Multifunctional mask and preparation method thereof

Abstract

A multifunctional mask and a preparation method thereof, wherein the activated carbon layer, the HEPA high-efficiency filtration layer and the antibacterial layer are quality-tested based on machine vision to determine whether the surface quality of the activated carbon layer, the HEPA high-efficiency filtration layer and the antibacterial layer is qualified; after determining that the surface quality of the activated carbon layer, the HEPA high-efficiency filtration layer and the antibacterial layer is qualified, the activated carbon layer, the HEPA high-efficiency filtration layer and the antibacterial layer are stacked by a stacking process to obtain a multi-layer filtration structure; the multi-layer filtration structure is installed on a plane mask layer; the bridge of the nose is installed on the plane mask layer; the first ear band and the second ear band are attached to the left and right sides of the plane mask layer. In this way, machine vision technology can be used to intelligently detect the surface quality of the carbon layers of different filtration layers of the multifunctional mask, thereby ensuring that the surface quality of the mask filtration layer meets the standard and improving the filtration efficiency and performance of the mask.

Description

Technical Field

[0001] The present application relates to the technical field of multifunctional masks, and more specifically, to a multifunctional mask and a preparation method thereof. Background Art

[0002] As air pollution worsens, people's demand for respiratory health is growing. Masks, as an important personal protective equipment, are essential to prevent the spread of viruses and bacteria and are widely used in daily life and specific workplaces. Especially for multifunctional masks, it is crucial to ensure their filtration efficiency and effectiveness.

[0003] However, the current multifunctional masks have some limitations in terms of filtration efficiency and effect. This is because the traditional multifunctional mask preparation process usually relies on manual quality inspection of different filter layers, which not only leads to low production efficiency and cannot meet the needs of large-scale production, but also lacks effective quality control methods and cannot accurately evaluate the quality of the filter layer, resulting in unstable filtration efficiency and performance of the mask. In other words, in the preparation of traditional masks, the quality of the filter layer may vary, and the quality of some filter layers does not meet the standards, affecting the overall filtration effect of the mask.

[0004] Therefore, a preparation scheme for an optimized multifunctional mask is desired. Summary of the invention

[0005] In order to solve the above technical problems, this application is proposed. This application provides a multifunctional mask and a preparation method thereof, which performs quality inspection on the activated carbon layer, the HEPA high-efficiency filtration layer and the antibacterial layer based on machine vision to determine whether the surface quality of the activated carbon layer, the HEPA high-efficiency filtration layer and the antibacterial layer is qualified; after determining that the surface quality of the activated carbon layer, the HEPA high-efficiency filtration layer and the antibacterial layer is qualified, the activated carbon layer, the HEPA high-efficiency filtration layer and the antibacterial layer are stacked by a stacking process to obtain a multi-layer filtration structure; the multi-layer filtration structure is installed on the plane mask layer; the bridge of the nose is installed on the plane mask layer; the first ear strap and the second ear strap are attached to the left and right sides of the plane mask layer. In this way, machine vision technology can be used to intelligently detect the surface quality of the carbon layer of different filter layers of the multifunctional mask, thereby ensuring that the surface quality of the mask filter layer meets the standard and improving the filtration efficiency and performance of the mask.

[0006] In a first aspect, a multifunctional mask is provided, comprising:

[0007] A planar mask body, wherein the planar mask body comprises a planar mask layer, a multi-layer filter structure and a nose bridge strip, wherein the nose bridge strip is arranged at the middle position of the upper part of the planar mask layer, and the multi-layer filter structure comprises an activated carbon layer, a HEPA high-efficiency filter layer and an antibacterial layer stacked on each other; and

[0008] A first ear strap and a second ear strap are arranged on opposite left and right sides of the flat mask body.

[0009] In a second aspect, a method for preparing a multifunctional mask is provided, comprising:

[0010] Provide activated carbon layer, HEPA high-efficiency filtration layer and antibacterial layer;

[0011] Performing quality inspection on the activated carbon layer, the HEPA high-efficiency filtration layer and the antibacterial layer based on machine vision to determine whether the surface quality of the activated carbon layer, the HEPA high-efficiency filtration layer and the antibacterial layer is qualified;

[0012] After determining that the surface qualities of the activated carbon layer, the HEPA high-efficiency filter layer and the antibacterial layer are qualified, the activated carbon layer, the HEPA high-efficiency filter layer and the antibacterial layer are stacked by a stacking process to obtain the multi-layer filter structure;

[0013] Installing the multi-layer filtering structure on the planar cover layer;

[0014] Mounting a nose bridge strip on the planar mask layer; and

[0015] Attach the first ear strap and the second ear strap to the left and right sides of the planar cover layer.

