File cabinet production and processing management system and method

Abstract

The invention relates to the field of file cabinet production and detection, and provides a file cabinet production and processing management system and method, which adopt an artificial intelligence detection technology based on deep learning to judge whether the spraying quality of a file cabinet is qualified or not by analyzing a multi-surface image of a target detection file cabinet. Therefore, the automatic detection of the spraying quality of the file cabinet can be realized, and the efficiency and the accuracy of the spraying quality detection are greatly improved.

Description

Technical Field

[0001] The present application relates to the field of intelligent detection, and more specifically, to a file cabinet production and processing management system and method. Background Art

[0002] File cabinets are cabinets for storing documents and materials. They are generally used in offices, archives, data rooms, and storage rooms. Steel file cabinets are the most popular. Steel file cabinets are green and environmentally friendly, easy to move and transport over long distances, and are mainly used to store long-term documents. File cabinets usually need to be sprayed with a protective film on their surface to isolate air and moisture, prevent rust and corrosion on the metal surface, and thus extend the service life of the file cabinet.

[0003] Traditional file cabinet spray quality inspection still relies on manual visual judgment. Manual inspection is easily affected by subjective factors, leading to misjudgment, and manual inspection efficiency is low.

[0004] Therefore, an optimized file cabinet production and processing management solution is needed. Summary of the invention

[0005] In view of the shortcomings of the prior art, this application provides a file cabinet production and processing management system and method, which adopts artificial intelligence detection technology based on deep learning, and analyzes the multi-faceted images of the target detection file cabinet to judge whether the file cabinet spraying quality is qualified. In this way, the automatic detection of the file cabinet spraying quality can be realized, which greatly improves the efficiency and accuracy of the spraying quality detection.

[0006] According to one aspect of the present application, a file cabinet production and processing management system is provided, which includes:

[0007] A file cabinet image data acquisition module, used to obtain multi-faceted images of the target detection file cabinet;

[0008] A file cabinet image preprocessing module, used for performing image preprocessing on the multi-faceted images of the target detection file cabinet to obtain a plurality of preprocessed file cabinet detection images;

[0009] A file cabinet spray quality detection feature generation module, used for performing spray quality feature encoding on the plurality of pre-processed file cabinet detection images to obtain a file cabinet spray quality detection feature vector;

[0010] The file cabinet spraying quality detection result generation module is used to determine whether the target detection file cabinet spraying is qualified based on the file cabinet spraying quality detection feature vector.

[0011] According to another aspect of the present application, a method for managing the production and processing of a file cabinet is provided, comprising:

[0012] Obtain multi-faceted images of the target detection filing cabinet;

[0013] Performing image preprocessing on the multi-faceted images of the target detection file cabinet to obtain a plurality of preprocessed file cabinet detection images;

[0014] Performing spraying quality feature encoding on the plurality of pre-processed file cabinet detection images to obtain a file cabinet spraying quality detection feature vector;

[0015] Based on the file cabinet spraying quality detection feature vector, it is determined whether the target detection file cabinet spraying is qualified.

[0016] This application has significant technical effects due to the adoption of the above technical solutions:

[0017] The file cabinet production and processing management system and method provided by the present application adopts artificial intelligence detection technology based on deep learning, and analyzes the multi-faceted images of the target detection file cabinet to judge whether the spraying quality of the file cabinet is qualified. In this way, the automatic detection of the spraying quality of the file cabinet can be realized, which greatly improves the efficiency and accuracy of the spraying quality detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0019] Figure 1 The present invention is a flowchart of a method for managing the production and processing of file cabinets according to an embodiment of the present application.

[0020] Figure 2 This is a flowchart of performing spray quality feature encoding on the plurality of pre-processed file cabinet detection images to obtain a file cabinet spray quality detection feature vector in a file cabinet production and processing management method according to an embodiment of the present application.

[0021] Figure 3 This is a flowchart of performing feature extraction on the multiple pre-processed file cabinet detection images to obtain multiple file cabinet surface spraying global feature images in the file cabinet production and processing management method according to an embodiment of the present application.

[0022] Figure 4 This is a flowchart of performing Euclidean distance calculation on the global feature maps of surface spraying of multiple file cabinets to obtain the feature vector of the file cabinet spraying quality detection in the file cabinet production and processing management method according to an embodiment of the present application.

[0023] Figure 5 This is a flowchart of judging whether the spraying of the target detected file cabinet is qualified based on the file cabinet spraying quality detection feature vector in the file cabinet production and processing management method according to an embodiment of the present application.

