Chassis defect detection method, device, medium and equipment
By using defect position prior information for pixel interception and neural network model detection in chassis defect detection, the problem of low detection accuracy in the prior art is solved, and higher detection accuracy and fewer false detection are achieved.
Patent Information
- Application Number
- CN202410107425.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-01-25
- Publication Date
- 2025-07-25
AI Technical Summary
The existing chassis defect detection methods rely on vehicle positioning accuracy, resulting in low detection accuracy and prone to false detection.
The chassis image is pixel-intercepted using defect position prior information, the area image of the target component is obtained, and defect detection is performed through the neural network model to eliminate false detection of non-target components.
It improves the accuracy of chassis defect detection, reduces the requirements for vehicle positioning accuracy, and reduces the false detection of non-target components.
Smart Images
Figure CN120374487A_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the field of automotive technologies, and in particular, to a method, device, medium, and equipment for detecting chassis defects. Background Art
[0002] In recent years, with the continuous development of automotive technologies and the improvement of economic levels, automobiles, as major consumer goods, have entered thousands of households.
[0003] After all assemblies of a vehicle are completed in the general assembly workshop, on-road tests, rain tests, and other on-site tests before leaving the factory will be carried out in the factory, and then the integrity of the chassis will be detected on the final inspection line of the inspection line in the general assembly workshop, so as to timely detect various quality defects and abnormalities such as scratches, glue overflow, bottom leakage, and liquid leakage that may exist on the chassis. Since the areas where defects may occur cover the entire chassis, it is difficult to detect all defects within a limited time.
[0004] Currently, there is an automatic chassis defect detection method based on vision. The vehicle to be tested is parked at the detection position, a chassis image is obtained by taking a photo, and the chassis image is analyzed and compared with a standard image to determine the defects existing on the chassis. However, the existing detection method is highly dependent on the accuracy of vehicle positioning. Slightly larger vehicle positioning deviation will easily lead to misdetection, resulting in the technical problem of low accuracy in chassis defect detection. Summary of the Invention
[0005] In order to solve the above technical problems, the present disclosure provides a method, device, medium, and equipment for detecting chassis defects to improve the accuracy of chassis defect detection.
[0006] The present disclosure provides a method for detecting chassis defects, including:
[0007] Performing pixel interception on the chassis image according to the prior information of the defect position to obtain at least one regional image representing the target component;
[0008] Performing defect detection on at least one regional image to identify whether there are defects on the target component.
[0009] In some embodiments, the prior information of the defect position includes pre-determined pixel position information;
[0010] Before performing pixel extraction on the chassis image according to the prior information of the defect position, it further includes:
[0011] Determining the prior information of the defect position according to the photographing position of the camera relative to the chassis, and the pixel position information corresponding to different photographing positions is different.
[0012] In some embodiments, the prior information of the defect position includes pre-determined target visual feature information of the target component;
[0013] Before pixel cropping the chassis image according to the prior information of the defect position, it further includes:
[0014] Identify the chassis image to determine the recognition visual feature information and envelope position information of the components included in the chassis image;
[0015] The pixel cropping of the chassis image according to the prior information of the defect position includes: screening the recognition visual feature information by using the target visual feature information to determine the recognition visual feature information representing the target component;
[0016] Perform pixel cropping on the chassis image based on the envelope position information associated with the recognition visual feature information representing the target component.
[0017] In some embodiments, the defect detection of at least one region image to identify whether there is a defect on the target component includes:
[0018] Perform defect recognition on at least one of the region images to obtain recognition defect information, where the recognition defect information includes the category of the recognized defect;
[0019] Filter the category of the recognized defect by using the prior information of the defect category to obtain the retained defect, where the prior information of the defect category is determined according to the type of the target component;
[0020] In the case where there is the retained defect, use the retained defect as the defect on the target component.
[0021] In some embodiments, the defect detection of at least one region image to identify whether there is a defect on the target component includes:
[0022] Perform defect recognition on at least one of the region images to obtain recognition defect information, where the recognition defect information includes the size of the recognized defect;
[0023] Filter the size of the recognized defect by using the prior information of the defect size to obtain the retained defect;
[0024] In the case where there is the retained defect, use the retained defect as the defect on the target component.
[0025] In some embodiments, the defect recognition of at least one of the region images to obtain recognition defect information includes:
[0026] Perform defect recognition on at least one of the region images to obtain a defect bounding box enclosing the recognized defect;
[0027] Based on the pixel size of the defect bounding box and the pre-determined pixel-physical size conversion relationship, determine the size of the target box, and use the size of the target box as the size of the identified defect.
