Vehicle plastic part detection method and system, storage medium and computer

Abstract

The invention provides a vehicle plastic part detection method and system, a storage medium and a computer, and the method comprises the steps: obtaining a to-be-detected image of a vehicle plastic part in real time, constructing a corresponding separation threshold value, and carrying out the feature separation of the to-be-detected image according to the separation threshold value, and obtaining a feature image; adding regression prediction in the convolutional neural network model to obtain a deep learning model; obtaining a detected marked image, carrying out fuzzy processing on the marked image, carrying out ambiguity calculation on the obtained blurred image and the marked image, and inputting the obtained ambiguity into a deep learning model for optimization so as to obtain a deep learning optimization model; and inputting the feature image into the deep learning optimization model for image processing, and generating a detection result of the to-be-detected image according to an image processing result. According to the method, the deep learning model is optimized by using the ambiguity, and the feature image is processed by using the constructed deep learning model, so that the corresponding detection result is generated.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, in particular to a vehicle plastic part detection method and system, a storage medium and a computer. BACKGROUND

[0002] With the rapid development of science and technology and the improvement of people's living standards, vehicles have become an indispensable part of people's lives.

[0003] In the vehicle production process, the detection of vehicle plastic parts as one of the necessary processes, usually uses visual image acquisition, uses the abnormal database collected by manual collection to compare the collected images, and uses threshold segmentation algorithm to identify potential defects in the image. This method relies on manually collected data and template matching, and has poor fault tolerance for posture deviation and partial occlusion, resulting in insufficient algorithm robustness. Moreover, this method cannot achieve real-time detection of vehicle plastic parts and cannot be combined with the production line. SUMMARY

[0004] Therefore, the purpose of the present application is to provide a vehicle plastic part detection method, system, storage medium and computer to at least solve the above technical problems.

[0005] The present application provides a vehicle plastic part detection method, comprising: real-time acquisition of a vehicle plastic part detection image, and construction of a corresponding separation threshold, feature separation of the detection image according to the separation threshold to obtain a corresponding feature image; construction of a convolutional neural network model, and addition of regression prediction in the convolutional neural network model to obtain a deep learning model; acquisition of a labeled image that has completed detection, and fuzzy processing of the labeled image, fuzzy degree calculation of the obtained fuzzy image and the labeled image, input of the obtained fuzzy degree into the deep learning model for optimization to obtain a deep learning optimization model; image processing of the feature image in the deep learning optimization model, and generation of a detection result of the detection image according to the image processing result.

[0006] Further, the step of real-time acquisition of a detection image and construction of a corresponding separation threshold, feature separation of the detection image according to the separation threshold to obtain a corresponding feature image comprises: acquisition of the image ratio of a target image and a background image in a plurality of standard images based on a knowledge graph, and construction of a corresponding background threshold according to the image ratio; According to the background threshold, a foreground image and a background image in the image to be detected are separated, and the number of pixel points of the foreground image and the background image and the corresponding average gray value are counted respectively; An optimal separation threshold is calculated according to the number of pixel points of the foreground image and the background image and the corresponding average gray value, and the image to be detected is separated by using the separation threshold, so as to obtain a corresponding feature image.

[0007] Further, the calculation formula of the background threshold is: ; In the formula, , indicates the histogram distribution of the standard image, ; , indicates the size of the standard image; , indicates the image ratio of the target image and the background image; The calculation formula of the separation threshold is: ; In the formula, , , and respectively indicate the number of pixel points of the foreground image and the background image, , , and respectively indicate the average gray value of the foreground image and the background image.

[0008] Further, the steps of constructing a convolutional neural network model and adding a regression prediction in the convolutional neural network model to obtain a deep learning model include: The maximum pooling layer in the convolutional neural network model is replaced by a processing layer containing a region detection algorithm, wherein the input data of the convolutional neural network model includes source data and a region detection window generated by the source data; A regression prediction is added to the output of the processing layer to construct a corresponding regression prediction layer to obtain a deep learning model.

[0009] Further, the steps of performing blur processing on the marked image and calculating the blur degree of the obtained blurred image and the marked image include: The marked image is subjected to blur processing to construct degradation information of the marked image, convolution processing is performed based on the degradation information, and the absolute values of the elements of the result matrix are accumulated to obtain a first blur value; The degradation information is subjected to gray value calculation, and a corresponding second blur value is calculated according to the calculated gray value, and a corresponding blur degree is calculated according to the first blur value and the second blur value.

