Vehicle defect identification method, device, equipment and storage medium

Abstract

The present invention relates to a vehicle defect recognition method, device, equipment and storage medium, wherein the vehicle defect recognition method comprises: obtaining a vehicle image of a vehicle to be detected; inputting the vehicle image into a well-trained defect recognition model to perform defect type recognition and defect component recognition, and obtaining corresponding defect type recognition results and defect component recognition results; associating the defect type recognition results and the defect component recognition results based on a preset association rule to obtain a vehicle defect recognition result of the vehicle to be detected, wherein the vehicle defect recognition result includes defect components and the defect type of the defect components. Through this application, the shortcoming of the related art that it is mainly time-consuming to manually identify vehicle defects by inspection personnel is solved, the purpose of quickly identifying defects in used cars is achieved, and the defect recognition efficiency and recognition accuracy are greatly improved.

Description

Technical Field

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

[0002] In the used car market, vehicle defects are a very important factor affecting the value of used cars, and are also one of the key factors affecting whether a used car is sold or not. Therefore, accurately displaying vehicle defects in the actual used car transaction process is the most critical step. Used car defect identification is a complex business scenario. The locations where used cars may have defects cover almost every part of the car body, and the amount of data is large.

[0003] At present, the method of identifying defects in used cars usually uses manual identification to determine the defective parts of the vehicle. Due to the large amount of data, it takes a long time to take photos of vehicle defects and accurately determine the defective parts and the corresponding defect types in this process, which is the most time-consuming part in the current vehicle defect identification process. At present, the method of manually identifying vehicle defects mainly relying on inspection personnel in related technologies is time-consuming and no effective solution has been proposed. Summary of the invention

[0004] The embodiments of the present application provide a vehicle defect identification method, device, equipment and storage medium to at least solve the problem in the related art that vehicle defects of used cars mainly rely on manual identification methods, which is time-consuming.

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

[0006] Acquire a vehicle image of a vehicle to be detected;

[0007] Inputting the vehicle image into a well-trained defect recognition model to perform defect type recognition and defect component recognition, and obtaining corresponding defect type recognition results and defect component recognition results;

[0008] The defect type identification result and the defect component identification result are associated based on a preset association rule to obtain a vehicle defect identification result of the vehicle to be inspected, wherein the vehicle defect identification result includes a defect component and a defect type of the defect component.

[0009] In some embodiments, the defective component identification result includes all components of the vehicle to be inspected, and the all components are arranged in descending order from high to low according to the predicted probability values.

[0010] In some embodiments, the defect type identification result and the defect component identification result are associated based on a preset association rule to obtain the vehicle defect identification result of the vehicle to be inspected, including:

[0011] Selecting a candidate defect type from the defect type identification results;

[0012] The candidate defect types are sequentially associated with the components arranged in descending order until an optimal component that meets a preset association rule is matched, thereby obtaining a vehicle defect identification result of the vehicle to be inspected, wherein the vehicle defect identification result includes the optimal component and the candidate defect type corresponding to the optimal component.

[0013] In some embodiments, the candidate defect types are sequentially associated with the components arranged in descending order until an optimal component that meets a preset association rule is matched, including:

[0014] Creating a temporary defective component identification result and initializing it to the defective component identification result;

[0015] Selecting a temporary defective component with the highest current prediction probability value from the temporary defective component identification results;

[0016] Determining whether the temporary defective component and the candidate defect type meet a preset association rule;

[0017] If it is in compliance, the temporary defective component is used as the optimal component, and the next new candidate defect type is continuously selected from the defect type identification result;

[0018] If not, after deleting the current temporary defective component from the temporary defective component identification result, continue to select the next temporary defective component with the highest prediction probability value until the optimal component that meets the preset association rule is matched.

[0019] In some embodiments, the preset association rule is defect type matching corresponding defective parts, and the preset association rule includes: paint peeling matches car door and / or paint peeling matches bumper.