[0028] The present disclosure also provides a chassis defect detection device, including:
[0029] An interception module, configured to perform pixel interception on the chassis image according to the prior information of the defect position, so as to obtain at least one region image representing the target component;
[0030] A detection module, configured to perform defect detection on at least one region image to identify whether there is a defect on the target component.
[0031] The present disclosure also provides a computer-readable storage medium, where the computer-readable storage medium stores a program or instruction, and the program or instruction enables a computer to execute the steps of any one of the above methods.
[0032] The present disclosure also provides an electronic device, including:
[0033] One or more processors;
[0034] A memory, configured to store one or more programs or instructions;
[0035] The processor is configured to execute the steps of any one of the above methods by calling the program or instruction stored in the memory.
[0036] The technical solution provided by the embodiment of the present disclosure has the following advantages compared with the prior art:
[0037] For the technical solution provided by the embodiment of the present disclosure, first, pixel interception is performed on the chassis image according to the prior information of the defect position to obtain at least one region image representing the target component, and then defect detection is performed on at least one region image to identify whether there is a defect on the target component. Using the prior information of the defect position for pixel interception can locate at least one region image of the target component. Therefore, the requirement for vehicle positioning accuracy is not high, and non-target components, that is, components that do not need to be detected, are excluded. Then, only defect detection is performed on the target component, thereby excluding false detection of defects that may appear on non-target components, and solving the technical problem of low accuracy existing in the prior art. Description of the Drawings
[0038] The drawings here are incorporated into the specification and constitute a part of this specification, showing embodiments consistent with the present disclosure, and are used together with the specification to explain the principles of the present disclosure.
[0039] To more clearly illustrate the technical solutions in the embodiments of the present disclosure or the prior art, the following will briefly introduce the accompanying drawings required for the description of the embodiments or the prior art. Obviously, for those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.
[0040] Figure 1 Flowchart of a chassis defect detection method provided by an embodiment of the present disclosure;
[0041] Figure 2 Another flowchart of a chassis defect detection method provided by an embodiment of the present disclosure;
[0042] Figure 3 Schematic diagram of an application example of the chassis defect detection method in an embodiment of the present disclosure;
[0043] Figure 4 Another schematic diagram of an application example of the chassis defect detection method in an embodiment of the present disclosure;
[0044] Figure 5 Another schematic diagram of an application example of the chassis defect detection method in an embodiment of the present disclosure;
[0045] Figure 6 Another schematic diagram of an application example of the chassis defect detection method in an embodiment of the present disclosure;
[0046] Figure 7 Block diagram of the structure of a chassis defect detection device provided by an embodiment of the present disclosure;
[0047] Figure 8 Schematic diagram of the structure of an electronic device provided by an embodiment of the present disclosure. Detailed implementation manners
[0048] In order to more clearly understand the above objects, features, and advantages of the present disclosure, the following will further describe the solutions of the present disclosure. It should be noted that, without conflict, the embodiments of the present disclosure and the features in the embodiments can be combined with each other.
[0049] Many specific details are set forth in the following description in order to fully understand the present disclosure, but the present disclosure can also be implemented in other ways different from those described herein; obviously, the embodiments in the specification are only a part of the embodiments of the present disclosure, rather than all of the embodiments.
[0050] Figure 1 Flowchart of a chassis defect detection method provided by an embodiment of the present disclosure. This method is applicable to the defect detection of vehicle chassis, and this method can be executed by a chassis defect detection device, which can be implemented in software and / or hardware. As Figure 1As shown, the chassis defect detection method includes the following steps:
[0051] S110. Pixel-intercept the chassis image according to the prior information of the defect position to obtain at least one regional image representing the target component.
[0052] A 2D image acquisition device such as a camera can be used to acquire images of the vehicle chassis. For different vehicle models, there may be 20 to 40 acquisition points. The prior information of the defect position includes the category of the target component to be detected at each acquisition point. Each acquisition point corresponds to different target components. Some acquisition points correspond to one target component, and some acquisition points correspond to multiple target components. That is to say, each acquisition point has its own different prior information of the defect position. The prior information of the defect position includes the category of the components to be detected in the chassis. The prior information of the defect position can be established and saved in advance for use when the chassis defect detection method runs. Pixel-interception is performed on the images acquired at each acquisition point using this step to obtain at least one regional image representing the target component at each acquisition point.