[0016] Compared with the prior art, the multifunctional mask and its preparation method provided by the present application perform quality inspection on the activated carbon layer, HEPA high-efficiency filtration layer and antibacterial layer based on machine vision to determine whether the surface quality of the activated carbon layer, HEPA high-efficiency filtration layer and antibacterial layer is qualified; after determining that the surface quality of the activated carbon layer, HEPA high-efficiency filtration layer and antibacterial layer is qualified, the activated carbon layer, HEPA high-efficiency filtration layer and antibacterial layer are stacked by a stacking process to obtain a multi-layer filtration structure; the multi-layer filtration structure is installed on the plane mask layer; the bridge of the nose is installed on the plane mask layer; the first ear band and the second ear band are attached to the left and right sides of the plane mask layer. In this way, machine vision technology can be used to intelligently detect the surface quality of the carbon layer of different filter layers of the multifunctional mask, thereby ensuring that the surface quality of the mask filter layer meets the standard and improving the filtration efficiency and performance of the mask. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0018] Figure 1 Schematic diagram of a multifunctional mask according to an embodiment of the present application.

[0019] Figure 2 Flow chart of a method for preparing a multifunctional mask according to an embodiment of the present application.

[0020] Figure 3 This is a flowchart of sub-step 120 of the method for preparing a multifunctional mask according to an embodiment of the present application.

[0021] Figure 4 It is a block diagram of a system for preparing a multifunctional mask according to an embodiment of the present application.

[0022] Figure 5 This is a schematic diagram of a scenario of a method for preparing a multifunctional mask according to an embodiment of the present application. DETAILED DESCRIPTION

[0023] The following will describe the technical solutions in the embodiments of the present application in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0024] Unless otherwise specified, all technical and scientific terms used in the embodiments of the present application have the same meaning as those commonly understood by those skilled in the art of the present application. The terms used in this application are only for the purpose of describing specific embodiments and are not intended to limit the scope of this application.

[0025] In the embodiments of the present application, it should be noted that, unless otherwise specified and limited, the term "connection" should be understood in a broad sense. For example, it can be an electrical connection or a connection between two components. It can be a direct connection or an indirect connection through an intermediate medium. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to the specific circumstances.

[0026] It should be noted that the terms "first\second\third" involved in the embodiments of the present application are only used to distinguish similar objects, and do not represent a specific order for the objects. It is understandable that the specific order or sequence of "first\second\third" can be interchanged where permitted. It should be understood that the objects distinguished by "first\second\third" can be interchanged where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein.

[0027] The current multifunctional masks have some limitations in terms of filtration efficiency and effectiveness, mainly due to the reliance on filter layer quality testing during the traditional preparation process. These include: limitations of manual quality testing, for example, manual quality testing is a time-consuming and laborious process that cannot meet the needs of large-scale production; manual testing is prone to subjectivity and inconsistency, and lacks effective quality control measures; manual testing cannot accurately evaluate the quality of each filter layer, resulting in possible differences in the quality of the filter layers of different masks.

[0028] The impact of substandard filter layer quality. For example, substandard quality of some filter layers will affect the overall filtration efficiency of the mask, resulting in unstable filtration effect; poor quality of the filter layer will reduce the protective performance of the mask and will not be able to effectively prevent the spread of viruses and bacteria.

[0029] In order to overcome these limitations, advanced mask manufacturing technologies are needed, such as using automated equipment to test the quality of the filter layer, improving production efficiency and quality control. Real-time monitoring of the filter layer production process ensures that each filter layer meets quality standards. By adopting these advanced technologies, the filtration efficiency and effectiveness of multifunctional masks can be improved, providing more reliable protection for individuals and the people around them.

[0030] In one embodiment of the present application, a multifunctional anti-smoke and anti-harmful gas mask and a preparation method thereof are provided, which relates to a safety device that can effectively filter harmful particles and gases in the air. The mask design adopts a multi-layer filtration system, including an activated carbon layer, a HEPA high-efficiency filtration layer and an antibacterial layer, which can effectively block smoke, dust, bacteria and other harmful substances. The preparation method includes selecting a suitable filter material to ensure that it has efficient filtration performance and good air permeability. In addition, the design of the mask also takes into account the comfort of the wearer, and adopts an ergonomic design to adapt to users with different face shapes. Through this design, a multifunctional protective mask that is both safe and comfortable can be provided.