[0020] In some embodiments, the defect recognition model includes a defect type recognition model and a defect component recognition model;

[0021] The backbone network of the defect type recognition model is built using a single-stage detector method;

[0022] The backbone network of the defective component recognition model is built using a convolutional neural network.

[0023] In some of the embodiments, the method further includes: binding the vehicle defect identification result with the vehicle image, and verifying the actual defect of the vehicle according to the vehicle image to determine the accuracy of the vehicle defect identification result;

[0024] If the vehicle defect recognition result is consistent with the actual defect situation of the vehicle, the vehicle image and the vehicle defect recognition result are directly stored in the inspection service backend;

[0025] If the vehicle defect identification result does not conform to the actual defect situation of the vehicle, the vehicle defect identification result is corrected according to the actual defect situation of the vehicle and stored in the inspection service backend.

[0026] In a second aspect, an embodiment of the present application provides a vehicle defect recognition device, comprising:

[0027] A vehicle image detection module, used to obtain a vehicle image of a vehicle to be detected;

[0028] A vehicle defect recognition module, used for inputting the vehicle image into a well-trained defect recognition model, performing defect type recognition and defect component recognition, and obtaining corresponding defect type recognition results and defect component recognition results;

[0029] The vehicle defect association module is used to associate the defect type identification result and the defect component identification result based on a preset association rule to obtain the vehicle defect identification result of the vehicle to be detected, wherein the vehicle defect identification result includes the defect component and the defect type of the defect component.

[0030] In a third aspect, an embodiment of the present application provides a vehicle defect identification device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the vehicle defect identification method described in the first aspect above.

[0031] In a fourth aspect, an embodiment of the present application provides a storage medium on which a computer program is stored, and when the program is executed by a processor, the vehicle defect identification method as described in the first aspect above is implemented.

[0032] Compared with the related art, the vehicle defect recognition method, device, equipment and storage medium provided in the embodiments of the present application obtain defect type recognition results and defect component recognition results of the vehicle image by inputting the acquired vehicle image of the vehicle to be inspected into a well-trained defect recognition model, and associating the defect type recognition results and the defect component recognition results to obtain the defect component corresponding to the vehicle image and the defect type of the defect component, which solves the shortcoming of the related art that mainly relies on manual identification of vehicle defects by inspectors and is time-consuming, achieves the purpose of quickly identifying defects of used cars, and greatly improves the defect recognition efficiency and recognition accuracy.

[0033] Details of one or more embodiments of the present application are set forth in the following drawings and description to make other features, objects, and advantages of the present application more readily apparent. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0035] Figure 1 It is a hardware structure block diagram of a terminal of the vehicle defect identification method according to an embodiment of the present application;

[0036] Figure 2 is a flow chart of a vehicle defect identification method provided according to an embodiment of the present application;

[0037] Figure 3 is a flow chart of a vehicle defect identification method provided according to another embodiment of the present application;

[0038] Figure 4 is a flow chart of a vehicle defect identification method provided according to another embodiment of the present application;

[0039] Figure 5 It is a structural block diagram of a vehicle defect identification device provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0040] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application is described and illustrated below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application. Based on the embodiments provided in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application. In addition, it can also be understood that although the efforts made in this development process may be complex and lengthy, for ordinary technicians in the field related to the contents disclosed in the present application, some changes such as design, manufacturing or production based on the technical contents disclosed in the present application are only conventional technical means, and should not be understood as insufficient contents disclosed in the present application.

[0041] Reference to "embodiments" in this application means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those of ordinary skill in the art that the embodiments described in this application may be combined with other embodiments without conflict.