[0053] S120. Perform defect detection on at least one regional image to identify whether there are defects on the target component.
[0054] Defect detection can use a neural network model to detect the target components in the image. The detection results can include the positions of all detected defects and the category of each detected defect. Typical neural network models include Faster RCNN, Yolo series, and SSD, etc.
[0055] In the embodiments of the present disclosure, pixel-interception is performed using the prior information of the defect position, which can locate at least one regional image of the target component. Therefore, the requirement for vehicle positioning accuracy is not high, and non-target components, that is, components that do not need to be detected, are excluded. Then, only defect detection is performed on the target components, thereby excluding false detections of defects that may appear on non-target components, and solving the technical problem of low accuracy existing in the prior art.
[0056] In some embodiments, the prior information of the defect position includes pre-determined pixel position information. Before step S110, it further includes:
[0057] Determine the prior information of the defect position according to the photographing position of the camera relative to the chassis.
[0058] Among them, the pixel position information corresponding to the same photographing position is different. The vehicle and the camera can be pre-positioned in advance. A relatively strict pre-positioning of the vehicle's position is required, and the positions of each acquisition point are also fixed in advance, so that the relative positions of the vehicle and each acquisition point are fixed. Then the range of the chassis image collected at each acquisition point is also fixed. The prior information of the defect position is the pixels at some specific positions in the chassis image. In the chassis image collected at one acquisition point, the target component must appear at a fixed position in the chassis image. Then the prior information of the defect position corresponding to this acquisition point is the pixel position information of the target component, such as the pixels from the nth row to the mth row and the pixels from the jth column to the kth column. The pixel position information of the target component is within this range.
[0059] In some embodiments, the prior information of the defect position includes the target visual feature information of the predetermined target component.
[0060] Before step S110, it further includes:
[0061] Identify the chassis image to determine the recognition visual feature information and the envelope position information of the components included in the chassis image.
[0062] The recognition of the chassis image can be based on the visual processing of deep learning, an artificial intelligence solution trained based on a large amount of two-dimensional image data, used to complete tasks such as classifying, detecting, and entity segmentation of targets in the image. The recognition of the recognition visual feature information belongs to the classification task. A convolutional neural network (Convolutional Neural Network, abbreviated as CNN) or other methods are used to classify the overall content of the image. The classification result of each image includes the category of the image. Typical neural network models include ResNet18, etc.
[0063] Step S110 includes:
[0064] Use the target visual feature information to screen the recognition visual feature information to determine the recognition visual feature information representing the target component; perform pixel interception on the chassis image based on the envelope position information associated with the recognition visual feature information representing the target component.
[0065] Use the recognition visual feature information to screen out the target components that need to be detected, determine the recognition visual feature information representing the target component, then combine the envelope position information to determine the pixel position corresponding to the target component, and then perform pixel interception on the chassis image based on the envelope position information associated with the recognition visual feature information.
[0066] In some embodiments, pixel interception can be performed by entity segmentation. For example, the target component is retained as the foreground in the image, and other areas are treated as irrelevant background parts and blackened. A CNN network or other methods can be used to perform entity segmentation on the target component in the image. The segmentation result includes the positions of all target components in the image. Typical neural network models include Mask RCNN, etc.
[0067] In some embodiments, step S120 includes:
[0068] Perform defect recognition on at least one regional image to obtain recognition defect information, where the recognition defect information includes the category of the recognized defect.
[0069] Use the prior information of defect categories to filter the categories of the recognized defects to obtain retained defects.
[0070] Among them, the prior information of defect categories is determined according to the type of the target component. According to the category of each recognized defect, compare it with the prior information of defect categories. If the category of a certain defect does not belong to the defects that may occur in the current target component, that is, the current target component cannot have such defects, then this defect belongs to the defect of the invalid category and will be regarded as a false detection and excluded. Finally, only the defects of the categories that may occur in the current target component are retained as the retained defects, so that the false detection rate can be significantly reduced.
[0071] In the case of existing retained defects, the retained defects are used as the defects on the target component.
[0072] If all defects are excluded and there are finally no defects belonging to the valid category, the defect detection at the current acquisition point is ended, and the chassis defect detection method is restarted at the next acquisition point.
[0073] In some embodiments, step S120 further includes:
[0074] Perform defect recognition on at least one regional image to obtain recognition defect information, where the recognition defect information includes the size of the recognized defect.