[0024] Figure 6 It is a system block diagram of a file cabinet production and processing management system according to an embodiment of the present application. DETAILED DESCRIPTION

[0025] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0026] As an indispensable storage tool in offices, archives, data rooms, and storage rooms, the design and material selection of filing cabinets are crucial to ensure the safety of files and extend their service life. Among many materials, steel filing cabinets are popular because of their durability, environmental protection, and convenience for long-distance transportation. These filing cabinets are usually used to store important documents and materials that need to be kept for a long time, so their protection measures are particularly important.

[0027] In order to ensure the long-term use of steel filing cabinets, manufacturers will apply a layer of spray protective film on their surface. This process can not only isolate air and moisture, prevent rust and corrosion on the metal surface, but also improve the aesthetics of the filing cabinet to a certain extent. The application of spray protective film is a key step in extending the service life of filing cabinets, and it also reflects the dual pursuit of product durability and aesthetics in the modern office furniture manufacturing industry.

[0028] However, traditional file cabinet spraying quality inspection often relies on manual visual judgment, which has obvious limitations. Manual inspection is not only susceptible to the subjective judgment of inspectors, resulting in an increased possibility of misjudgment, but also inefficiency that is difficult to meet the needs of modern fast-paced office work. Therefore, an optimized file cabinet production and processing management solution is expected.

[0029] In recent years, deep learning and neural networks have been widely used in computer vision, natural language processing, text signal processing and other fields. In addition, deep learning and neural networks have also shown a level close to or even beyond that of humans in image classification, object detection, semantic segmentation, text translation and other fields. The development of deep learning and neural networks has provided new solutions and solutions for the production and processing management of file cabinets.

[0030] Figure 1 Flow chart of the file cabinet production and processing management method according to the embodiment of the present application. Figure 1As shown, the file cabinet production and processing management method according to the embodiment of the present application includes: S110, obtaining a multi-faceted image of the target detection file cabinet; S120, performing image preprocessing on the multi-faceted image of the target detection file cabinet to obtain a plurality of preprocessed file cabinet detection images; S130, performing spray quality feature encoding on the plurality of preprocessed file cabinet detection images to obtain a file cabinet spray quality detection feature vector; S140, judging whether the spraying of the target detection file cabinet is qualified based on the file cabinet spray quality detection feature vector.

[0031] Specifically, in the technical solution of the present application, a multi-faceted image of a target detection file cabinet is obtained. It should be understood that the file cabinet image contains rich information, which is crucial for judging the quality of spraying. Specifically, the image contains the color distribution, glossiness change and various possible defects (such as bubbles, scratches, spray leakage areas, uneven colors, etc.) on the surface of the file cabinet, and these information provide a basic basis for subsequent image processing and analysis. By analyzing the image of the file cabinet, it can be used to determine whether the spraying of the file cabinet is qualified. And because the file cabinet is a three-dimensional object, its spraying quality is not only related to the situation of a certain side, but also to the coordination, consistency of the overall appearance and whether the transition between the various faces is natural. Therefore, only obtaining an image of a certain side of the file cabinet is not enough to fully evaluate its spraying quality. In order to more accurately determine whether the spraying of the file cabinet is qualified, it is necessary to obtain its multi-faceted images, including the front, side, back and possible other special faces (such as top, bottom, etc.). In particular, in a specific embodiment of the present application, a high-definition camera can be used to collect multi-faceted images of the target detection file cabinet.

[0032] The original collected file cabinet image is affected by many factors, such as lighting conditions, shooting angle, camera performance, etc., which will cause the image to have problems such as noise, blur, and insufficient contrast. These problems directly affect the accuracy and efficiency of subsequent image processing and analysis. Therefore, in the technical solution of this application, it is necessary to perform image preprocessing on the multi-faceted images of the target detection file cabinet to obtain multiple preprocessed file cabinet detection images. By preprocessing the original file cabinet image, the image quality can be improved, the processability of the original file cabinet image and the accuracy of information extraction can be improved, thereby providing more reliable basic data for subsequent spray quality judgment.

[0033] In an embodiment of the present application, a possible implementation method for performing image preprocessing on the multi-faceted images of the target detection file cabinet to obtain multiple preprocessed file cabinet detection images may be: a. Noise removal: A filtering algorithm may be applied to the original image according to actual needs to remove noise in the original file cabinet image, such as applying a median filtering algorithm, which can better maintain edge information while removing noise; b. Color correction: A color correction algorithm is used to adjust the image in the original file cabinet to make it more accurate.