[0031] Figure 1 Schematic diagram of a multifunctional mask according to an embodiment of the present application. Figure 1 As shown, the multifunctional mask is characterized in that it includes: a planar mask body 1, wherein the planar mask body 1 includes a planar mask body layer 11, a multi-layer filtering structure 12 and a nose bridge strip 13, the nose bridge strip 13 is arranged in the middle position of the upper part of the planar mask body layer 11, and the multi-layer filtering structure 12 includes an activated carbon layer 121, a HEPA high-efficiency filter layer 122 and an antibacterial layer 123 stacked on each other; and a first ear band 14 and a second ear band 15 arranged on the opposite left and right sides of the planar mask body 1.

[0032] Figure 2Flow chart of the method for preparing a multifunctional mask according to an embodiment of the present application. Figure 2 As shown, the preparation method of the multifunctional mask includes: 110, providing an activated carbon layer, a HEPA high-efficiency filter layer and an antibacterial layer; 120, performing quality inspection on the activated carbon layer, the HEPA high-efficiency filter layer and the antibacterial layer based on machine vision to determine whether the surface quality of the activated carbon layer, the HEPA high-efficiency filter layer and the antibacterial layer is qualified; 130, after determining that the surface quality of the activated carbon layer, the HEPA high-efficiency filter layer and the antibacterial layer is qualified, the activated carbon layer, the HEPA high-efficiency filter layer and the antibacterial layer are stacked by a stacking process to obtain the multi-layer filter structure; 140, installing the multi-layer filter structure on the planar mask layer; 150, installing the nose bridge strip on the planar mask layer; and, 160, attaching the first ear strap and the second ear strap to the left and right sides of the planar mask layer.

[0033] It should be understood that the activated carbon layer, the HEPA high-efficiency filter layer and the antibacterial layer are important components of an efficient mask. The activated carbon layer can absorb harmful gases and odors, the HEPA high-efficiency filter layer can effectively filter tiny particles, and the antibacterial layer helps to inhibit bacterial reproduction and improve the hygienic performance of the mask. Therefore, in order to ensure that the surface quality of the mask filter layer meets the standards, thereby improving the filtration efficiency and performance of the mask, it is necessary to use machine vision technology to automatically perform quality inspections on the activated carbon layer, the HEPA high-efficiency filter layer and the antibacterial layer to ensure that the surface quality of the activated carbon layer, the HEPA high-efficiency filter layer and the antibacterial layer is qualified.

[0034] Figure 3 1 is a flow chart of sub-step 120 of the method for preparing a multifunctional mask according to an embodiment of the present application. Figure 3As shown, 120, based on machine vision, the quality of the activated carbon layer, the HEPA high-efficiency filtration layer and the antibacterial layer is inspected to determine whether the surface quality of the activated carbon layer, the HEPA high-efficiency filtration layer and the antibacterial layer is qualified, including: 121, obtaining a surface state image of the activated carbon layer collected by a camera; 122, performing multi-scale feature extraction on the surface state image of the activated carbon layer through an image multi-scale feature extractor based on a deep neural network to obtain a shallow feature map of the surface state of the activated carbon layer and a deep feature map of the surface state of the activated carbon layer; 123, the deep feature map of the surface state of the activated carbon layer is passed through a local feature salient device based on a convolution kernel attention network to obtain a significant deep feature map of the surface state of the activated carbon layer; 124, the significant deep feature map of the surface state of the activated carbon layer and the shallow feature map of the surface state of the activated carbon layer are passed through a cross-depth feature interactive fusion device to obtain a multi-scale feature map of the surface state of the activated carbon layer as the multi-scale feature of the surface state of the activated carbon layer; 125, based on the multi-scale feature of the surface state of the activated carbon layer, determining whether the surface quality of the activated carbon layer is qualified.

[0035] Among them, the image multi-scale feature extractor based on deep neural network is an image multi-scale feature extractor based on pyramid network.

[0036] Specifically, in the technical solution of the present application, first, the surface state image of the activated carbon layer collected by the camera is obtained. Then, a convolutional neural network model with excellent performance in implicit feature extraction of the image is used to extract the features of the surface state image of the activated carbon layer. In particular, considering that the surface of the activated carbon layer not only has shallow features such as color, texture and edge, but also has deep features such as structure and shape, these features may be manifested under different levels of image features. Therefore, in order to be able to perform comprehensive feature extraction and analysis of the activated carbon layer from different levels and angles, so as to more accurately evaluate the surface quality and performance of the activated carbon layer, in the technical solution of the present application, the surface state image of the activated carbon layer is further passed through an image multi-scale feature extractor based on a pyramid network to obtain a shallow feature map of the surface state of the activated carbon layer and a deep feature map of the surface state of the activated carbon layer. It should be understood that the multi-scale feature extractor based on the pyramid network can extract the feature information of the image at different levels and scales, thereby realizing a comprehensive analysis of the surface state of the activated carbon layer. By extracting features at different depths at multiple scales, more and richer characteristic information on the surface state of the activated carbon layer can be obtained, including subtle textures, pore structures, etc. This helps to more accurately describe the surface state of the activated carbon layer, thereby more comprehensively evaluating the quality of the activated carbon layer.

