality reduction processing on the first feature information and upsampling processing on the second feature information, it can ensure that the dimension of the sampled features is equal to the dimension of the low-dimensional features, which is beneficial to the generation of the fused feature information.

[0161] The segmentation unit 111 performs component segmentation processing on the vehicle image based on a pre-trained rib line segmentation model to obtain multiple vehicle rib lines of the damaged vehicle and the rib line contour of each vehicle rib line.

[0162] In at least one embodiment of the present invention, the rib line segmentation model is used to segment the rib line contour corresponding to the vehicle rib line in the vehicle image.

[0163] In this embodiment, since the network structure of the rib line segmentation model is similar to the network structure of the component segmentation model, therefore, the specific manner in which the segmentation unit 111 performs component segmentation processing on the vehicle image based on the rib line segmentation model will not be elaborated in this application.

[0164] The positioning unit 112 locates the target contour of the target rib line among the multiple vehicle rib lines according to the component contour of each vehicle component and the rib line contour of each vehicle rib line.

[0165] In at least one embodiment of the present invention, the target rib line refers to the vehicle rib line within the component contour.

[0166] In at least one embodiment of the present invention, the positioning unit 112 locating the target contour of the target rib line among the plurality of vehicle rib lines according to the part contour of each vehicle part and the rib line contour of each vehicle rib line includes:

[0167] Locating the part position of each vehicle part in the vehicle image according to the part contour;

[0168] Locating the rib line position of each vehicle rib line in the vehicle image according to the rib line contour;

[0169] Identifying the vehicle rib line within the part contour as the target rib line according to the rib line position and the part position;

[0170] Determining the rib line contour corresponding to the target rib line as the target contour.

[0171] By combining the rib line position and the part position, the vehicle rib line within the part contour can be accurately identified, thereby improving the screening accuracy of the target rib line and thus improving the accuracy of the target contour.

[0172] The detection unit 113 performs damage detection on the vehicle image based on a pre-trained sheet metal part damage detection model to obtain a sheet metal part damage box of the vehicle image.

[0173] In at least one embodiment of the present invention, the damage detection model includes a plurality of detection convolutional layers.

[0174] The sheet metal part damage box refers to the damage detection box corresponding to the sheet metal part in the damaged vehicle. The damaged sheet metal part is included in the sheet metal part damage box.

[0175] In at least one embodiment of the present invention, the detection unit 113 performing damage detection on the vehicle image based on a pre-trained sheet metal part damage detection model to obtain a sheet metal part damage box of the vehicle image includes:

[0176] Performing feature extraction on the vehicle image based on the plurality of detection convolutional layers to obtain damage detection features of each detection convolutional layer;

[0177] Performing a weighted sum operation on the damage detection features based on the network layer weights of the plurality of detection convolutional layers to obtain damage feature information of the vehicle image;

[0178] Identifying a detection box corresponding to the damage feature information in the vehicle image as a damage detection box;

[0179] Locating the detection box position of the damage detection box in the vehicle image;

[0180] Obtain a preset sheet metal part, and obtain the sheet metal part position corresponding to the preset sheet metal part from the part positions.

[0181] According to the sheet metal part position and the detection frame position, determine the damage detection frame that intersects with the preset sheet metal part as the sheet metal part damage frame.

[0182] Wherein, the network layer weight refers to the weight corresponding to each detection convolutional layer in the sheet metal part damage detection model, and the network layer weight can be determined according to the detection ability or detection loss value of each detection convolutional layer in the sheet metal part damage detection model.

[0183] The preset sheet metal part generally includes vehicle parts such as fenders and trunk lids.

[0184] By combining the multiple detection convolutional layers and the network layer weights, the damage feature information of the vehicle image can be accurately extracted, thereby improving the recognition accuracy of the damage detection frame. Further, through the sheet metal part position and the detection frame position, the damage detection frame that intersects with the preset sheet metal part can be directly determined, improving the determination efficiency of the sheet metal part damage frame.

[0185] The detection unit 113 detects whether there is an intersection between the sheet metal part damage frame and the target contour.