[0034] Figure 2 The present invention is a flowchart of performing spraying quality feature encoding on the plurality of pre-processed file cabinet detection images to obtain a file cabinet spraying quality detection feature vector in the file cabinet production and processing management method according to an embodiment of the present application. Figure 2 As shown, the spray quality feature encoding of the multiple pre-processed file cabinet detection images to obtain the file cabinet spray quality detection feature vector includes: S131, feature extraction of the multiple pre-processed file cabinet detection images to obtain multiple file cabinet surface spray global feature maps; S132, Euclidean distance calculation of the multiple file cabinet surface spray global feature maps to obtain the file cabinet spray quality detection feature vector.

[0035] Figure 3 This is a flowchart of extracting features from the plurality of pre-processed file cabinet detection images to obtain a plurality of file cabinet surface spraying global feature images in the file cabinet production and processing management method according to an embodiment of the present application. Figure 3 As shown, the feature extraction of the multiple pre-processed file cabinet detection images to obtain multiple file cabinet surface spraying global feature images includes: S1311, respectively passing the multiple pre-processed file cabinet detection images through a file cabinet surface spraying quality feature encoder to obtain multiple file cabinet surface spraying local feature images; S1312, passing the multiple file cabinet surface spraying local feature images through a non-local neural network to obtain the multiple file cabinet surface spraying global feature images.

[0036] Considering that the evaluation of the quality of file cabinet spraying often involves complex visual features, such as color uniformity, glossiness, surface defects (such as scratches, bubbles, stains, etc.), etc., these features are difficult to accurately judge directly through simple pixel value comparison. Therefore, it is necessary to perform feature extraction on the preprocessed file cabinet detection image to convert these complex visual information in the image into a numerical form that can be understood by the computer, that is, the multiple preprocessed file cabinet detection images are respectively passed through the file cabinet surface spraying quality feature encoder to obtain multiple file cabinet surface spraying local feature maps. Through the processing of the file cabinet surface spraying quality feature encoder, the color features, defect features, etc. of the file cabinet surface can be extracted. In this application, the file cabinet surface spraying quality feature encoder is a convolutional neural network model using convolution kernels of different sizes, wherein the size of the first convolution kernel is larger than the size of the second convolution kernel. It should be understandable that two convolution kernels of different sizes can capture different features in the file cabinet image. The first convolution kernel is larger and can capture more global and fine-grained features, such as the overall color distribution. The second convolution kernel is smaller in size and is better at capturing features such as small defects and local changes. This multi-scale feature extraction helps to better understand the surface features of the file cabinet. Through layer-by-layer convolution and pooling operations, the model can learn hierarchical feature representations of file cabinets from simple to complex. The convolution kernel of the first layer processes the original image and extracts low-level visual features; as the number of network layers increases, the convolution kernels of subsequent layers further extract higher-level and more abstract features based on the output of the previous layer. This hierarchical feature learning helps to improve the generalization and robustness of the model.

[0037] In particular, the process of using the file cabinet surface spray quality feature encoder to process multiple pre-processed file cabinet detection images is the same, so the present application takes the processing of a pre-processed file cabinet detection image as an example. In an embodiment of the present application, a possible implementation method of passing the preprocessed file cabinet detection image through a file cabinet surface spraying quality feature encoder to obtain a file cabinet surface spraying local feature map can be: using each layer of the file cabinet surface spraying quality feature encoder to respectively perform the following on the input data in the forward pass of the layer: using a first convolution kernel to convolve the input data to obtain a first convolution feature map; using a second convolution kernel to convolve the first convolution feature map to obtain a second convolution feature map, wherein the size of the first convolution kernel is larger than the size of the second convolution kernel; performing pooling processing on the second convolution feature map to obtain a pooling feature map; and, performing activation processing on the pooling feature map to obtain an activation feature map; wherein the output of the last layer of the file cabinet surface spraying quality feature encoder is the file cabinet surface spraying local feature map, the input of the first layer of the file cabinet surface spraying quality feature encoder is the preprocessed file cabinet detection image, the size of the first convolution kernel is 5×5, and the size of the second convolution kernel is 3×3.