[0042] Unless otherwise defined, the technical terms or scientific terms involved in this application should be understood by people with ordinary skills in the technical field to which this application belongs. The words "one", "a", "a", "the" and the like involved in this application do not indicate a quantitative limitation, and may represent the singular or plural. The terms "include", "comprise", "have" and any of their variations involved in this application are intended to cover non-exclusive inclusions; for example, a process, method, system, product or device that includes a series of steps or modules (units) is not limited to the listed steps or units, but may also include steps or units that are not listed, or may also include other steps or units inherent to these processes, methods, products or devices. The words "connect", "connected", "coupled" and the like involved in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The "multiple" involved in this application refers to greater than or equal to two. "And / or" describes the association relationship of associated objects, indicating that there may be three relationships, for example, "A and / or B" can represent: A exists alone, A and B exist at the same time, and B exists alone. The terms "first", "second", "third" and the like involved in the present application are merely used to distinguish similar objects and do not represent a specific ordering of the objects.

[0043] The method embodiment provided in this embodiment can be executed in a terminal, a computer or a similar computing device. Taking running on a terminal as an example, Figure 1 FIG. 1 is a hardware structure block diagram of a terminal of a vehicle defect identification method according to an embodiment of the present application. Figure 1 As shown, the terminal may include one or more ( Figure 1 Only one is shown in the figure) a processor 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data. Optionally, the terminal may also include a transmission device 106 and an input / output device 108 for communication functions. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above terminal. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.

[0044] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the vehicle defect identification method in the embodiment of the present invention. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, to implement the above method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0045] The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (Radio Frequency, referred to as RF) module, which is used to communicate with the Internet wirelessly.

[0046] The present application embodiment provides a vehicle defect identification method, Figure 2 is a flow chart of a vehicle defect identification method according to an embodiment of the present invention. Figure 2 As shown, the process includes the following steps:

[0047] Step S201, obtaining a vehicle image of a vehicle to be detected.

[0048] In the process of used car transactions, used car defect detection is an indispensable step and an extremely complex business scenario. The type and location of the vehicle defect itself is one of the important factors affecting the transaction value of used cars. The locations where vehicle defects usually appear include: structural parts of the vehicle itself (such as: ABC columns, longitudinal beams, shock absorber seats, etc.), reinforcements (such as: front and rear anti-collision beams, inner edges of fenders, etc.), covering parts (such as: engine hood, fenders, doors, trunk lids, etc.), interior parts (such as: workbenches, main and co-driver seats, etc.). Vehicle defect types usually include: "scratches", "scratches", "paint peeling", "potholes", "wear and tear", etc.

[0049] The vehicle defect and the location of the defective component can be reflected in the captured vehicle image, and the relevant defect content can be obtained by analyzing the vehicle image. In practice, the vehicle image of the vehicle to be inspected can be directly captured by the inspector holding a front-end inspection tool (such as a camera) or automatically acquired by an image acquisition system.

[0050] Step S202: input the vehicle image into a well-trained defect recognition model to perform defect type recognition and defect component recognition, and obtain corresponding defect type recognition results and defect component recognition results.

[0051] In an embodiment of the present application, the above-mentioned defect recognition model includes a defect type recognition model and a defect component recognition model. The backbone network of the defect type recognition model can be built using a single-stage detector (Single Shot MultiBoxDetector, referred to as SSD) method; the backbone network of the defect component recognition model can be built using a convolutional neural network (Convolutional Neural Network, referred to as CNN).

[0052] The defect recognition model is based on a multi-task learning architecture and includes two branches: defect recognition and component recognition. Among them, the defect recognition branch is used to detect defect information in vehicle images, and can output defect locations and types. The defect recognition branch can be constructed using the SSD method; the defect component recognition branch is used to identify the target component type in vehicle images. The defect component recognition branch can be constructed using a fully connected neural network layer. The backbone network of the defect recognition model is built using CNN, and the network structure can use ResNet-50 (residual network). The loss function can use a linear combination of the MSE (mean square error) loss function and the cross entropy loss function. Based on the labeled defect recognition training data set, the model training is completed using a supervised learning method.