[0010] The application further provides a vehicle plastic part detection system, which comprises: a feature separation module, configured to acquire a to-be-detected image of a vehicle plastic part in real time, and construct a corresponding separation threshold, and perform feature separation on the to-be-detected image according to the separation threshold to obtain a corresponding feature image; a model construction module, configured to construct a convolutional neural network model, and add a regression prediction in the convolutional neural network model to obtain a deep learning model; a model optimization module, configured to acquire a labeled image after detection is completed, perform blur processing on the labeled image, perform blur degree calculation on a blur image obtained and the labeled image, input the obtained blur degree to the deep learning model for optimization to obtain a deep learning optimization model; an image detection module, configured to input the feature image to the deep learning optimization model for image processing, and generate a detection result of the to-be-detected image according to the image processing result.

[0011] Further, the feature separation module comprises: a threshold construction unit, configured to acquire an image ratio of a target image and a background image in a plurality of standard images based on a knowledge graph, and construct a corresponding background threshold according to the image ratio; a data processing unit, configured to separate a foreground image and a background image in the to-be-detected image according to the background threshold, and respectively count pixel point numbers and corresponding average gray values of the foreground image and the background image; a feature separation unit, configured to calculate an optimal separation threshold according to the pixel point numbers and the corresponding average gray values of the foreground image and the background image, and perform feature separation on the to-be-detected image by using the separation threshold to obtain a corresponding feature image.

[0012] Further, the model construction module comprises: a structure replacement unit, configured to replace a maximum pooling layer in the convolutional neural network model with a processing layer containing a region detection algorithm, wherein input data of the convolutional neural network model comprises source data and a region detection window generated by the source data; a regression prediction unit, configured to add a regression prediction at an output of the processing layer, and construct a corresponding regression prediction layer to obtain a deep learning model.

[0013] Further, the model optimization module comprises: a blur processing unit, configured to perform blur processing on the labeled image to construct degradation information of the labeled image, perform convolution processing based on the degradation information, and accumulate absolute values of elements of a result matrix to obtain a first blur value; A blur degree calculation unit is configured to calculate a gray value of the degradation information, calculate a second blur value corresponding to the calculated gray value, and calculate a blur degree corresponding to the first blur value and the second blur value.

[0014] The application further provides a storage medium, which has a computer program stored thereon, and the computer program is executed by a processor to realize the detection method of the vehicle plastic part.

[0015] The application further provides a computer, which comprises a memory, a processor and a computer program stored in the memory and executable on the processor, and the processor realizes the detection method of the vehicle plastic part when executing the computer program.

[0016] The detection method of the vehicle plastic part, the system, the storage medium and the computer in the application separate the features of the to-be-detected image by constructing a separation threshold, increase regression prediction in the constructed convolutional neural network model to obtain a deep learning model, so that the processing accuracy and efficiency of the model are improved, thereby saving time cost; the marked image after detection is completed is subjected to fuzzy processing, the blur degree of the marked image is calculated by using the fuzzy image obtained by the fuzzy processing, the two images are quickly distinguished by using the calculated blur degree, the deep learning model is optimized by using the blur degree, and the feature image is subjected to image processing by using the constructed deep learning model to generate a corresponding detection result. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 A flowchart of the detection method of the vehicle plastic part in the first embodiment of the application; Figure 2 A structural block diagram of the detection system of the vehicle plastic part in the second embodiment of the application; Figure 3 A structural block diagram of the computer in the third embodiment of the application.

[0018] The following specific embodiments will further illustrate the application in combination with the above-mentioned drawings. DETAILED DESCRIPTION

[0019] In order to facilitate the understanding of the application, the application will be described more fully below with reference to the related drawings. The drawings show several embodiments of the application. However, the application can be realized in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the disclosure of the application more thorough and comprehensive.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in the description herein is for describing particular embodiments only and is not intended to be limiting of the application. The use herein of the terms "and / or" includes a set of one or more associated listed items.

[0021] Embodiment one Please refer to Figure 1 , it is a detection method of vehicle plastic parts in the first embodiment of the application, the method specifically includes steps S101 to S104: S101, real-time acquisition of vehicle plastic parts to be detected image, and construct the corresponding separation threshold, according to the separation threshold of the image to be detected feature separation, to get the corresponding feature image; Further, the step S101 specifically includes steps S1011~S1013: S1011, based on the knowledge graph to obtain the image ratio of target image and background image in a plurality of standard images, and according to the image ratio to construct the corresponding background threshold; S1012, according to the background threshold, the foreground image and the background image in the image to be detected are separated, and the pixel point number and the corresponding average gray value of the foreground image and the background image are counted respectively; S1013, according to the pixel point number and the corresponding average gray value of the foreground image and the background image, the optimal separation threshold is calculated, and the feature separation of the image to be detected is carried out by using the separation threshold, to get the corresponding feature image.