[0038] Considering that the local feature map of the file cabinet surface spraying can capture the local details of the file cabinet surface, but often ignores the correlation and overall consistency between different local parts, in order to capture the long-range dependencies between different positions on the file cabinet surface, in the technical solution of the present application, the multiple local feature maps of the file cabinet surface spraying are passed through a non-local neural network to obtain multiple global feature maps of the file cabinet surface spraying. Non-local Neural Networks (NLNN) is a special neural network structure that enhances the feature extraction ability of the network by introducing non-local operations. The core idea of ​​non-local operations is to directly calculate the feature correlation between any two positions in the image and use this correlation for feature updating. Specifically, non-local operations can be regarded as weighted summation of the features of each position in the image, where the weight is determined by the feature similarity between the position and all other positions. This operation enables the network to capture the long-range dependencies in the image. Even if the two positions are far apart in space, as long as there is a similarity in features between them, they can be associated with each other in non-local operations. By inputting the multiple local feature maps of the file cabinet surface spraying into the non-local neural network, the expression of the surface features of the file cabinet on each surface can be enhanced, so as to better understand the global spraying quality of the file cabinet surface on each surface.

[0039] Similarly, the present application uses the processing of a local feature map of the surface spraying of a file cabinet as an example to illustrate the processing process. In an embodiment of the present application, a possible implementation method of passing the local feature map of the surface spraying of the file cabinet through a non-local neural network to obtain a global feature map of the surface spraying of the file cabinet may be: performing a first point convolution processing, a second point convolution processing, and a third point convolution processing on the local feature map of the surface spraying of the file cabinet respectively to obtain a first file cabinet surface spraying feature map, a second file cabinet surface spraying feature map, and a third file cabinet surface spraying feature map; calculating the position-weighted sum between the first file cabinet surface spraying feature map and the second file cabinet surface spraying feature map to obtain a file cabinet surface spraying fusion feature map; inputting the file cabinet surface spraying fusion feature map into a Softmax function to convert the file cabinet surface spraying fusion feature map into a Softmax function. The characteristic values ​​of each position in the figure are mapped to the probability space to obtain a normalized file cabinet surface spray fusion feature map; the position point multiplication between the normalized file cabinet surface spray fusion feature map and the third file cabinet surface spray feature map is calculated to obtain a file cabinet surface spray re-fusion feature map; the file cabinet surface spray re-fusion feature map is embedded in a Gaussian similarity function to obtain a global similarity feature map; the global similarity feature map is subjected to a fourth point convolution process to adjust the number of channels of the global similarity feature map to obtain a channel-adjusted global similarity feature map; the position-weighted sum of the channel-adjusted global similarity feature map and the file cabinet surface spray local feature map is calculated to obtain the file cabinet surface spray global feature map.

[0040] Figure 4 This is a flowchart of performing Euclidean distance calculation on the multiple file cabinet surface spraying global feature maps to obtain the file cabinet spraying quality detection feature vector in the file cabinet production and processing management method according to the embodiment of the present application. Figure 4 As shown, the Euclidean distance calculation of the multiple file cabinet surface spraying global feature maps to obtain the file cabinet spraying quality detection feature vector includes: S1321, expanding the multiple file cabinet surface spraying global feature maps to obtain multiple file cabinet surface spraying global feature vectors; S1322, calculating the Euclidean distance between any two file cabinet surface spraying global feature vectors in the multiple file cabinet surface spraying global feature vectors to obtain the file cabinet spraying quality detection feature vector composed of multiple Euclidean distances.

[0041] The global feature map of the file cabinet surface spraying contains detailed information on the spraying quality on each surface, such as color uniformity, glossiness, surface defects, etc. Since the global feature map of spraying has a high dimension, it is relatively complex in the model operation. In order to retain the information about the file cabinet surface spraying quality in the feature map and simplify the operation, in the technical solution of the present application, the multiple file cabinet surface spraying global feature maps are expanded to obtain multiple file cabinet surface spraying global feature vectors. By expanding, three-dimensional data can be converted into one-dimensional data, realizing the dimensionality reduction of the spraying quality feature.

[0042] Considering that the global feature vectors of the surface spraying of multiple file cabinets are highly abstract feature representations, they all independently represent the comprehensive information of the corresponding file cabinet surface in terms of spraying quality. In order to quantify the similarity between the spraying qualities of different surfaces of the file cabinet, and to characterize the overall spraying quality of the file cabinet, in the technical solution of the present application, the Euclidean distance between any two of the global feature vectors of the surface spraying of the file cabinet is calculated to obtain a file cabinet spraying quality detection feature vector composed of multiple Euclidean distances. Euclidean distance refers to the straight-line distance between two points in Euclidean space. In the present application, the global feature vector of the spraying of each file cabinet surface can be regarded as a point in a high-dimensional space, and the Euclidean distance is used to measure the distance between these points. The Euclidean distance can reflect the similarity between the feature vectors of the spraying quality of different file cabinet surfaces. When the Euclidean distance between two feature vectors is small, it means that they are close in the spraying quality characteristics and may belong to the same type of spraying quality; on the contrary, if the Euclidean distance is large, it means that there are significant differences in their spraying quality characteristics. When the Euclidean distance between the global feature vector of the spraying on the surface of a certain filing cabinet and the feature vectors of the surfaces of most filing cabinets is far, then the filing cabinet is likely to have a spraying quality problem.