[0053] The above defect type recognition result includes all defect types identified from the current vehicle image. The defect type recognition result may have no defect type or one or more defect types, depending on the specific defect situation in the vehicle image. The above defect component recognition result shows the defect component identified from the current vehicle image. The defect component may be a certain component or multiple possible components. According to the historical data provided by the second-hand car trading platform, a second-hand car has 400+ vehicle component labels and 2000+ defect labels. The defect type recognition result and the defect component recognition result can find the corresponding type and component in the above labels. For example, the defect type recognition result may have no defect type, or one or more of "scratches", "scratches", "paint peeling", "potholes", "wear and tear", etc.; the defect component recognition result may have one or more of ABC columns, longitudinal beams, shock absorber seats, front and rear anti-collision beams, inner edges of fenders, engine covers, fenders, doors, trunk lids, workbenches, main and co-driver seats, etc.

[0054] Step S203, associating the defect type identification result and the defect component identification result based on a preset association rule to obtain a vehicle defect identification result of the vehicle to be inspected, wherein the vehicle defect identification result includes a defect component and a defect type of the defect component.

[0055] The defect type identification result and the defect component identification result are two related features in the vehicle image. A certain defect type must be reflected on a certain defect component. Therefore, the two need to be associated based on the preset association rules. The main purpose of the preset association rules is to link the defect type identification result with the defect component identification result to show the defects of the vehicle image together. The preset association rules are mainly based on historical data to match the defect type and defect component. If the match is successful, the vehicle defect identification data can be directly output. If the match is unsuccessful, there may be no vehicle defect, or the identification result is incorrect. At this time, manual inspection of the relevant vehicle defects is required.

[0056] Through the above steps, the acquired vehicle image is input into a well-trained defect recognition model to determine the defect type recognition result and defect component recognition result of the vehicle image. In order to ensure the accuracy and completeness of the above recognition results and avoid missing relevant defect content, the defect recognition model outputs all possible recognition results, and associates the defect type recognition result and the defect component recognition result to obtain the defect component corresponding to the vehicle image and the defect type of the defect component, which solves the shortcoming of the related technology that mainly relies on inspection personnel to manually identify vehicle defects and is time-consuming, achieves the purpose of quickly identifying defects in used cars, and greatly improves the defect recognition efficiency and recognition accuracy.

[0057] In some embodiments, the defective component identification result includes all components of the vehicle to be inspected, and all components are arranged in descending order from high to low according to the predicted probability values.

[0058] The number of all parts of the above-mentioned vehicle to be detected is fixed and can be defined manually. For example, according to the historical data provided by a second-hand car trading platform, a second-hand car has 400+ vehicle part labels. The defective part identification result each time is all parts arranged in descending order from high to low according to the predicted probability value. For example, a vehicle image is input into the defect recognition model, and the output defect type recognition result is scratches, and the output defective part recognition results are: engine hood, 90%; fender, 70%; door, 60%; trunk lid, 40%; longitudinal beam 20%, etc. It is possible to judge whether the scratches meet the preset association rules with the engine hood, fender, door, and trunk lid to obtain the vehicle defect recognition result, or it is possible to judge whether they meet the preset association rules in the order of the predicted probability values ​​to obtain the vehicle defect recognition result.

[0059] Figure 3 FIG. 1 is a flow chart of a vehicle defect identification method provided according to another embodiment of the present application. Figure 3 As shown, the process includes the following steps in addition to the above steps S201 and S202:

[0060] Step S301, select a candidate defect type from the defect type identification result. The defect type identification result includes all defect types identified from the current vehicle image. The defect type identification result may contain no defect type, or one or more defect types, depending on the specific defect situation in the vehicle image. The purpose of selecting the candidate defect type is to match it with the defective component identification result to determine whether it meets the preset association rules. If the defect type identification result does not contain a defect type, this step is not executed and the process ends directly.