[0037] Furthermore, considering that in the deep feature map of the surface state of the activated carbon layer, there are not only key features related to the quality of the surface state of the activated carbon, but also some irrelevant interfering feature information. That is to say, different positions in the deep feature map of the surface state of the activated carbon layer have different degrees of importance. Therefore, in the technical solution of the present application, the deep feature map of the surface state of the activated carbon layer is further passed through a local feature salient device based on a convolution kernel attention network to obtain a significant deep feature map of the surface state of the activated carbon layer. It should be understood that the convolution kernel attention network can help the model learn and pay attention to the key features related to the quality of the surface state of the activated carbon in the image, so that the key information and features therein are highlighted and emphasized, thereby improving the capture and utilization of important information. Through local feature saliency, the key information related to the quality in the deep features of the surface state of the activated carbon can be enhanced, and the influence of noise and irrelevant information in the deep feature map can be reduced, thereby improving the ability to identify and capture the key features of the surface state of the activated carbon layer, as well as the accuracy of quality assessment. That is, by using the local feature saliency device to process the deep feature map of the surface state of the activated carbon layer, the goal of guiding the network through weight information, focusing more on extracting the convolution kernels of useful features, and ignoring the convolution kernels that learn useless background features or noise, is achieved. In this way, the part related to the surface quality of the activated carbon layer in the deep feature map can be highlighted, and the interference of irrelevant information can be reduced, thereby improving the accuracy and robustness of quality assessment.

[0038] In a specific embodiment of the present application, the deep feature map of the surface state of the activated carbon layer is passed through a local feature saliency device based on a convolution kernel attention network to obtain a salient activated carbon layer surface state deep feature map, including: performing a nonlinear convolution operation on the deep feature map of the surface state of the activated carbon layer through a multi-scale convolution kernel group of the convolution kernel attention network to obtain a first state semantic feature map, a second state semantic feature map, a third state semantic feature map and a fourth state semantic feature map; adding the first state semantic feature map, the second state semantic feature map, the third state semantic feature map and the fourth state semantic feature map by position to obtain a fused feature map; performing global mean pooling processing on each feature matrix along the channel dimension in the fused feature map to obtain a pooled feature vector; The pooled feature vector is activated and regularized by the fully connected layer in the convolution kernel attention network to obtain a fused feature vector; based on the fused feature vector, the weight vectors of the first state semantic feature map, the second state semantic feature map, the third state semantic feature map and the fourth state semantic feature map are calculated to obtain a first weight vector, a second weight vector, a third weight vector and a fourth weight vector; the first weight vector, the second weight vector, the third weight vector and the fourth weight vector are used as channel weighted vectors, and the weighted sum of the first state semantic feature map, the second state semantic feature map, the third state semantic feature map and the fourth state semantic feature map is calculated to obtain the deep feature map of the surface state of the significant activated carbon layer.

[0039] Further, performing global mean pooling processing on each feature matrix along the channel dimension in the fused feature map to obtain a pooled feature vector, including: performing global mean pooling processing on each feature matrix along the channel dimension in the fused feature map using the following global mean pooling formula to obtain the pooled feature vector; wherein the global mean pooling formula is:

[0040]

[0041] Among them, U c (i, j) is the eigenvalue of the (i, j)th position of each feature matrix along the channel dimension in the fused feature map, H and W are the height and width of the fused feature map, respectively, S c is the feature value of each position in the pooled feature vector.

[0042] It should be understood that, considering that the deep feature map of the surface state of the significant activated carbon layer and the shallow feature map of the surface state of the activated carbon layer respectively represent the deep structural features and shallow features of the surface state of the activated carbon layer in the surface state image of the activated carbon layer, and the features of different levels and scales are of great significance for the surface quality identification and detection of the activated carbon layer. Therefore, in order to combine the advantages of deep features and shallow features, realize the information interaction and fusion between features of different depths, so as to more comprehensively and accurately characterize the surface state of the activated carbon layer and improve the effect of quality assessment, in the technical solution of the present application, the deep feature map of the surface state of the significant activated carbon layer and the shallow feature map of the surface state of the activated carbon layer are further combined through a cross-depth feature interaction fuser to obtain a multi-scale feature map of the surface state of the activated carbon layer. Specifically, after the deep feature map of the surface state of the significant activated carbon layer is globally averaged pooled in the channel dimension, the two-dimensional feature matrix will be compressed into a real number, which has the global receptive field of each feature in the high-level feature map. This value is then used as attention information to multiply the different features of the shallow feature map of the surface state of the activated carbon layer, guiding the position information in the low-level shallow feature map to restore the semantics of the surface state of the activated carbon layer, thereby better capturing the diverse characteristics of the surface state of the activated carbon layer. Finally, the generated low-level feature map and high-level feature map are added and fused to obtain a multi-scale feature map of the surface state of the activated carbon layer with richer semantic information. In this way, the information interaction and complementarity between deep features and shallow features can be promoted, so that the correlation between features of different depths can be strengthened. This helps to improve the representation ability of features and the accuracy of quality assessment.