[0043] In an embodiment of the present application, a possible implementation method of calculating the Euclidean distance between any two of the plurality of file cabinet surface spraying global feature vectors to obtain a file cabinet spraying quality detection feature vector composed of a plurality of Euclidean distances may be: calculating the Euclidean distance between every two of the plurality of file cabinet surface spraying global feature vectors using the following distance calculation formula to obtain the plurality of Euclidean distances; wherein the distance calculation formula is:

[0044]

[0045] Among them, V i and V j Respectively represent any two file cabinet surface spraying global feature vectors among the plurality of file cabinet surface spraying global feature vectors, d(V i ,Vj ) represents calculating the Euclidean distance between every two file cabinet surface spraying global feature vectors in the plurality of file cabinet surface spraying global feature vectors, V ik and V jk Respectively represent the feature values ​​of each position of any two file cabinet surface spraying global feature vectors among the multiple file cabinet surface spraying global feature vectors.

[0046] Figure 5 This is a flowchart of judging whether the target detected file cabinet spraying is qualified based on the file cabinet spraying quality detection feature vector in the file cabinet production and processing management method according to the embodiment of the present application. Figure 5 As shown, judging whether the target detected file cabinet is spray-painted qualified based on the file cabinet spray-painted quality detection feature vector comprises: S141, performing classification error viewpoint measurement based on model interpretability reverse analysis on the file cabinet spray-painted quality detection feature vector to obtain an optimized file cabinet spray-painted quality detection feature vector; S142, passing the optimized file cabinet spray-painted quality detection feature vector through a spray-painted quality classifier to obtain a classification result, and the classification result is used to indicate whether the target detected file cabinet is spray-painted qualified.

[0047] In particular, in the technical solution of the present application, it is considered that if the number of samples of certain spray quality categories in the training data is much larger than that of other categories, the spray quality classifier may tend to recognize these common categories and inaccurately recognize uncommon categories. The model may perform well on the training data, but its generalization ability may be insufficient when faced with new or unseen spray quality problems. Feature selection and dimensionality reduction may be required in the process of generating feature vectors, which may lead to the loss of some important information, and further cause the file cabinet spray quality detection feature vector to have a category harmony offset relative to the predetermined class label of the classifier. Therefore, in order to reduce the category harmony offset, in the technical solution of the present application, the classification error viewpoint measurement based on the reverse analysis of model interpretability is performed on the file cabinet spray quality detection feature vector.

[0048] Specifically, in an embodiment of the present application, the classification error viewpoint measurement based on model interpretability reverse analysis is performed on the file cabinet spray quality detection feature vector to obtain an optimized file cabinet spray quality detection feature vector, including: performing probabilistic activation based on the Sigmoid function on the file cabinet spray quality detection feature vector to obtain a probabilistic file cabinet spray quality detection feature vector; inputting the file cabinet spray quality detection feature vector into the spray quality classifier to obtain an initial class label probability value; calculating the absolute value of the difference between the feature value of each position in the probabilistic file cabinet spray quality detection feature vector and the initial class label probability value to obtain a backward machine learning fine-grained error measurement vector; and using the The backward machine learning fine-grained error metric vector is used as a key vector, the file cabinet spray quality detection feature vector is used as a query vector and a value vector, and the key vector, the query vector and the value vector are input into a converter structure to obtain a backward learning type attention optimized file cabinet spray quality detection feature vector; the ratio between the sum of the backward machine learning fine-grained error metric vector and the length of the backward machine learning fine-grained error metric vector is calculated, and the natural exponential function value with the ratio as a power is calculated as an optimized scaling factor; the product between the optimized scaling factor and the backward learning type attention optimized file cabinet spray quality detection feature vector is calculated to obtain the optimized file cabinet spray quality detection feature vector.