[0061] Step S302, the candidate defect type is sequentially associated with the components arranged in descending order until the optimal component that meets the preset association rule is matched, and the vehicle defect recognition result of the vehicle to be detected is obtained, and the vehicle defect recognition result includes the optimal component and the candidate defect type corresponding to the optimal component. Among the components arranged in descending order from high to low according to the predicted probability value, the higher the predicted probability value of the component, the greater the possibility that the component is the real component shown in the vehicle image. More importantly, the component with a higher predicted probability value must meet the preset association rule with the candidate defect type, and the component with a higher predicted probability value is the optimal component. For example, the candidate defect type is "wear", and the components arranged in descending order are: engine hood, 90%; fender, 70%; door, 60%; trunk lid, 40%; longitudinal beam 20%...etc. "Wear" matches the engine hood and fender in sequence, and neither meets the preset association rule. "Wear" does not meet the preset association rule until it matches the door, then the door is the optimal component.

[0062] Figure 4 FIG. 1 is a flow chart of a vehicle defect identification method provided according to another embodiment of the present application. Figure 4 As shown, in addition to the above steps S201, S202, and S301, the process also includes the step S302 of sequentially associating the candidate defect types with the components arranged in descending order until the optimal component that meets the preset association rule is matched. The specific steps are:

[0063] S401, creating a temporary defective component identification result and initializing it to the defective component identification result.

[0064] Each defect type identification result needs to be associated with the components arranged in descending order according to the predicted probability value to determine whether it meets the preset association rules. A temporary defective component identification result is created and initialized as the defective component identification result. After each candidate defect type identification result is selected, the candidate defect type identification result can be matched with the complete defective component identification result in turn to avoid missing related components.

[0065] S402: Select a temporary defective component with the highest current prediction probability value from the temporary defective component identification results.

[0066] S403: Determine whether the temporary defective component and the candidate defect type meet a preset association rule.

[0067] S404: If it is in compliance, the temporary defective component is used as the optimal component, and the next new candidate defect type is selected from the defect type identification results.

[0068] S405, if not, after deleting the current temporary defective component from the temporary defective component identification result, continue to select the next temporary defective component with the highest prediction probability value until the optimal component that meets the preset association rule is matched.

[0069] For example, the defect type recognition result includes "scratches", "scratches", "paint peeling", "potholes", and "wear and tear"; the defect component recognition result includes components in descending order, which are: engine hood, 90%; fender, 70%; door, 60%; trunk lid, 40%; longitudinal beam 20%... etc. Create a temporary defect component recognition result and initialize it to: engine hood, 90%; fender, 70%; door, 60%; trunk lid, 40%; longitudinal beam 20%... etc. The selected candidate defect type is "wear and tear". "Wear and tear" is matched with the engine hood and fender in turn, and neither meets the preset association rule. The engine hood and fender are deleted from the temporary defect component recognition result. When "wear and tear" matches with the door, it meets the preset association rule, and the door is the optimal component. Then, reselect a candidate defect type "scratches" and match it with the engine hood, fender, door, etc. in the initialized temporary defect component recognition result in turn, and repeat the above operation.

[0070] In another embodiment of the present application, the preset association rule is defect type matching corresponding defective parts, which may include: paint peeling matching car doors and / or paint peeling matching bumpers.

[0071] In another embodiment of the present application, the application further includes: binding the vehicle defect identification result with the vehicle image, and verifying the actual defect of the vehicle based on the vehicle image to determine the accuracy of the vehicle defect identification result;

[0072] If the vehicle defect recognition result is consistent with the actual defect situation of the vehicle, the vehicle image and the vehicle defect recognition result are directly stored in the inspection service backend;

[0073] If the vehicle defect identification result does not conform to the actual defect situation of the vehicle, the vehicle defect identification result is corrected according to the actual defect situation of the vehicle and stored in the inspection service backend.

[0074] Since vehicle defects are an important reference standard for determining the price of used cars, and vehicle photo recognition is also the most numerous and time-consuming step in used car defect recognition, currently there are an average of 40 or more photos of defect types in a used car, and vehicle defect recognition takes a total of 25-30 minutes. By using a defect recognition model on vehicle photos, automatic identification of vehicle defects (name of vehicle defective parts and defect type information of the defective parts) can be completed, and the time consumed for vehicle defect recognition can be controlled to less than 20-25 minutes, which can reduce the time consumption by about 5 minutes and improve the work efficiency of inspection personnel by more than 20%. In addition, using this vehicle defect recognition method, the accuracy of identifying defect types can be increased to more than 90%, the accuracy of identifying defective parts can be increased to more than 95%, and the accuracy of identifying both defect types and defective parts can be increased to more than 80%.