[0043] In a specific embodiment of the present application, the significant activated carbon layer surface state deep feature map and the activated carbon layer surface state shallow feature map are subjected to a cross-depth feature interactive fusion device to obtain an activated carbon layer surface state multi-scale feature map as the activated carbon layer surface state multi-scale feature, including: performing global mean pooling processing on each feature matrix along the channel dimension in the significant activated carbon layer surface state deep feature map to obtain an activated carbon layer surface state deep pooling feature vector; passing the activated carbon layer surface state deep pooling feature vector through a fully connected layer to perform fully connected encoding on the activated carbon layer surface state deep pooling feature vector to obtain an activated carbon layer surface state deep semantic fully connected feature vector; performing weighted fusion based on the channel dimension on the activated carbon layer surface state deep semantic fully connected feature vector and the activated carbon layer surface state shallow feature map to obtain a fused feature map; fusing the fused feature map with the significant activated carbon layer surface state deep feature map to obtain the activated carbon layer surface state multi-scale feature map.

[0044] Then, the multi-scale feature map of the surface state of the activated carbon layer is passed through a classifier-based quality detector to obtain a quality detection result, which is used to indicate whether the surface quality of the activated carbon layer is qualified. In other words, the surface quality of the activated carbon layer is detected by using the multi-level semantic fusion features of the surface state of the activated carbon layer for classification processing to ensure that the surface quality of the activated carbon layer is qualified. Similarly, the surface quality of the HEPA high-efficiency filter layer and the antibacterial layer can be detected in a similar manner to ensure that the surface quality of the mask filter layer meets the standards.

[0045] In a specific embodiment of the present application, whether the surface quality of the activated carbon layer is qualified is determined based on the multi-scale characteristics of the surface state of the activated carbon layer, including: passing the multi-scale characteristic map of the surface state of the activated carbon layer through a classifier-based quality detector to obtain a quality detection result, and the quality detection result is used to indicate whether the surface quality of the activated carbon layer is qualified.

[0046] In one embodiment of the present application, the preparation method of the multifunctional mask also includes a training step: for training the pyramid network-based image multi-scale feature extractor, the convolution kernel attention network-based local feature salient device, the cross-depth feature interaction fusion device and the classifier-based quality detector; wherein the training step includes: obtaining a training surface state image of the activated carbon layer collected by a camera; performing multi-scale feature extraction on the training surface state image of the activated carbon layer through the deep neural network-based image multi-scale feature extractor to obtain a shallow feature map of the training activated carbon layer surface state and a deep feature map of the training activated carbon layer surface state; passing the deep feature map of the training activated carbon layer surface state through the local feature salient device based on the convolution kernel attention network to obtain a training salient activated carbon layer surface state map. deep feature map; passing the training salient activated carbon layer surface state deep feature map and the training activated carbon layer surface state shallow feature map through the cross-depth feature interaction fuser to obtain a training activated carbon layer surface state multi-scale feature map; optimizing the training activated carbon layer surface state multi-scale feature map with the feature matrix distribution of the training activated carbon layer surface state multi-scale feature map to obtain an optimized training activated carbon layer surface state multi-scale feature map; passing the optimized training activated carbon layer surface state multi-scale feature map through the classifier-based quality detector to obtain a classification loss function value; based on the classification loss function value, training the pyramid network-based image multi-scale feature extractor, the convolution kernel attention network-based local feature salient device, the cross-depth feature interaction fuser and the classifier-based quality detector.

[0047] In the above technical solution, the shallow feature map of the surface state of the training activated carbon layer and the deep feature map of the surface state of the training activated carbon layer respectively represent the image semantic features of the surface state image of the training activated carbon layer based on different convolutional coding depths. After the deep feature map of the surface state of the training activated carbon layer passes through the local feature salient device based on the convolution kernel attention network, the local spatial distribution of the features in the feature matrix of the deep feature map of the surface state of the training activated carbon layer will be further enhanced. However, considering that the shallow feature map of the surface state of the training activated carbon layer and the deep feature map of the surface state of the training activated carbon layer themselves have differences in feature order and resolution, the local feature saliency device based on the convolution kernel attention network also introduces significant differences in the local spatial distribution of features when processing the deep feature map of the surface state of the training activated carbon layer, which will cause more significant feature distribution differences between the corresponding feature matrices of the deep feature map of the surface state of the training significant activated carbon layer and the shallow feature map of the surface state of the training activated carbon layer, thereby affecting the class fusion effect based on the eigenvalue numerical distribution of each feature matrix of the multi-scale feature map of the surface state of the training activated carbon layer obtained by the cross-depth feature interactive fuser, and affecting the accuracy of the classification results of the multi-scale feature map of the surface state of the training activated carbon layer obtained by the classifier-based quality detector.