[0049] In an embodiment of the present application, another expression method of performing classification error viewpoint measurement based on model interpretability reverse analysis on the file cabinet spray quality detection feature vector to obtain an optimized file cabinet spray quality detection feature vector can be: processing the file cabinet spray quality detection feature vector with the following formula to obtain the file cabinet spray quality detection feature vector; wherein, the formula is:

[0050]

[0051] Among them, V c represents the probabilistic file cabinet spray quality detection feature vector, P represents the initial class label probability value, Indicates subtraction by position, V e represents the fine-grained error metric vector for backward machine learning, v i represents the eigenvalue of the ith position of the backward machine learning fine-grained error metric vector, L represents the length of the backward machine learning fine-grained error metric vector, T represents the transpose of the vector, represents vector multiplication, Softmax represents the normalized exponential function, ⊙ represents point multiplication by position, V ’ C represents the optimized file cabinet spray quality detection feature vector.

[0052] Based on this, in the technical solution of the present application, the classification error viewpoint measurement based on the reverse analysis of model interpretability is performed on the file cabinet spray quality detection feature vector, which first implements the probabilistic activation based on the Sigmoid function on the file cabinet spray quality detection feature vector to obtain a probabilistic file cabinet spray quality detection feature vector, and inputs the file cabinet spray quality detection feature vector into the spray quality classifier to obtain the initial class label probability value. Then, the error between the probabilistic file cabinet spray quality detection feature vector and the initial class label probability value is calculated through Backward Error Analysis to generate a fine-grained error measurement vector, which quantifies the viewpoint measurement contribution of each feature position to the classification error.

[0053] Then, the backward machine learning fine-grained error metric vector is used as a key vector, the file cabinet spray quality detection feature vector is used as a query vector and a value vector, and the key vector, the query vector and the value vector are input into a transformer structure to perform backward machine learning based on an error metric-based class attention mechanism on the file cabinet spray quality detection feature vector through a class Transformer architecture. In this way, the file cabinet spray quality detection feature vector is subjected to a backward interpretability analysis so that the obtained backward learning class attention optimized file cabinet spray quality detection feature vector has a backward learning sensitive preference relative to the predetermined class label of the spray quality classifier. And further, the ratio of the sum of the error metric vector and the vector length is calculated, and the optimized scaling factor obtained by calculating the natural exponential function value with this ratio as an exponent is used to perform backward scaling optimization on the backward learning class attention optimized file cabinet spray quality detection feature vector. In this way, the sensitivity preference and robustness of the file cabinet spray quality detection feature vector relative to the predetermined class label of the spray quality classifier are improved, thereby improving the accuracy of the classification result.

[0054] Finally, the optimized file cabinet spray quality detection feature vector is passed through the spray quality classifier to obtain a classification result, and the classification result is used to indicate whether the target detection file cabinet spray is qualified. It should be understood that the spray quality classifier is a model based on a machine learning algorithm, which accepts the optimized file cabinet spray quality detection feature vector as input and outputs a classification result to indicate whether the target detection file cabinet spray is qualified. This model is usually trained on a large amount of labeled file cabinet surface data and corresponding spray quality labels, and can learn the characteristic differences between qualified and unqualified spray quality. Through the spray quality classifier, rapid and automated detection of the spray quality of a large number of file cabinets can be achieved. Compared with manual detection, this method not only improves the detection efficiency, but also reduces the errors caused by human factors.

[0055] In an embodiment of the present application, the optimized file cabinet spray quality detection feature vector is passed through a spray quality classifier to obtain a classification result, and the classification result is used to indicate whether the target detection file cabinet spraying is qualified. One possible implementation method may be: using the spray quality classifier to process the optimized file cabinet spray quality detection feature vector using the following classification formula to generate the classification result; wherein the classification formula is:

[0056] O=softmax{M c ,B c )|V' c}

[0057] Wherein O is the classification result, V' c represents the optimized file cabinet spray quality detection feature vector, M c is the weight matrix of the fully connected layer, B c Represents the bias vector of the fully connected layer, and softmax is the normalized exponential function.

[0058] In summary, the file cabinet production and processing management method based on the embodiment of the present application is explained, which adopts artificial intelligence detection technology based on deep learning, and analyzes the multi-faceted images of the target detection file cabinet to judge whether the file cabinet spraying quality is qualified. In this way, the automatic detection of the file cabinet spraying quality can be realized, which greatly improves the efficiency and accuracy of the spraying quality detection.

[0059] Figure 6 1 is a system block diagram of a file cabinet production and processing management system according to an embodiment of the present application. Figure 6As shown, according to the file cabinet production and processing management system 100 of the embodiment of the present application, it includes: a file cabinet image data acquisition module 110, which is used to obtain multi-faceted images of the target detection file cabinet; a file cabinet image preprocessing module 120, which is used to perform image preprocessing on the multi-faceted images of the target detection file cabinet to obtain multiple preprocessed file cabinet detection images; a file cabinet spray quality detection feature generation module 130, which is used to perform spray quality feature encoding on the multiple preprocessed file cabinet detection images to obtain a file cabinet spray quality detection feature vector; a file cabinet spray quality detection result generation module 140, which is used to determine whether the target detection file cabinet is sprayed qualified based on the file cabinet spray quality detection feature vector.