[0186] In at least one embodiment of the present invention, the electronic device detecting whether there is an intersection between the sheet metal part damage frame and the target contour includes:

[0187] Obtain the first pixel point in the sheet metal part damage frame from the vehicle image;

[0188] Obtain the second pixel point in the target contour from the vehicle image;

[0189] Detect whether there is a pixel point in the first pixel points that is at the same pixel position as the second pixel point;

[0190] If there is a pixel point in the first pixel points that is at the same pixel position as the second pixel point, determine that there is an intersection between the sheet metal part damage frame and the target contour; or

[0191] If there is no pixel point in the first pixel points that is at the same pixel position as the second pixel point, determine that there is no intersection between the sheet metal part damage frame and the target contour.

[0192] Among them, the first pixel points refer to the pixel points in the vehicle image that form the damage frame of the sheet metal part, and the second pixel points refer to the pixel points in the vehicle image that form the target contour.

[0193] Through the above embodiments, since the rib lines are relatively small in feature compared to the vehicle, by detecting whether there is an intersection between the damage frame of the sheet metal part and the target contour based on pixel points, the detection accuracy can be improved.

[0194] If there is no intersection between the damage frame of the sheet metal part and the target contour, the generating unit 114 generates the painting result of the damaged vehicle according to the damage frame of the sheet metal part and the part contour of each vehicle part.

[0195] It should be emphasized that to further ensure the privacy and security of the above painting result, the above painting result can also be stored in a node of a blockchain.

[0196] In at least one embodiment of the present invention, the painting result includes a partial painting method and a full painting method. Among them, the partial painting method refers to performing local painting on the damaged vehicle, and the full painting method refers to performing overall painting on the damaged vehicle.

[0197] In at least one embodiment of the present invention, the part contour corresponds to the part position, and the generating unit 114 generating the painting result of the damaged vehicle according to the damage frame of the sheet metal part and the part contour of each vehicle part includes:

[0198] Obtain the damage position of the sheet metal part of the damage frame of the sheet metal part from the detection frame position;

[0199] According to the damage position and the part position, count the number of vehicle parts included in the damage frame of the sheet metal part;

[0200] If the number of parts is less than or equal to a preset number, determine that the painting result is the partial painting method.

[0201] Among them, the preset number is usually set to 1.

[0202] When there is no intersection between the damage frame of the sheet metal part and the target contour, it indicates that there is no cross-part damage to the damaged vehicle. Further, by comparing the relationship between the number of parts and the preset number, it is possible to avoid inaccurate painting results caused by multiple non-adjacent part damages in the damaged vehicle.

[0203] In at least one embodiment of the present invention, if the number of parts is greater than the preset number, or there is an intersection between the damage frame of the sheet metal part and the target contour, the generating unit 114 determines that the painting result is the full painting method.

[0204] By the above embodiments, directly determining the painting result as the full painting method can not only avoid color difference in the damaged parts after repair caused by an overly large damaged area or crossing the rib lines on the vehicle parts, but also avoid inaccurate vehicle insurance claims settlement results.

[0205] It can be seen from the above technical solutions that the present application can accurately locate the target contour of the target rib line by combining the part contour of each vehicle part and the rib line contour of each vehicle rib line. Since the target rib line has a certain number of rib lines reduced compared to the multiple vehicle rib lines, it is not necessary to detect all vehicle rib lines, which can improve the intersection detection efficiency between the sheet metal part damage frame and the target contour, thereby improving the generation efficiency of the painting result. When there is no intersection between the sheet metal part damage frame and the target contour, further combining the part contour of each vehicle part to detect the sheet metal part damage frame can improve the generation accuracy of the painting result, and thus improve the accuracy of vehicle insurance claims settlement.

[0206] As Figure 3 shown, it is a schematic structural diagram of an electronic device of a preferred embodiment for implementing the vehicle painting method recognition method of the present invention.

[0207] In an embodiment of the present invention, the electronic device 1 includes, but is not limited to, a memory 12, a processor 13, and computer-readable instructions stored in the memory 12 and executable on the processor 13, such as a vehicle painting method recognition program.

[0208] Those skilled in the art can understand that the schematic diagram is only an example of the electronic device 1, and does not constitute a limitation on the electronic device 1. It may include more or fewer components than shown, or combine certain components, or different components. For example, the electronic device 1 may further include input / output devices, network access devices, buses, etc.

[0209] The processor 13 may be a Central Processing Unit (CPU), or may also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The processor 13 is the operation core and control center of the electronic device 1, connecting various parts of the entire electronic device 1 through various interfaces and circuits, and executing the operating system of the electronic device 1 and various installed application programs, program codes, etc.