[0022] In the specific implementation, the image acquisition device installed on the production line is used to collect the image of the vehicle plastic parts, and the image ratio of target image and background image in a plurality of standard images is obtained based on the knowledge graph of the vehicle plastic parts, wherein the standard image is the image data of the segmented foreground and background, the same type of standard image as the vehicle plastic parts is used to obtain the corresponding image ratio, and the background threshold for distinguishing foreground and background is constructed by using the image ratio: ; In the formula, , represents the histogram distribution of the standard image, ; , represents the size of the standard image; , represents the image ratio of target image and background image; Specifically, the foreground image and the background image in the to-be-detected image are separated according to the background threshold value, and the number of pixel points and the corresponding average gray value of the foreground image and the background image are counted respectively, and the optimal separation threshold value is constructed according to the number of pixel points and the average gray value, wherein the calculation formula of the separation threshold value is: ; In the formula, 、 respectively represent the number of pixel points of the foreground image and the background image, 、 respectively represent the average gray value of the foreground image and the background image.

[0023] In the calculation process, the gray value is traversed from 0 to 255, and the separation threshold value calculated each time is calculated, the gray value when the maximum value is selected as the optimal threshold value, and the optimal separation threshold value obtained is used to separate the features of the to-be-detected image to obtain the corresponding feature image. It can be understood that the above-mentioned method can increase the difference between the foreground and the background in the to-be-detected image, and the binary method can have better robustness.

[0024] S102, constructing a convolutional neural network model, and adding a regression prediction in the convolutional neural network model to obtain a deep learning model; Further, the step S102 specifically includes steps S1021-S1022: S1021, replacing the maximum pooling layer in the convolutional neural network model with a processing layer containing a region detection algorithm, wherein the input data of the convolutional neural network model includes source data and a region detection window generated by the source data; S1022, adding a regression prediction to the output of the processing layer to construct a corresponding regression prediction layer to obtain a deep learning model.

[0025] In specific implementation, a preliminary convolutional neural network model is constructed based on a convolutional neural network (any one of CNN algorithm, RNN algorithm, GAN algorithm and MLP algorithm is used in the embodiment), the network structure used is any one of LeNet-5 convolutional neural network structure, AlexNet convolutional neural network structure and VGGNet convolutional neural network structure, the maximum pooling layer in the convolutional neural network model is replaced with a processing layer containing a region detection algorithm, wherein the input data of the convolutional neural network model includes source data and a region detection window generated by the source data, a regression prediction is added to the output of the processing layer to construct a corresponding regression prediction layer to obtain a deep learning model; It can be understood that by replacing the max pooling layer, the caching processing of the region detection window feature can be avoided, and the increase of the regression prediction in the output of the processing layer can convert the traditional hierarchical structure training into an end-to-end training mode.

[0026] Specifically, in the constructed deep learning model, the network output layer includes two, one is a target classification layer, and the other is a regression prediction layer, the target classification layer is used to output the image classification result, and the regression prediction layer is used to output the probability vector of the region detection window generated by the image , wherein, The value is transmitted through the full connection layer of the neural network, and the Softmax layer is calculated.

[0027] S103, obtaining a labeled image that has completed detection, and performing blur processing on the labeled image, calculating the blur degree of the obtained blurred image and the labeled image, inputting the obtained blur degree into the deep learning model for optimization to obtain a deep learning optimization model; Further, the step S103 specifically includes steps S1031-S1032: S1031, performing blur processing on the labeled image to construct degradation information of the labeled image, performing convolution processing based on the degradation information, and accumulating the absolute values of the elements of the result matrix to obtain a first blur value; S1032, calculating the gray value of the degradation information, and calculating a corresponding second blur value according to the calculated gray value, and calculating a corresponding blur degree according to the first blur value and the second blur value.