[0060] Here, those skilled in the art will appreciate that the specific functions and operations of the various units and modules in the above-mentioned file cabinet production and processing management system 100 have been described in detail above. Figures 1 to 5 The description of the file cabinet production and processing management method has been introduced in detail, and therefore, its repeated description will be omitted.

[0061] In summary, the file cabinet production and processing management system 100 based on the embodiment of the present application is explained, which adopts artificial intelligence detection technology based on deep learning, and analyzes the multi-faceted images of the target detection file cabinet to judge whether the file cabinet spraying quality is qualified. In this way, the automatic detection of the file cabinet spraying quality can be realized, which greatly improves the efficiency and accuracy of the spraying quality detection.

[0062] As described above, the file cabinet production and processing management system 100 according to the embodiment of the present application can be implemented in various wireless terminals, such as a server for file cabinet production and processing management. In one example, the file cabinet production and processing management system 100 according to the embodiment of the present application can be integrated into the wireless terminal as a software module and / or a hardware module. For example, the file cabinet production and processing management system 100 can be a software module in the operating system of the wireless terminal, or can be an application developed for the wireless terminal; of course, the file cabinet production and processing management system 100 can also be one of the many hardware modules of the wireless terminal.

[0063] Alternatively, in another example, the file cabinet production and processing management system 100 and the wireless terminal may also be separate devices, and the file cabinet production and processing management system 100 may be connected to the wireless terminal via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0064] The basic principles of the present application are described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, effects, etc. mentioned in the present application are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. are required by each embodiment of the present application. In addition, the specific details of the above application are only for the purpose of illustration and ease of understanding, rather than limitation, and the above details do not limit the present application to the use of the above specific details to be implemented.

[0065] In this application, words such as "including", "comprising", "having", etc. are open words, meaning "including but not limited to", and can be used interchangeably therewith. The words "or" and "and" used herein mean the words "and / or" and can be used interchangeably therewith, unless the context clearly indicates otherwise. The word "such as" used herein means the phrase "such as but not limited to", and can be used interchangeably therewith.

[0066] Additionally, as used herein, "or" used in a list of items beginning with "at least one" indicates a separate list, so that, for example, a list of "at least one of A, B, or C" means A or B or C, or AB or AC or BC, or ABC (i.e., A and B and C). Furthermore, the word "exemplary" does not mean that the example described is preferred or better than other examples.

[0067] It should also be noted that in the system and method of the present application, each component or each step can be decomposed and / or recombined. Such decomposition and / or recombination should be regarded as equivalent solutions of the present application.

[0068] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of the present application. Therefore, the present application is not intended to be limited to the aspects shown herein, but rather to the widest scope consistent with the principles and novel features of the present application.

[0069] The above description has been given for the purpose of illustration and description. In addition, this description is not intended to limit the embodiments of the present application to the forms disclosed herein. Although multiple example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, changes, additions and sub-combinations thereof.

Claims

1. A method for the production and processing management of a file cabinet, characterized in that: include: Obtain multi-faceted images of the target detection filing cabinet; Performing image preprocessing on the multi-faceted images of the target detection file cabinet to obtain a plurality of preprocessed file cabinet detection images; Performing spraying quality feature encoding on the plurality of pre-processed file cabinet detection images to obtain a file cabinet spraying quality detection feature vector; Based on the file cabinet spraying quality detection feature vector, it is determined whether the target detection file cabinet spraying is qualified.

2. The method for manufacturing and managing file cabinets according to claim 1, characterized in that: Performing spraying quality feature encoding on the plurality of pre-processed file cabinet detection images to obtain a file cabinet spraying quality detection feature vector includes: Performing feature extraction on the plurality of pre-processed file cabinet detection images to obtain a plurality of file cabinet surface spraying global feature images; Euclidean distance calculation is performed on the multiple file cabinet surface spraying global feature maps to obtain the file cabinet spraying quality detection feature vector.

3. The file cabinet production and processing management method according to claim 2 is characterized in that: Performing feature extraction on the plurality of preprocessed file cabinet detection images to obtain a plurality of file cabinet surface spraying global feature images, including: The plurality of pre-processed file cabinet detection images are respectively passed through a file cabinet surface spraying quality feature encoder to obtain a plurality of file cabinet surface spraying local feature images; The local feature maps of the surface spraying of the multiple file cabinets are passed through a non-local neural network to obtain the global feature maps of the surface spraying of the multiple file cabinets.