[0075] Another embodiment of the present application is described below in combination with a specific application scenario in practice. The process of the vehicle defect recognition scenario is as follows:

[0076] S501, the inspection personnel use the inspection front-end tool (such as a camera) to collect vehicle images;

[0077] S502, the detection front-end tool calls the detection service back-end interface to upload the collected vehicle image to the detection service back-end;

[0078] S503, the inspection service backend calls the defect recognition model to complete vehicle defect recognition, identifies defective parts in the vehicle image and the defect types of the defective parts, and transmits the recognition result information to the inspection front-end tool through the network. The inspection front-end tool receives the recognition result information and stores it locally in the inspection front-end tool;

[0079] S504, the inspector clicks on the vehicle image to enter the photo preview page and verifies the defect information according to the actual situation of the vehicle: when the defect information is consistent with the actual situation of the vehicle, the inspector clicks on the list page to complete the inspection, and directly stores the vehicle image and the vehicle defect identification result to the inspection service backend; when the defect information does not conform to the actual situation of the vehicle, the inspector can change the name of the defective part of the vehicle and the defect type corresponding to the defective part on the photo preview page, click on the list page to complete the inspection, and store the changed name of the defective part of the vehicle and the defect type corresponding to the defective part to the inspection service backend.

[0080] The specific steps of the vehicle defect identification method performed in the inspection service background are as follows:

[0081] S601: Input a vehicle image to be detected.

[0082] S602, using a fully trained defect recognition model to identify defect types and defective parts of the vehicle image to be inspected, the defect type recognition result is recorded as R_defect, and the defective part recognition result is recorded as R_parts. The defective part recognition result includes all parts of the vehicle to be inspected, and all parts are arranged in descending order from high to low according to the predicted probability value.

[0083] S603, performing association processing on the recognition results of the defect recognition model, and outputting the final vehicle defect recognition result, which is recorded as R_final.

[0084] The specific association processing steps are as follows:

[0085] S603-1, select a defect type recognition result from the defect type recognition result R_defect as a candidate defect type recognition result, recorded as RD_candidate. If R_defect is empty, directly assign the defect type recognition result R_final to empty, and the process ends.

[0086] S603-2, create a temporary defective component identification result R_parts_temp and initialize it to the defective component identification result R_parts.

[0087] S603-3, select the part with the highest probability value from R_parts_temp as the optimal part, recorded as RP_candidate.

[0088] S603-4, according to the preset association rules, determine whether RD_candidate and RP_candidate meet the preset association rules. If so, update RD_candidate and RP_candidate as vehicle defect identification results to R_final, and delete RD_candidate from R_defect; if not, delete RP_candidate from R_parts_temp, and continue to execute steps S3.3 to S3.4.

[0089] S603-5, delete RD_candidate from R_defect, determine whether R_defect is empty, if not, continue to execute steps S3.1 to S3.4; if it is empty, output the final defect identification result R_final, and the process ends.

[0090] It should be noted that the steps shown in the above process or the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0091] This embodiment also provides a vehicle defect identification device, which is used to implement the above-mentioned embodiments and preferred implementation modes, and will not be repeated hereafter. As used below, the terms "module", "unit", "subunit", etc. may be a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0092] Figure 5 is a structural block diagram of a vehicle defect identification device provided according to an embodiment of the present application, such as Figure 5 As shown, the device includes: a vehicle image detection module 41, which is used to obtain a vehicle image of a vehicle to be detected;

[0093] A vehicle defect recognition module 42, configured to input the vehicle image into a well-trained defect recognition model, perform defect type recognition and defect component recognition, and obtain corresponding defect type recognition results and defect component recognition results; and

[0094] The vehicle defect association module 43 is used to associate the defect type identification result and the defect component identification result based on a preset association rule to obtain the vehicle defect identification result of the vehicle to be detected, and the vehicle defect identification result includes the defect component and the defect type of the defect component.