[0048] Therefore, the applicant of the present application optimizes the multi-scale characteristic map of the surface state of the training activated carbon layer based on the characteristic matrix distribution of the multi-scale characteristic map of the surface state of the training activated carbon layer when the multi-scale characteristic map of the surface state of the training activated carbon layer is classified and regressed by a mass detector based on a classifier, which is expressed as: the coefficients of the multi-scale characteristic map of the surface state of the training activated carbon layer are calculated by the following optimization formula; wherein the optimization formula is:

[0049]

[0050] Where m i,j is the eigenvalue of the (i, j)th position of each feature matrix of the multi-scale feature map of the surface state of the activated carbon layer, φ(m i,j ) represents the probabilistic function of the eigenvalue, that is, the eigenvalue m i,j A probabilistic function mapped to the interval [0,1], S is the scale of each feature matrix of the training multi-scale feature map of the surface state of the activated carbon layer, that is, the width multiplied by the height, p is the class probability value obtained by the classifier-based quality detector of the training multi-scale feature map of the surface state of the activated carbon layer, and α is a weight hyperparameter, ω is a coefficient; the corresponding feature matrix of the training multi-scale feature map of the surface state of the activated carbon layer is weightedly optimized with the coefficient ω to obtain the optimized training multi-scale feature map of the surface state of the activated carbon layer.

[0051] That is, for the feature scene corresponding to each feature matrix of the multi-scale feature map of the surface state of the training activated carbon layer, the class probability reasoning logic association of the scene saturation is undertaken through the probability distribution prospect constraint and the relative probability mapping response hypothesis, so as to give the scene concept ontology cognition to the feature set of each feature matrix of the multi-scale feature map of the surface state of the training activated carbon layer, that is, the overall distribution is internally aligned with the class probability logic reasoning based on the scene under the classification process, so as to improve the understanding ability of the feature matrix scene distribution of the multi-scale feature map of the surface state of the training activated carbon layer for class cognition. In this way, the corresponding feature matrix of the multi-scale feature map of the surface state of the training activated carbon layer is weighted optimized by the coefficient ω, so as to improve the accuracy of the classification result obtained by the classifier-based quality detector of the optimized multi-scale feature map of the surface state of the training activated carbon layer. In this way, machine vision technology can be used to intelligently detect the surface quality of the carbon layer of different filter layers of the multi-functional mask, so as to ensure that the surface quality of the filter layer of the mask meets the standard and improve the filtering efficiency and performance of the mask.

[0052] In summary, the intelligent control method of the lighting driving power supply based on the embodiment of the present application is explained. In order to ensure that the surface quality of the mask filter layer meets the standard, thereby improving the filtration efficiency and performance of the mask, it is necessary to use machine vision technology to automatically perform quality inspection on the activated carbon layer, the HEPA high-efficiency filtration layer and the antibacterial layer to ensure that the surface quality of the activated carbon layer, the HEPA high-efficiency filtration layer and the antibacterial layer is qualified.

[0053] In one embodiment of the present application, Figure 4 Schematic diagram of a system for preparing a multifunctional mask according to an embodiment of the present application. Figure 4 As shown, according to the preparation system 200 of the multifunctional mask of the embodiment of the present application, it includes: each layer providing module 210, which is used to provide an activated carbon layer, a HEPA high-efficiency filter layer and an antibacterial layer; a quality detection module 220, which is used to perform quality detection on the activated carbon layer, the HEPA high-efficiency filter layer and the antibacterial layer based on machine vision to determine whether the surface quality of the activated carbon layer, the HEPA high-efficiency filter layer and the antibacterial layer is qualified; a stacking processing module 230, which is used to stack the activated carbon layer, the HEPA high-efficiency filter layer and the antibacterial layer through a stacking process to obtain the multi-layer filter structure after determining that the surface quality of the activated carbon layer, the HEPA high-efficiency filter layer and the antibacterial layer is qualified; a multi-layer filter structure installation module 240, which is used to install the multi-layer filter structure on the planar mask layer; a nose bridge strip installation module 250, which is used to install the nose bridge strip on the planar mask layer; and an ear band attachment module 260, which is used to attach the first ear band and the second ear band to the left and right sides of the planar mask layer.