[0028] In specific implementation, a labeled image that has completed detection is obtained, and blur processing is performed on the labeled image to construct degradation information of the labeled image: ; In the formula, denotes a blur kernel, denotes a labeled image, denotes noise of the labeled image; In order to construct a corresponding blur degree by using the labeled image and the degradation information to realize the distinction of the two kinds of images, specifically, a LoG convolution kernel is used to perform convolution calculation on the obtained degradation information, and the absolute values of the elements of the obtained result matrix are accumulated to obtain a corresponding first blur value, wherein, denotes the gray value of the image at , denotes the size of the image, denotes the first blur value: ; In the formula,​ Represents the LoG convolution kernel; Furthermore, grayscale values ​​are calculated for the degraded information, and the corresponding second fuzzy value is calculated based on the calculated grayscale values. : ; Calculate the image in frequency coordinates Complex values ​​at the location (including amplitude and phase information): ; In the formula, The imaginary unit, This represents the basis functions of a two-dimensional sine wave. Calculate the corresponding weighted fuzzy value: ; ; ; In the formula, Indicates the phase angle. Indicates the intensity of the cosine component. Indicates the intensity of the sinusoidal component. This represents the power spectrum obtained by squaring the image after Fourier transform. This represents the high-frequency value obtained after calculation using the Gaussian weighting function. Represents local neighborhood The average phase.

[0029] The corresponding ambiguity is calculated based on the obtained first ambiguity value, second ambiguity value, and weighted ambiguity value: ; In the formula, Indicates adaptive weights, This represents the confidence level correction factor. This represents the variance correction factor. Indicates the first fuzzy value Second fuzzy value and weighted fuzzy value variance Indicates the first fuzzy value Second fuzzy value and weighted fuzzy value The maximum value, Indicates the first fuzzy value Second fuzzy value and weighted fuzzy value The mean.

[0030] Further, the obtained ambiguity is input into the deep learning model for optimization, and the ambiguity is used to improve the fuzzy adaptive convolution and fuzzy perception attention of the deep learning model, so as to obtain a deep learning optimization model.

[0031] S104, input the feature image into the deep learning optimization model for image processing, and generate a detection result of the to-be-detected image according to the image processing result.

[0032] In specific implementation, the obtained feature image is input into the deep learning optimization model for image processing, so as to process the image by using the feature algorithm in the deep learning optimization model, and generate a detection result of the to-be-detected image according to the image processing result.

[0033] In summary, the detection method of the vehicle plastic part in the above embodiment of the application separates the features of the to-be-detected image by constructing a separation threshold, and increases regression prediction in the constructed convolutional neural network model to obtain a deep learning model, so that the processing accuracy and efficiency of the model are improved, thereby saving time cost; the fuzzy processing is performed on the labeled image after the detection is completed, the ambiguity of the labeled image is calculated by using the fuzzy image obtained by the fuzzy processing, the two images are quickly distinguished by using the calculated ambiguity, the deep learning model is optimized by using the ambiguity, and the feature image is processed by using the constructed deep learning model to generate a corresponding detection result.

[0034] Embodiment two Another aspect of the application also provides a detection system for vehicle plastic parts, please refer to Figure 2 , which is a detection system for vehicle plastic parts in the second embodiment of the application, the system comprises: The feature separation module 11 is used for acquiring a to-be-detected image of a vehicle plastic part in real time, constructing a corresponding separation threshold, and separating the features of the to-be-detected image according to the separation threshold to obtain a corresponding feature image; Further, the feature separation module 11 comprises: The threshold construction unit is used for acquiring the image ratio of the target image and the background image in a plurality of standard images based on the knowledge graph, and constructing a corresponding background threshold according to the image ratio; The data processing unit is used for separating the foreground image and the background image in the to-be-detected image according to the background threshold, and respectively counting the pixel point number and the corresponding average gray value of the foreground image and the background image; A feature separation unit is configured to calculate an optimal separation threshold based on the number of pixels and the corresponding average gray value of the foreground image and the background image, and separate features of the image to be detected based on the separation threshold to obtain a corresponding feature image.

[0035] A model construction module 12 is configured to construct a convolutional neural network model and add a regression prediction in the convolutional neural network model to obtain a deep learning model. Further, the model construction module 12 includes: A structure replacement unit is configured to replace a maximum pooling layer in the convolutional neural network model with a processing layer including a region detection algorithm, wherein the input data of the convolutional neural network model includes source data and a region detection window generated based on the source data. A regression prediction unit is configured to add a regression prediction to an output of the processing layer to construct a corresponding regression prediction layer to obtain a deep learning model.