[0075] In some embodiments, first perform defect recognition on at least one regional image to obtain a defect bounding box that encloses the recognized defect; then, based on the pixel size of the defect bounding box and the pre-determined pixel-physical size conversion relationship, determine the size of the target box, and use the size of the target box as the size of the recognized defect. That is, calculate the scaling ratio according to the image size and actual size of the target component in the image, and then calculate the size of the target box of each defect according to the image size occupied by each defect bounding box in the image and the above scaling ratio, which is the actual size of the defect.
[0076] The size of the identified defects is filtered using the prior information on defect size to obtain the retained defects.
[0077] The prior information on defect size includes the negligible size thresholds for defects of each category. The prior information on defect size can also be established and saved in advance for use when the chassis defect detection method runs.
[0078] If the size of a certain defect is small, less than or equal to the negligible size threshold, then this defect can be ignored and regarded as non-existent, or it can also be considered equivalent to regarding this defect as a false detection, thereby further eliminating some false detections of defects to solve the technical problem of low accuracy existing in the prior art.
[0079] In the case of having retained defects, the retained defects are taken as the defects on the target component.
[0080] If all defects are ignored due to their small sizes, the defect detection at the current acquisition point is ended, and the chassis defect detection method is restarted at the next acquisition point.
[0081] The following describes the chassis defect detection method provided by the embodiments of the present disclosure in combination with application examples:
[0082] After establishing and saving the prior information on defect location, prior information on defect category, and prior information on defect size, the robotic arm drives the 2D image acquisition device to perform traversal image acquisition on the vehicle chassis. Since the acquisition points of the robotic arm are fixed for each vehicle, the relative positions where the vehicle stops each time cannot vary too much, but it is not necessary to be overly precise. Generally, the repeat positioning error does not exceed 50 mm. Vehicle positioning can be achieved by adding a vehicle stop limiter at the detection station, and the defect detection process is as Figure 2 shown.
[0083] The currently acquired chassis image is as Figure 3 shown, and target detection is performed on the currently acquired chassis image.
[0084] Based on the prior information on defect location, it can be known that only the rear subframe assembly in this chassis image is the target component to be detected. Therefore, the area of the rear subframe assembly in this image is segmented based on the deep learning model for the location of chassis components. As Figure 4 shown, the rear subframe assembly is retained as the foreground in the figure, and other irrelevant parts are used as the background and blackened. If the target component to be detected is not detected, the robotic arm drives the 2D image acquisition device to run to the next acquisition point and returns to the initial step.
[0085] Using a defect detector, the segmented rear subframe assembly area is subjected to defect detection. Among them, the defect detector includes, but is not limited to, detectors or detection networks trained by methods such as deep learning and machine learning. If there are defects in the foreground area, the target area (Region Of Interest, abbreviated as ROI) of the 2D positions of all detected defects in the figure is given. The ROI is a defect bounding box that demarcates the defect with a rectangular box, as Figure 5 shown by the square box. If no defects are detected in the foreground area, the robotic arm drives the 2D image acquisition device to run to the next acquisition point, and returns to the initial step.
[0086] After scaling the image areas corresponding to all defect ROIs to a unified size, they are sent to a defect classifier for recognition to obtain the categories of all defects. Among them, the defect classifier includes, but is not limited to, classifiers or classification networks trained by methods such as deep learning and machine learning. Then, based on the prior information of the defect categories, filter out the defect categories that are recognized by the classifier as existing but cannot actually appear in the rear subframe assembly area, and retain the possible defect categories. If all defects belong to the defect categories that cannot appear, the robotic arm drives the 2D image acquisition device to run to the next acquisition point, and returns to the initial step.
[0087] Calculate the size of each defect, as Figure 6 shown, find the width u and height v of the minimum enclosing surface of the segmented rear subframe assembly area. By comparing with the actual width w and height h of the known rear subframe assembly, the ratio u / w or v / h of the area image of this area to the actual size can be obtained, and then the actual size of the above defect ROI can be calculated. Then, based on the prior information of the defect size, filter out the defects with small sizes that can be ignored. If all defects are ignored due to small sizes, the robotic arm drives the 2D image acquisition device to run to the next acquisition point, and returns to the initial step.
[0088] The finally retained defects are the actually detected defects in the rear subframe assembly area, and record the positions and types of the current defects.
[0089] It should be noted that the above embodiments only use the rear subframe assembly as an example of defect detection. In other examples, it can also be components of the vehicle chassis such as the left front shock absorber lower bracket assembly and the left front lower control arm assembly, which are not limited here.