[0054] Here, those skilled in the art will appreciate that the specific functions and operations of the various units and modules in the preparation system of the multifunctional mask have been described above with reference to Figures 1 to 3 The preparation method of the multifunctional mask has been described in detail, and therefore, its repeated description will be omitted.

[0055] As described above, the system 200 for preparing a multifunctional mask according to an embodiment of the present application can be implemented in various terminal devices, such as a server for preparing a multifunctional mask. In one example, the system 200 for preparing a multifunctional mask according to an embodiment of the present application can be integrated into a terminal device as a software module and / or a hardware module. For example, the system 200 for preparing a multifunctional mask can be a software module in the operating system of the terminal device, or can be an application developed for the terminal device; of course, the system 200 for preparing a multifunctional mask can also be one of the many hardware modules of the terminal device.

[0056] Alternatively, in another example, the multifunctional mask preparation system 200 and the terminal device may also be separate devices, and the multifunctional mask preparation system 200 may be connected to the terminal device via a wired and / or wireless network, and transmit interactive information in accordance with an agreed data format.

[0057] Figure 5 Schematic diagram of a method for preparing a multifunctional mask according to an embodiment of the present application. Figure 5 As shown, in this application scenario, first, a surface state image of the activated carbon layer captured by a camera is obtained (for example, Figure 5 Then, the acquired surface state image is input to a server (for example, Figure 5 In S) shown in , the server is capable of processing the surface state image based on the preparation algorithm of the multifunctional mask to determine whether the surface quality of the activated carbon layer is qualified.

[0058] It should also be noted that in the apparatus, device 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.

[0059] 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 disclosed herein.

[0060] 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 preparing a multifunctional mask, characterized in that: The multifunctional mask comprises: A planar mask body, wherein the planar mask body comprises a planar mask layer, a multi-layer filter structure and a nose bridge strip, wherein the nose bridge strip is arranged at the middle position of the upper part of the planar mask layer, and the multi-layer filter structure comprises an activated carbon layer, a HEPA high-efficiency filter layer and an antibacterial layer stacked on each other; and A first ear strap and a second ear strap are arranged on the opposite left and right sides of the flat mask body; The preparation method of the multifunctional mask comprises: Provide activated carbon layer, HEPA high-efficiency filtration layer and antibacterial layer; Performing quality inspection on the activated carbon layer, the HEPA high-efficiency filtration layer and the antibacterial layer based on machine vision to determine whether the surface quality of the activated carbon layer, the HEPA high-efficiency filtration layer and the antibacterial layer is qualified; After determining that the surface qualities of the activated carbon layer, the HEPA high-efficiency filter layer and the antibacterial layer are qualified, the activated carbon layer, the HEPA high-efficiency filter layer and the antibacterial layer are stacked by a stacking process to obtain the multi-layer filter structure; Installing the multi-layer filtering structure on the planar cover layer; Mounting a nose bridge strip on the planar mask layer; and Attaching the first ear strap and the second ear strap to the left and right sides of the planar mask layer; The activated carbon layer, the HEPA high-efficiency filtration layer, and the antibacterial layer are subjected to quality inspection based on machine vision to determine whether the surface quality of the activated carbon layer, the HEPA high-efficiency filtration layer, and the antibacterial layer is qualified, including: Acquire a surface state image of the activated carbon layer collected by a camera; Performing multi-scale feature extraction on the surface state image of the activated carbon layer by using a multi-scale feature extractor for an image based on a deep neural network to obtain a shallow feature map of the surface state of the activated carbon layer and a deep feature map of the surface state of the activated carbon layer; The deep feature map of the surface state of the activated carbon layer is passed through a local feature salient device based on a convolution kernel attention network to obtain a salient activated carbon layer surface state deep feature map; The significant activated carbon layer surface state deep feature map and the activated carbon layer surface state shallow feature map are subjected to a cross-depth feature interactive fusion device to obtain an activated carbon layer surface state multi-scale feature map as an activated carbon layer surface state multi-scale feature; Determining whether the surface quality of the activated carbon layer is qualified based on the multi-scale characteristics of the surface state of the activated carbon layer; The image multi-scale feature extractor based on deep neural network is an image multi-scale feature extractor based on pyramid network; The deep feature map of the surface state of the activated carbon layer is passed through a local feature salient device based on a convolution kernel attention network to obtain a salient activated carbon layer surface state deep feature map, including: The deep feature map of the surface state of the activated carbon layer is subjected to a nonlinear convolution operation through the multi-scale convolution kernel group of the convolution kernel attention network to obtain a first state semantic feature map, a second state semantic feature map, a third state semantic feature map and a fourth state semantic feature map; Adding the first state semantic feature map, the second state semantic feature map, the third state semantic feature map and the fourth state semantic feature map by position to obtain a fused feature map; Performing global mean pooling processing on each feature matrix along the channel dimension in the fused feature map to obtain a pooled feature vector; Passing the pooled feature vector through the fully connected layer in the convolution kernel attention network to activate and regularize the pooled feature vector to obtain a fused feature vector; Based on the fused feature vector, calculating weight vectors of the first state semantic feature map, the second state semantic feature map, the third state semantic feature map, and the fourth state semantic feature map to obtain a first weight vector, a second weight vector, a third weight vector, and a fourth weight vector; Taking the first weight vector, the second weight vector, the third weight vector and the fourth weight vector as channel weight vectors, the weighted sum of the first state semantic feature map, the second state semantic feature map, the third state semantic feature map and the fourth state semantic feature map is calculated to obtain the deep feature map of the surface state of the significant activated carbon layer.