[0036] A model optimization module 13 is configured to obtain a labeled image after detection is completed, blur the labeled image, calculate a blur degree of the blurred image and the labeled image, input the calculated blur degree to the deep learning model for optimization to obtain an optimized deep learning model. Further, the model optimization module 13 includes: A blurring unit is configured to blur the labeled image to construct degradation information of the labeled image, perform convolution based on the degradation information, and accumulate absolute values of elements of a result matrix to obtain a first blur value. A blur degree calculation unit is configured to calculate a gray value of the degradation information, calculate a second blur value based on the calculated gray value, and calculate a blur degree based on the first blur value and the second blur value.

[0037] An image detection module 14 is configured to input the feature image to the optimized deep learning model for image processing, and generate a detection result of the image to be detected based on the image processing result.

[0038] The functions or operation steps realized when the above modules and units are executed are substantially the same as those of the above method embodiments, and thus will not be described herein.

[0039] The vehicle plastic part detection system provided in the embodiments of the present application has the same implementation principle and technical effects as the above method embodiments, and for brevity, the system embodiments not mentioned in the above description can be referred to the corresponding contents in the above method embodiments.

[0040] Embodiment Three The application further provides a computer, please refer to Figure 3 Fig. 3 shows a computer in the third embodiment of the application, which comprises a memory 10, a processor 20, and a computer program 30 stored in the memory 10 and executable on the processor 20, and the processor 20 implements the above-mentioned detection method of vehicle plastic parts when executing the computer program 30.

[0041] The memory 10 comprises at least one type of storage medium, such as flash memory, hard disk, multimedia card, card-type memory (e.g. SD or DX memory, etc.), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 10 can be an internal storage unit of the computer, such as a hard disk of the computer. In other embodiments, the memory 10 can also be an external storage device, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. Further, the memory 10 can comprise both an internal storage unit and an external storage device of the computer. The memory 10 can be used not only to store application software and various data installed on the computer, but also to temporarily store data that has been output or will be output.

[0042] The processor 20 can be an electronic control unit (ECU), a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip, which is used to run program codes or process data stored in the memory 10, such as executing access restriction programs, in some embodiments.

[0043] It should be noted that, Figure 3 The structures shown do not constitute a limitation on the computer, and in other embodiments, the computer can comprise fewer or more components than shown, or combine certain components, or have different component arrangements.

[0044] The embodiments of the application further provide a storage medium having a computer program stored thereon, and the program is executed by a processor to implement the above-mentioned detection method of vehicle plastic parts.

[0045] Those skilled in the art can appreciate that the logic and / or steps represented in the flow diagrams, or otherwise described herein, for example, can be considered as a list of executable instructions for implementing logic functions, and can be embodied in any computer-readable medium for use by an instruction execution system, apparatus, or device, such as a computer-based system, processor- based system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions, or a combination of the above. As used in this description, a "computer-readable medium" can be any apparatus that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device.

[0046] More specific examples (a non-exhaustive list) of the computer-readable medium include the following: an electrical connection (electronic) having one or more wires, a portable computer diskette (magnetic), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium can also be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, for example, via an optical scanner, then compiled, interpreted, or otherwise processed, and stored in a computer memory in a form that is then employable by a computer.

[0047] It should be understood that aspects of the application can be implemented in hardware, software, firmware or combinations thereof. In the above embodiments, various steps or methods can be implemented in software or firmware that is stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, any of the following technologies, known in the art, or combinations thereof, can be used: a discrete logic circuit having logic gates for implementing logic functions upon data signals, an application specific integrated circuit having appropriate combinational logic gates, a programmable gate array (PGA), a field programmable gate array (FPGA), or the like.

[0048] The technical features of the above-described embodiments can be combined in any manner, and for the sake of brevity, not all possible combinations are described, however, any combination of the technical features is considered to be within the scope of the present specification.

[0049] The above embodiments only express several implementation ways of the present application, and the description is more specific and detailed, but it should not be understood as a limitation to the patent scope of the application. It should be pointed out that for ordinary skilled in the art, without departing from the concept of the present application, several modifications and improvements can be made, which are all within the protection scope of the present application. Therefore, the protection scope of the patent of the present application should be subject to the appended claims.

Claims

1. A method of inspecting a vehicle plastic part, characterized in that, The method comprises the following steps: real-time acquisition of a vehicle plastic part to be detected image, and constructing a corresponding separation threshold, and performing feature separation on the to-be-detected image according to the separation threshold to obtain a corresponding feature image; constructing a convolutional neural network model and adding a regression prediction in the convolutional neural network model to obtain a deep learning model; obtaining a labeled image after detection is completed, and performing blur processing on the labeled image, calculating the blur degree of the obtained blurred image and the labeled image, inputting the obtained blur degree into the deep learning model for optimization to obtain a deep learning optimization model; inputting the feature image into the deep learning optimization model for image processing, and generating a detection result of the to-be-detected image according to the image processing result.