[0090] Corresponding to the chassis defect detection method provided by the embodiments of the present disclosure, the embodiments of the present disclosure also provide a chassis defect detection device. Figure 7 For the structural block diagram of the chassis defect detection device provided by the embodiments of the present disclosure, as Figure 7 shown, the device includes:
[0091] The extraction module 71 is configured to perform pixel extraction on the chassis image according to the prior information of the defect position, so as to obtain at least one region image representing the target component;
[0092] The detection module 72 is configured to perform defect detection on at least one region image to identify whether there are defects on the target component.
[0093] In some embodiments, the prior information of the defect position includes pre-determined pixel position information;
[0094] The extraction module 71 is further configured to: before performing pixel extraction on the chassis image according to the prior information of the defect position, determine the prior information of the defect position according to the photographing position of the camera relative to the chassis, and the pixel position information corresponding to different photographing positions is different.
[0095] In some embodiments, the prior information of the defect position includes pre-determined target visual feature information of the target component;
[0096] The extraction module 71 is further configured to: before performing pixel extraction on the chassis image according to the prior information of the defect position, identify the chassis image to determine the identified visual feature information and envelope position information of the components included in the chassis image;
[0097] Specifically, the extraction module 71 is further configured to: screen the identified visual feature information by using the target visual feature information to determine the identified visual feature information representing the target component; perform pixel extraction on the chassis image based on the envelope position information associated with the identified visual feature information representing the target component.
[0098] In some embodiments, the detection module 72 is specifically configured to:
[0099] Perform defect identification on at least one region image to obtain identified defect information, where the identified defect information includes the category of the identified defect; filter the category of the identified defect by using the prior information of the defect category to obtain the retained defect, and the prior information of the defect category is determined according to the type of the target component; in the case of the existence of the retained defect, use the retained defect as the defect on the target component.
[0100] In some embodiments, the detection module 72 is specifically configured to:
[0101] Perform defect identification on at least one region image to obtain identified defect information, where the identified defect information includes the size of the identified defect; filter the size of the identified defect by using the prior information of the defect size to obtain the retained defect; in the case of the existence of the retained defect, use the retained defect as the defect on the target component.
[0102] In some embodiments, the detection module 72 is further specifically configured to:
[0103] Defect recognition is performed on at least one regional image to obtain a defect bounding box enclosing the recognized defect; based on the pixel size of the defect bounding box and a pre-determined pixel-physical size conversion relationship, the size of the target box is determined, and the size of the target box is used as the size of the recognized defect.
[0104] The chassis defect detection device disclosed in the above embodiments can execute the chassis defect detection methods disclosed in the above respective embodiments, and has the same or corresponding beneficial effects. To avoid repetition, it will not be elaborated here.
[0105] The embodiments of the present disclosure also provide a computer-readable storage medium. The computer-readable storage medium stores a program or instructions, and the program or instructions cause a computer to execute the steps of any of the above methods.
[0106] Based on the prior information of the defect location, target detection is performed on the current chassis image to obtain the components to be detected in the current chassis image;
[0107] Defect detection is performed on the component to be detected to detect the defects of the component to be detected;
[0108] The defect category of each defect is identified, and based on the prior information of the defect category, the defects belonging to the invalid category are excluded, and the defects belonging to the valid category are retained.
[0109] Optionally, when executed by a computer processor, the computer-executable instructions can also be used to execute the technical solutions of any of the above chassis defect detection methods provided by the embodiments of the present disclosure to achieve the corresponding beneficial effects.
[0110] From the above description of the embodiments, those skilled in the art can clearly understand that the embodiments of the present disclosure can be implemented by means of software and necessary general-purpose hardware. Of course, it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the embodiments of the present disclosure, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as a floppy disk, a read-only memory (ROM), a random access memory (RAM), a flash memory (FLASH), a hard disk, or an optical disc of a computer, etc., including several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in the various embodiments of the present disclosure.
[0111] Embodiments of the present disclosure also provide an electronic device, including: one or more processors; a memory for storing one or more programs or instructions; the processor is configured to execute the steps of any of the above methods by invoking the programs or instructions stored in the memory, so as to achieve corresponding beneficial effects.
[0112] Figure 8 It is a schematic diagram of the hardware structure of the electronic device provided by the embodiments of the present disclosure. As Figure 8 shown, the electronic device includes one or more processors 801 and a memory 802.
[0113] The processor 801 may be a central processing unit (CPU) or other forms of processing units with data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device to perform desired functions.