[0095] In an embodiment of the present application, the defective component identification result includes all components of the vehicle to be inspected, and the all components are arranged in descending order from high to low according to the predicted probability values.

[0096] In an embodiment of the present application, the vehicle defect association module 43 is used to select a candidate defect type from the defect type identification result; associate the candidate defect type with the components arranged in descending order in sequence until the optimal component that meets the preset association rule is matched, and obtain the vehicle defect identification result of the vehicle to be inspected, and the vehicle defect identification result includes the optimal component and the candidate defect type corresponding to the optimal component.

[0097] In an embodiment of the present application, the vehicle defect association module 43 is also used to create a temporary defective component identification result and initialize it to the defective component identification result; select the temporary defective component with the highest current predicted probability value from the temporary defective component identification result; determine whether the temporary defective component and the candidate defect type meet the preset association rules; if yes, use the temporary defective component as the optimal component and continue to select the next new candidate defect type from the defect type identification result; if not, delete the current temporary defective component from the temporary defective component identification result, and continue to select the next temporary defective component with the highest predicted probability value until the optimal component that meets the preset association rules is matched.

[0098] In an embodiment of the present application, the preset association rule is defect type matching corresponding defective parts, including: paint peeling matching car doors and / or paint peeling matching bumpers.

[0099] In an embodiment of the present application, the defect recognition model includes a defect type recognition model and a defect component recognition model; the backbone network of the defect type recognition model is built using a single-stage detector method; the backbone network of the defect component recognition model is built using a convolutional neural network.

[0100] In an embodiment of the present application, the device further includes an accuracy determination module 44, which is used to bind the vehicle defect identification result with the vehicle image, and verify the actual defect of the vehicle according to the vehicle image to determine the accuracy of the vehicle defect identification result;

[0101] If the vehicle defect recognition result is consistent with the actual defect situation of the vehicle, the vehicle image and the vehicle defect recognition result are directly stored in the inspection service backend;

[0102] If the vehicle defect identification result does not conform to the actual defect situation of the vehicle, the vehicle defect identification result is corrected according to the actual defect situation of the vehicle and stored in the inspection service backend.

[0103] It should be noted that the above modules can be functional modules or program modules, and can be implemented by software or hardware. For modules implemented by hardware, the above modules can be located in the same processor; or the above modules can be located in different processors in any combination.

[0104] This embodiment further provides an electronic device, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0105] Optionally, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0106] Optionally, in this embodiment, the processor may be configured to perform the following steps through a computer program:

[0107] S1, obtaining a vehicle image of a vehicle to be detected.

[0108] S2, inputting the vehicle image into a well-trained defect recognition model to perform defect type recognition and defect component recognition, and obtaining corresponding defect type recognition results and defect component recognition results.

[0109] S3, associating the defect type identification result and the defect component identification result based on a preset association rule to obtain a vehicle defect identification result of the vehicle to be inspected, wherein the vehicle defect identification result includes a defect component and a defect type of the defect component.

[0110] It should be noted that the specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementation modes, and this embodiment will not be described in detail here.

[0111] In addition, in combination with the vehicle defect identification method in the above embodiment, the embodiment of the present application can provide a storage medium for implementation. The storage medium stores a computer program; when the computer program is executed by a processor, any one of the vehicle defect identification methods in the above embodiment is implemented.