2. The method of claim 1, wherein The step of real-time acquisition of a to-be-detected image and constructing a corresponding separation threshold, and performing feature separation on the to-be-detected image according to the separation threshold to obtain a corresponding feature image comprises: obtaining the image ratio of a target image and a background image in a plurality of standard images based on a knowledge graph, and constructing a corresponding background threshold according to the image ratio; separating the foreground image and the background image in the to-be-detected image according to the background threshold, and respectively counting the pixel point number and the corresponding average gray value of the foreground image and the background image; calculating the optimal separation threshold according to the pixel point number and the corresponding average gray value of the foreground image and the background image, and performing feature separation on the to-be-detected image by using the separation threshold to obtain a corresponding feature image.

3. The method of claim 2, wherein The calculation formula of the background threshold is: ； wherein represents a histogram distribution of the standard image, ; represents a size of the standard image; represents an image ratio of the target image to the background image; The calculation formula of the separation threshold is: ； In the formula, , respectively represent the number of pixel points of the foreground image and the background image, , respectively represent the average gray value of the foreground image and the background image.

4. The method of claim 1, wherein The step of constructing a convolutional neural network model and adding a regression prediction in the convolutional neural network model to obtain a deep learning model comprises: replacing the maximum pooling layer in the convolutional neural network model with a processing layer containing a region detection algorithm, wherein the input data of the convolutional neural network model includes source data and a region detection window generated by the source data; adding a regression prediction to the output of the processing layer to construct a corresponding regression prediction layer to obtain a deep learning model.

5. The method of claim 1, wherein The step of performing blur processing on the labeled image and calculating the blur degree of the obtained blurred image and the labeled image comprises: performing blur processing on the labeled image to construct degradation information of the labeled image, performing convolution processing based on the degradation information, and accumulating the absolute values of the elements of the result matrix to obtain a first blur value; calculating the gray value of the degradation information, calculating a corresponding second blur value according to the calculated gray value, and calculating a corresponding blur degree according to the first blur value and the second blur value.

6. A system for detecting a vehicle plastic part, characterized in that The method comprises the following steps: a feature separation module for real-time acquisition of a vehicle plastic part to be detected image, and constructing a corresponding separation threshold, and performing feature separation on the to-be-detected image according to the separation threshold to obtain a corresponding feature image; a model construction module for constructing a convolutional neural network model and adding a regression prediction in the convolutional neural network model to obtain a deep learning model; The model optimization module is configured to obtain a labeled image after detection is completed, perform blur processing on the labeled image, perform blur degree calculation on the obtained blurred image and the labeled image, input the obtained blur degree to the deep learning model for optimization to obtain a deep learning optimization model; The image detection module is configured to input the feature image into the deep learning optimization model for image processing, and generate a detection result of the to-be-detected image according to the image processing result.

7. The system for detecting a vehicle plastic part of claim 6, wherein, The feature separation module comprises: A threshold construction unit is configured to obtain an image ratio of a target image and a background image in a plurality of standard images based on a knowledge graph, and construct a corresponding background threshold according to the image ratio; A data processing unit is configured to separate a foreground image and a background image in the to-be-detected image according to the background threshold, and respectively count pixel point numbers and corresponding average gray values of the foreground image and the background image; A feature separation unit is configured to calculate an optimal separation threshold according to the pixel point numbers and the corresponding average gray values of the foreground image and the background image, and perform feature separation on the to-be-detected image by using the separation threshold to obtain a corresponding feature image.

8. The system for detecting vehicle plastic parts of claim 6, wherein, The model construction module comprises: A structure replacement unit is configured to replace a maximum pooling layer in the convolutional neural network model with a processing layer comprising a region detection algorithm, wherein input data of the convolutional neural network model comprises source data and a region detection window generated by the source data; A regression prediction unit is configured to add regression prediction to an output of the processing layer to construct a corresponding regression prediction layer to obtain a deep learning model.

9. A storage medium having stored thereon a computer program, characterized in that The program is executed by the processor to implement the vehicle plastic part detection method of any one of claims 1 to 5.

10. A computer comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the vehicle plastic part detection method of any one of claims 1 to 5.

Images