[0210] Exemplarily, the computer-readable instructions may be divided into one or more modules / units. The one or more modules / units are stored in the memory 12 and executed by the processor 13 to complete the present invention. The one or more modules / units may be a series of computer-readable instruction segments capable of performing specific functions, and these computer-readable instruction segments are used to describe the execution process of the computer-readable instructions in the electronic device 1. For example, the computer-readable instructions may be divided into an acquisition unit 110, a segmentation unit 111, a positioning unit 112, a detection unit 113, and a generation unit 114.

[0211] The memory 12 can be used to store the computer-readable instructions and / or modules. The processor 13 realizes various functions of the electronic device 1 by running or executing the computer-readable instructions and / or modules stored in the memory 12, and by calling the data stored in the memory 12. The memory 12 may mainly include a program storage area and a data storage area. Among them, the program storage area may store the operating system, application programs required for at least one function (such as a sound playback function, an image playback function, etc.); the data storage area may store data created according to the use of the electronic device. The memory 12 may include non-volatile and volatile memories, such as: hard disks, memories, plug-in hard disks, SmartMedia Cards (SMCs), Secure Digital (SD) cards, Flash Cards, at least one magnetic disk storage device, flash memory device, or other storage devices.

[0212] The memory 12 can be an external memory and / or an internal memory of the electronic device 1. Further, the memory 12 can be a memory in physical form, such as a memory stick, a TF card (Trans-flash Card), etc.

[0213] If the modules / units integrated in the electronic device 1 are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, to implement all or part of the processes in the above-described embodiment methods of the present invention, it can also be completed by instructing relevant hardware through computer-readable instructions. The computer-readable instructions can be stored in a computer-readable storage medium. When the computer-readable instructions are executed by a processor, the steps of the above various method embodiments can be implemented.

[0214] Among them, the computer-readable instructions include computer-readable instruction codes, and the computer-readable instruction codes can be in the form of source code, object code, executable files, or some intermediate forms, etc. The computer-readable medium can include: any entity or device capable of carrying the computer-readable instruction codes, recording media, USB flash drives, mobile hard disks, magnetic disks, optical disks, computer memories, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory).

[0215] The blockchain referred to in the present invention is a new application mode of computer technologies such as distributed vehicle painting method identification, peer-to-peer transmission, consensus mechanism, and encryption algorithm. Blockchain, in essence, is a decentralized database, a string of data blocks generated by using cryptographic methods. Each data block contains information about a batch of network transactions, used to verify the validity of the information (anti-counterfeiting) and generate the next block. The blockchain can include a blockchain underlying platform, a platform product service layer, and an application service layer, etc.

[0216] Combined with Figure 1 , the memory 12 in the electronic device 1 stores computer-readable instructions to implement a vehicle painting method identification method, and the processor 13 can execute the computer-readable instructions to thereby implement:

[0217] When receiving a vehicle painting method identification request, obtain a vehicle image of the damaged vehicle according to the vehicle painting method identification request;

[0218] Based on a pre-trained component segmentation model, perform component segmentation processing on the vehicle image to obtain multiple vehicle components of the damaged vehicle and the component contours of each vehicle component;

[0219] Perform component segmentation processing on the vehicle image based on a pre-trained rib line segmentation model to obtain multiple vehicle rib lines of the damaged vehicle and the rib line contour of each vehicle rib line;

[0220] Locate the target contour of the target rib line among the multiple vehicle rib lines according to the component contour of each vehicle component and the rib line contour of each vehicle rib line;

[0221] Perform damage detection on the vehicle image based on a pre-trained sheet metal part damage detection model to obtain the sheet metal part damage box of the vehicle image;

[0222] Detect whether there is an intersection between the sheet metal part damage box and the target contour;

[0223] If there is no intersection between the sheet metal part damage box and the target contour, generate the painting result of the damaged vehicle according to the sheet metal part damage box and the component contour of each vehicle component.

[0224] Specifically, for the specific implementation method of the above computer-readable instructions by the processor 13, reference can be made to Figure 1 the description of the relevant steps in the corresponding embodiments, which will not be elaborated here.