[0114] The memory 802 may include one or more computer program products, and the computer program products may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory, etc. The non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage media, and the processor 801 may run the program instructions to implement the chassis defect detection method of the embodiments of the present disclosure described above, and / or other desired functions. Various contents such as input signals, signal components, and noise components may also be stored in the computer-readable storage media.
[0115] In one example, the electronic device may further include: an input device 803 and an output device 804, and these components are interconnected through a bus system and / or other forms of connection mechanisms (not shown).
[0116] In addition, the input device 803 may further include, for example, a keyboard, a mouse, and so on.
[0117] The output device 804 may output various information to the outside, including the determined distance information, direction information, etc. The output device 804 may include, for example, a display, a speaker, a printer, and a communication network and its connected remote output devices, and so on.
[0118] Of course, for simplicity, Figure 8 only some of the components related to the present disclosure in the electronic device are shown, and components such as buses, input / output interfaces, etc. are omitted. In addition, according to specific application scenarios, the electronic device may further include any other appropriate components.
[0119] It should be noted that in this document, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, such that a process, method, article or device comprising a series of elements not only includes those elements but also other elements not expressly listed, or elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising one..." does not exclude the presence of additional identical elements in the process, method, article or device comprising said element.
[0120] The above are only specific embodiments of the present disclosure, enabling those skilled in the art to understand or implement the present disclosure. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present disclosure. Therefore, the present disclosure will not be limited to these embodiments described herein, but rather will conform to the broadest scope consistent with the principles and novel features disclosed herein.
Claims
1. A method for detecting chassis defects, characterized in that, Including: Performing pixel interception on the chassis image according to the prior information of the defect position to obtain at least one regional image representing the target component; Performing defect detection on at least one regional image to identify whether there are defects on the target component.
2. The method according to claim 1, characterized in that, The prior information of the defect position includes pre-determined pixel position information; Before performing pixel extraction on the chassis image according to the prior information of the defect position, it further includes: Determining the prior information of the defect position according to the photographing position of the camera relative to the chassis, and the pixel position information corresponding to different photographing positions is different.
3. The method according to claim 1, characterized in that, The prior information of the defect position includes pre-determined target visual feature information of the target component; Before performing pixel interception on the chassis image according to the prior information of the defect position, it further includes: Identifying the chassis image to determine the identified visual feature information and envelope position information of the components included in the chassis image; The pixel interception of the chassis image according to the prior information of the defect position includes: screening the identified visual feature information by using the target visual feature information to determine the identified visual feature information representing the target component; Performing pixel interception on the chassis image based on the envelope position information associated with the identified visual feature information representing the target component.
4. The method according to any one of claims 1 to 3, characterized in that, The performing defect detection on at least one regional image to identify whether there are defects on the target component includes: Performing defect identification on at least one of the regional images to obtain identified defect information, and the identified defect information includes the category of the identified defect; Filtering the category of the identified defect by using the prior information of the defect category to obtain the retained defect, and the prior information of the defect category is determined according to the type of the target component; In the case of the existence of the retained defect, taking the retained defect as the defect on the target component.
5. The method according to any one of claims 1-3, characterized in that, The performing defect detection on at least one regional image to identify whether there are defects on the target component includes: Performing defect identification on at least one of the regional images to obtain identified defect information, and the identified defect information includes the size of the identified defect; Filtering the size of the identified defect by using the prior information of the defect size to obtain the retained defect; In the case of the existence of the retained defect, taking the retained defect as the defect on the target component.
6. The method according to claim 5, wherein The performing defect identification on at least one of the regional images to obtain identified defect information includes: Performing defect identification on at least one of the regional images to obtain a defect bounding box enclosing the identified defect; Based on the pixel size of the defect bounding box and the pre-determined pixel-physical size conversion relationship, determining the size of the target box and taking the size of the target box as the size of the identified defect.
7. A chassis defect detection device, characterized in that, Including: An interception module, configured to perform pixel interception on the chassis image according to the prior information of the defect position to obtain at least one regional image representing the target component; A detection module, configured to perform defect detection on at least one regional image to identify whether there are defects on the target component.
8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program or instructions, and the program or instructions cause the computer to execute the steps of the method according to any one of claims 1 to 6.
9. An electronic device, characterized in that, Including: One or more processors; A memory for storing one or more programs or instructions; The processor is configured to execute the steps of the method according to any one of claims 1 to 6 by calling the programs or instructions stored in the memory.