2. The method for preparing the multifunctional mask according to claim 1, characterized in that: Performing global mean pooling processing on each feature matrix along the channel dimension in the fused feature map to obtain a pooled feature vector, including: performing global mean pooling processing on each feature matrix along the channel dimension in the fused feature map according to the following global mean pooling formula to obtain the pooled feature vector; Among them, the global mean pooling formula is: Among them, U c (i, j) is the eigenvalue of the (i, j)th position of each feature matrix along the channel dimension in the fused feature map, H and W are the height and width of the fused feature map, respectively, S c is the eigenvalue of each position in the pooled feature vector.

3. The method for preparing the multifunctional mask according to claim 2, characterized in that: The significant activated carbon layer surface state deep feature map and the activated carbon layer surface state shallow feature map are subjected to a cross-depth feature interactive fusion device to obtain an activated carbon layer surface state multi-scale feature map as an activated carbon layer surface state multi-scale feature, including: Performing global mean pooling processing on each feature matrix along the channel dimension in the significant activated carbon layer surface state deep feature map to obtain an activated carbon layer surface state deep pooling feature vector; Passing the deep pooling feature vector of the surface state of the activated carbon layer through a fully connected layer to perform full connection encoding on the deep pooling feature vector of the surface state of the activated carbon layer to obtain a deep semantic fully connected feature vector of the surface state of the activated carbon layer; Performing a weighted fusion based on the channel dimension on the deep semantic fully connected feature vector of the surface state of the activated carbon layer and the shallow feature map of the surface state of the activated carbon layer to obtain a fused feature map; The fused feature map and the significant activated carbon layer surface state deep feature map are fused to obtain the activated carbon layer surface state multi-scale feature map.

4. The method for preparing the multifunctional mask according to claim 3, characterized in that: Based on the multi-scale characteristics of the surface state of the activated carbon layer, determining whether the surface quality of the activated carbon layer is qualified includes: passing the multi-scale characteristic map of the surface state of the activated carbon layer through a quality detector based on a classifier to obtain a quality detection result, and the quality detection result is used to indicate whether the surface quality of the activated carbon layer is qualified.

5. The method for preparing the multifunctional mask according to claim 4, characterized in that: The method further includes a training step for training the pyramid network-based image multi-scale feature extractor, the convolution kernel attention network-based local feature salient device, the cross-depth feature interaction fusion device, and the classifier-based quality detector; Wherein, the training step includes: Acquire a training surface state image of the activated carbon layer collected by a camera; Performing multi-scale feature extraction on the training surface state image of the activated carbon layer by the image multi-scale feature extractor based on the deep neural network to obtain a shallow feature map of the surface state of the training activated carbon layer and a deep feature map of the surface state of the training activated carbon layer; Passing the training activated carbon layer surface state deep feature map through the local feature salient device based on the convolution kernel attention network to obtain a training salient activated carbon layer surface state deep feature map; The deep feature map of the surface state of the training significant activated carbon layer and the shallow feature map of the surface state of the training activated carbon layer are passed through the cross-depth feature interactive fusion device to obtain a multi-scale feature map of the surface state of the training activated carbon layer; Optimizing the training activated carbon layer surface state multi-scale feature map with the feature matrix distribution of the training activated carbon layer surface state multi-scale feature map to obtain an optimized training activated carbon layer surface state multi-scale feature map; Passing the optimized trained activated carbon layer surface state multi-scale feature map through the classifier-based quality detector to obtain a classification loss function value; The pyramid network-based image multi-scale feature extractor, the convolution kernel attention network-based local feature salient device, the cross-depth feature interaction fusion device and the classifier-based quality detector are trained based on the classification loss function value.

Images