[0225] In several embodiments provided by the present invention, it should be understood that the disclosed system, device and method can be implemented in other ways. For example, the device embodiments described above are only illustrative. For example, the division of the modules is only a logical function division, and there may be other division methods in actual implementation.

[0226] Computer-readable instructions are stored on the computer-readable storage medium, wherein when the computer-readable instructions are executed by the processor 13, the following steps are implemented:

[0227] When a vehicle painting method recognition request is received, obtain the vehicle image of the damaged vehicle according to the vehicle painting method recognition request;

[0228] Perform component segmentation processing on the vehicle image based on a pre-trained component segmentation model to obtain multiple vehicle components of the damaged vehicle and the component contour of each vehicle component;

[0229] Perform component segmentation processing on the vehicle image based on a pre-trained rib line segmentation model to obtain multiple vehicle rib lines of the damaged vehicle and the rib line contour of each vehicle rib line;

[0230] Locate the target contour of the target rib line among the multiple vehicle rib lines according to the component contour of each vehicle component and the rib line contour of each vehicle rib line;

[0231] Perform damage detection on the vehicle image based on a pre-trained sheet metal part damage detection model to obtain the sheet metal part damage box of the vehicle image;

[0232] Detect whether there is an intersection between the sheet metal part damage box and the target contour;

[0233] If there is no intersection between the sheet metal part damage box and the target contour, generate a painting result for the damaged vehicle according to the sheet metal part damage box and the part contour of each vehicle part.

[0234] The module described as a separate component description may or may not be physically separated. The components shown as modules may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0235] In addition, in each embodiment of the present invention, the functional modules can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above integrated unit can be implemented in the form of hardware or in the form of a hardware plus software functional module.

[0236] Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be encompassed by the present invention. Any associated drawing marks in the claims should not be regarded as limiting the claims involved.

[0237] In addition, obviously the word "including" does not exclude other units or steps, and the singular does not exclude the plural. The described multiple units or devices can also be implemented by one unit or device through software or hardware. Words such as first and second are used to represent names and do not represent any specific order.

[0238] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for identifying a vehicle painting method, characterized in that, the vehicle painting method identification method includes: When receiving a vehicle painting method identification request, obtaining a vehicle image of the damaged vehicle according to the vehicle painting method identification request; Performing component segmentation processing on the vehicle image based on a pre-trained component segmentation model to obtain multiple vehicle components of the damaged vehicle and the component contour of each vehicle component; Performing component segmentation processing on the vehicle image based on a pre-trained rib line segmentation model to obtain multiple vehicle rib lines of the damaged vehicle and the rib line contour of each vehicle rib line; Locating the target contour of the target rib line among the multiple vehicle rib lines according to the component contour of each vehicle component and the rib line contour of each vehicle rib line; Performing damage detection on the vehicle image based on a pre-trained sheet metal part damage detection model to obtain a sheet metal part damage box of the vehicle image; Detecting whether there is an intersection between the sheet metal part damage box and the target contour, including: obtaining a first pixel point in the sheet metal part damage box from the vehicle image; obtaining a second pixel point in the target contour from the vehicle image; detecting whether there is a pixel point in the first pixel points that is at the same pixel position as the second pixel point; if there is a pixel point in the first pixel points that is at the same pixel position as the second pixel point, it is determined that there is an intersection between the sheet metal part damage box and the target contour; or if there is no pixel point in the first pixel points that is at the same pixel position as the second pixel point, it is determined that there is no intersection between the sheet metal part damage box and the target contour; If there is no intersection between the sheet metal part damage box and the target contour, generating a painting result of the damaged vehicle according to the sheet metal part damage box and the component contour of each vehicle component.

2. The vehicle painting method identification method according to claim 1, characterized in that, the component segmentation model includes an encoding layer, a feature extraction layer, a standard convolution layer and a prediction contour output layer, and the performing component segmentation processing on the vehicle image based on a pre-trained component segmentation model to obtain multiple vehicle components of the damaged vehicle and the component contour of each vehicle component includes: Performing encoding processing on the vehicle image based on the pixel vector mapping table in the encoding layer to obtain an image encoding vector; Extracting first feature information of the image encoding vector based on the feature extraction layer; Performing convolution processing on the first feature information based on the standard convolution layer to obtain second feature information; Fusing the first feature information and the second feature information to obtain fused feature information; Performing mapping processing on the fused feature information based on the prediction contour output layer to obtain the component contour.