4. The method for managing the production and processing of file cabinets according to claim 3 is characterized in that: Performing Euclidean distance calculation on the multiple file cabinet surface spraying global feature maps to obtain the file cabinet spraying quality detection feature vector, including: Expanding the plurality of file cabinet surface spraying global feature maps to obtain a plurality of file cabinet surface spraying global feature vectors; The Euclidean distance between any two of the plurality of file cabinet surface spraying global feature vectors is calculated to obtain the file cabinet spraying quality detection feature vector composed of a plurality of Euclidean distances.

5. The method for managing the production and processing of file cabinets according to claim 4, characterized in that: The file cabinet surface spray quality feature encoder is a convolutional neural network model using convolution kernels of different sizes, wherein the size of the first convolution kernel is larger than the size of the second convolution kernel.

6. The method for managing the production and processing of file cabinets according to claim 5, characterized in that: Based on the file cabinet spraying quality detection feature vector, judging whether the target detection file cabinet spraying is qualified includes: Performing classification error viewpoint measurement based on model interpretability reverse analysis on the file cabinet spray quality detection feature vector to obtain an optimized file cabinet spray quality detection feature vector; The optimized file cabinet spraying quality detection feature vector is passed through a spraying quality classifier to obtain a classification result, and the classification result is used to indicate whether the target detection file cabinet spraying is qualified.

7. The method for managing the production and processing of file cabinets according to claim 6, characterized in that: The classification error viewpoint measurement based on the model interpretability reverse analysis is performed on the file cabinet spray quality detection feature vector to obtain an optimized file cabinet spray quality detection feature vector, including: Performing probabilistic activation on the file cabinet spray quality detection feature vector based on a Sigmoid function to obtain a probabilistic file cabinet spray quality detection feature vector; Inputting the file cabinet spray quality detection feature vector into the spray quality classifier to obtain an initial class label probability value; Calculate the absolute value of the difference between the feature value of each position in the probabilistic file cabinet spray quality detection feature vector and the initial class label probability value to obtain a backward machine learning fine-grained error measurement vector; Using the backward machine learning fine-grained error metric vector as a key vector and the file cabinet spray quality detection feature vector as a query vector and a value vector, the key vector, the query vector and the value vector are input into a converter structure to obtain a backward learning class attention optimized file cabinet spray quality detection feature vector; Calculating a ratio between a sum of the backward machine learning fine-grained error metric vector and a length of the backward machine learning fine-grained error metric vector, and calculating a natural exponential function value raised to a power of the ratio as an optimization scaling factor; The product of the optimized scaling factor and the backward learning type attention optimized file cabinet spray quality detection feature vector is calculated to obtain the optimized file cabinet spray quality detection feature vector.

8. A file cabinet production and processing management system, characterized in that: include: A file cabinet image data acquisition module, used to obtain multi-faceted images of the target detection file cabinet; A file cabinet image preprocessing module, used for performing image preprocessing on the multi-faceted images of the target detection file cabinet to obtain a plurality of preprocessed file cabinet detection images; A file cabinet spray quality detection feature generation module, used for performing spray quality feature encoding on the plurality of pre-processed file cabinet detection images to obtain a file cabinet spray quality detection feature vector; The file cabinet spraying quality detection result generation module is used to determine whether the target detection file cabinet spraying is qualified based on the file cabinet spraying quality detection feature vector.

9. The file cabinet production and processing management system according to claim 8, characterized in that: The file cabinet spray quality detection feature generation module includes: A file cabinet detection image feature extraction unit, used for extracting features from the plurality of pre-processed file cabinet detection images to obtain a plurality of file cabinet surface spraying global feature images; The Euclidean distance calculation unit is used to perform Euclidean distance calculation on the multiple file cabinet surface spraying global feature maps to obtain the file cabinet spraying quality detection feature vector.

10. The file cabinet production and processing management system according to claim 9, characterized in that: The file cabinet spraying quality detection result generation module includes: A file cabinet spray quality detection feature optimization unit, used for performing classification error viewpoint measurement based on model interpretability reverse analysis on the file cabinet spray quality detection feature vector to obtain an optimized file cabinet spray quality detection feature vector; The optimized file cabinet spray quality detection feature classification unit is used to pass the optimized file cabinet spray quality detection feature vector through the spray quality classifier to obtain a classification result, and the classification result is used to indicate whether the target detection file cabinet spray is qualified.

Images