[0112] Those skilled in the art should understand that the various technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the various technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0113] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A vehicle defect identification method, characterized in that: include: Acquire a vehicle image of a vehicle to be detected; Inputting the vehicle image into a well-trained defect recognition model to perform defect type recognition and defect component recognition, and obtaining corresponding defect type recognition results and defect component recognition results; Associating the defect type identification result and the defect component identification result based on a preset association rule to obtain a vehicle defect identification result of the vehicle to be inspected, wherein the vehicle defect identification result includes a defect component and a defect type of the defect component; The defective component identification result includes all components of the vehicle to be inspected, and all components are arranged in descending order from high to low according to the predicted probability values; The step of associating the defect type identification result and the defect component identification result based on a preset association rule to obtain a vehicle defect identification result of the vehicle to be inspected includes: Selecting a candidate defect type from the defect type identification results; The candidate defect types are sequentially associated with the components arranged in descending order until an optimal component that meets a preset association rule is matched, thereby obtaining a vehicle defect identification result of the vehicle to be inspected, wherein the vehicle defect identification result includes the optimal component and the candidate defect type corresponding to the optimal component.

2. The vehicle defect identification method according to claim 1, characterized in that: The candidate defect types are sequentially associated with the components arranged in descending order until an optimal component that meets a preset association rule is matched, including: Creating a temporary defective component identification result and initializing it to the defective component identification result; Selecting a temporary defective component with the highest current prediction probability value from the temporary defective component identification results; Determining whether the temporary defective component and the candidate defect type meet a preset association rule; If it is in compliance, the temporary defective component is used as the optimal component, and the next new candidate defect type is continuously selected from the defect type identification result; If not, after deleting the current temporary defective component from the temporary defective component identification result, continue to select the next temporary defective component with the highest prediction probability value until the optimal component that meets the preset association rule is matched.

3. The vehicle defect identification method according to any one of claims 1 to 2, characterized in that: The preset association rule is that defect types match corresponding defective parts, and the preset association rule includes: paint peeling matches car doors and / or paint peeling matches bumpers.

4. The vehicle defect identification method according to any one of claims 1 to 2, characterized in that: The defect recognition model includes a defect type recognition model and a defect component recognition model; The backbone network of the defect type recognition model is built using a single-stage detector method; The backbone network of the defective component recognition model is built using a convolutional neural network.

5. The vehicle defect identification method according to any one of claims 1 to 2, characterized in that: Also includes: Binding the vehicle defect identification result with the vehicle image, and verifying the actual defect of the vehicle according to the vehicle image to determine the accuracy of the vehicle defect identification result; If the vehicle defect recognition result is consistent with the actual defect situation of the vehicle, the vehicle image and the vehicle defect recognition result are directly stored in the inspection service backend; If the vehicle defect identification result does not conform to the actual defect situation of the vehicle, the vehicle defect identification result is corrected according to the actual defect situation of the vehicle and stored in the inspection service backend.

6. A vehicle defect identification device, characterized in that: include: A vehicle image detection module, used to obtain a vehicle image of a vehicle to be detected; A vehicle defect recognition module, used for inputting the vehicle image into a well-trained defect recognition model, performing defect type recognition and defect component recognition, and obtaining corresponding defect type recognition results and defect component recognition results; A vehicle defect association module, used for associating the defect type identification result and the defect component identification result based on a preset association rule to obtain a vehicle defect identification result of the vehicle to be detected, wherein the vehicle defect identification result includes a defect component and a defect type of the defect component; the defect component identification result includes all components of the vehicle to be detected, and all components are arranged in descending order from high to low according to the predicted probability value; The vehicle defect association module is specifically used to: select a candidate defect type from the defect type identification result; associate the candidate defect type with the components arranged in descending order in sequence until an optimal component that meets a preset association rule is matched, and obtain a vehicle defect identification result of the vehicle to be inspected, wherein the vehicle defect identification result includes the optimal component and the candidate defect type corresponding to the optimal component.

7. A vehicle defect recognition device, comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to run the computer program to execute the vehicle defect identification method according to any one of claims 1 to 5.

8. A storage medium, characterized in that: The storage medium stores a computer program, wherein the computer program is configured to execute the vehicle defect identification method according to any one of claims 1 to 5 when running.

Images