3. The vehicle painting method identification method according to claim 1, characterized in that, the locating the target contour of the target rib line among the multiple vehicle rib lines according to the component contour of each vehicle component and the rib line contour of each vehicle rib line includes: Locating the component position of each vehicle component in the vehicle image according to the component contour; Locate the position of each vehicle rib line in the vehicle image according to the rib line contour; Identify the vehicle rib lines within the component contour as the target rib lines based on the rib line position and the component position; Determine the rib line contour corresponding to the target rib lines as the target contour.

4. The vehicle painting method identification method according to claim 3, characterized in that the damage detection model includes a plurality of detection convolutional layers, and performing damage detection on the vehicle image based on the pre-trained sheet metal part damage detection model to obtain the sheet metal part damage box of the vehicle image includes: Performing feature extraction on the vehicle image based on the plurality of detection convolutional layers to obtain damage detection features of each detection convolutional layer; Performing weighted sum operation on the damage detection features based on the network layer weights of the plurality of detection convolutional layers to obtain damage feature information of the vehicle image; Identifying a detection box corresponding to the damage feature information in the vehicle image as a damage detection box; Locating the detection box position of the damage detection box in the vehicle image; Obtaining a preset sheet metal part, and obtaining the sheet metal part position corresponding to the preset sheet metal part from the component positions; According to the sheet metal part position and the detection box position, determining the damage detection box having an intersection with the preset sheet metal part as the sheet metal part damage box.

5. The vehicle painting method identification method according to claim 4, characterized in that the component contour corresponds to the component position, and generating the painting result of the damaged vehicle according to the sheet metal part damage box and the component contour of each vehicle component includes: Obtaining the sheet metal part damage position of the sheet metal part damage box from the detection box position; According to the sheet metal part damage position and the component position, counting the number of components of the vehicle components included in the sheet metal part damage box; If the number of components is less than or equal to the preset number, determining the painting result as a semi-spraying method.

6. The vehicle painting method identification method according to claim 5, characterized in that the vehicle painting method identification method further includes: If the number of components is greater than the preset number, or the sheet metal part damage box and the target contour have an intersection, determining the painting result as a full-spraying method.

7. A vehicle painting method identification device, characterized in that the vehicle painting method identification device includes: An acquisition unit, configured to obtain a vehicle image of a damaged vehicle according to the vehicle painting method identification request when receiving the vehicle painting method identification request; A segmentation unit, configured to perform component segmentation processing on the vehicle image based on a pre-trained component segmentation model to obtain a plurality of vehicle components of the damaged vehicle and the component contour of each vehicle component; The segmentation unit is further configured to perform component segmentation processing on the vehicle image based on a pre-trained rib line segmentation model to obtain a plurality of vehicle rib lines of the damaged vehicle and the rib line contour of each vehicle rib line; A positioning unit, configured to locate the target contour of the target rib line among the plurality of vehicle rib lines according to the component contour of each vehicle component and the rib line contour of each vehicle rib line; A detection unit for performing damage detection on the vehicle image based on a pre-trained sheet metal part damage detection model to obtain a sheet metal part damage box of the vehicle image; The detection unit is further configured to detect whether there is an intersection between the sheet metal part damage box and the target contour, including: obtaining a first pixel point in the sheet metal part damage box from the vehicle image; obtaining a second pixel point in the target contour from the vehicle image; detecting whether there is a pixel point in the first pixel points that is at the same pixel position as the second pixel point; if there is a pixel point in the first pixel points that is at the same pixel position as the second pixel point, determining that there is an intersection between the sheet metal part damage box and the target contour; or if there is no pixel point in the first pixel points that is at the same pixel position as the second pixel point, determining that there is no intersection between the sheet metal part damage box and the target contour; A generation unit for generating a painting result of the damaged vehicle according to the sheet metal part damage box and the part contour of each vehicle part if there is no intersection between the sheet metal part damage box and the target contour.

8. An electronic device, characterized in that, the electronic device includes: a memory storing computer-readable instructions; and a processor that executes the computer-readable instructions stored in the memory to implement the vehicle painting method identification method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that: the computer-readable storage medium stores computer-readable instructions, and the computer-readable instructions are executed by a processor in an electronic device to implement the vehicle painting method identification method according to any one of claims 1 to 6.

Images