Intelligent conveying system and method for intelligent cosmetic packaging
Through deep learning computer vision technology, the surface state of cosmetic packaging hose is analyzed, and intelligent power control instructions for dust collection devices are generated, which solves the problem of unstable power adjustment of dust collection devices in the prior art, and achieves the optimization of cleaning effect and energy efficiency .
Patent Information
- Application Number
- CN202411143202.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-20
- Publication Date
- 2024-11-26
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
In the existing cosmetic packaging hose cleaning device, the power adjustment of the dust collection mechanism depends on manual experience, resulting in unstable cleaning effect and waste of energy, and lack of intelligent dynamic adjustment mechanism.
Using computer vision technology based on deep learning, the camera collects the front and back comparison and analysis of the surface state image of cosmetic packaging hose through the camera, digs out the surface state characteristics before and after dust removal and cleaning, and generates power control instructions for the dust collection mechanism to achieve intelligent control and dynamic optimization.
It realizes intelligent power adjustment of the dust collection mechanism, optimizes the cleaning effect and energy efficiency ratio, and ensures the stability of the cleaning effect and energy utilization efficiency.
Smart Images

Figure BDA0005001504340000081 
Figure BDA0005001504340000101 
Figure BDA0005001504340000102
Abstract
Description
Technical Field
[0001] The present application relates to the field of intelligent transportation, and specifically to an intelligent transportation system and method for intelligent packaging of cosmetics. Background Art
[0002] In the cosmetics industry, the quality of product packaging is directly related to the brand image and market acceptance of the product. In particular, the surface cleanliness of cosmetic hoses, as one of the common packaging forms, not only affects the appearance of the product, but also may affect the hygiene and safety of the product and the consumer's experience. If the product packaging contains impurities such as dust, lint, and hair, it will not only reduce the appearance quality of the product, but may also cause dissatisfaction and complaints from consumers.
[0003] In this regard, the invention patent with publication number CN111634610A discloses a negative pressure automatic conveying device for the production of cosmetic packaging hoses, which achieves the cleaning treatment of the surface of the cosmetic packaging hose by setting a dust collection mechanism in the conveying device. More specifically, during the operation of the dust collection mechanism, its operating power is one of the important factors affecting the cleaning effect. Usually, in order to improve energy efficiency while ensuring the cleaning effect, it is necessary to adjust the power of the dust collection mechanism. However, in the existing technology, the power adjustment of the dust collection mechanism often relies on manual experience and judgment, which is not only inefficient, but also prone to errors, making it difficult to ensure the stability of the cleaning effect, resulting in poor cleaning effect or energy waste.
[0004] Therefore, an optimized intelligent conveying method for intelligent packaging of cosmetics is expected. Summary of the invention
[0005] This application is made in consideration of the above problems. One object of this application is to provide an intelligent conveying system and method for intelligent packaging of cosmetics.
[0006] The embodiment of the present application provides an intelligent conveying system for intelligent packaging of cosmetics, including a conveyor frame, one end of the conveyor frame is movably mounted with an active rotating shaft, the other end of the conveyor frame is movably mounted with a driven rotating shaft, the bottom of the conveyor frame is fixedly mounted with a supporting leg, the bottom of the supporting leg is fixedly mounted with a supporting foot pad, the bottom of the supporting foot pad is fixedly mounted with an anti-slip rubber pad, a negative pressure mechanism is provided at the outer middle position of the conveyor frame, the negative pressure mechanism includes a negative pressure box, a blower, a dust collection mechanism, an exhaust branch pipe and an air hood, the dust collection mechanism includes a dust collection cylinder, a dust filter hood, an exhaust main pipe, an air duct, a bag filter and an air filter, a cooling mechanism is provided at the other side of the upper part of the negative pressure box, the cooling mechanism includes a filter box, a first cooling air duct, A second cooling air duct, a first supporting and fixing rod, a second supporting and fixing rod and a filter net, hose flow grooves are provided on both side surfaces of the negative pressure box, and cameras are provided above both side areas of the negative pressure box, wherein the front area of the negative pressure box is the first area, and the rear area of the negative pressure box is the second area, wherein the camera is used to collect a first surface state image of the distribution of the cluster of cosmetic packaging hoses in the first area, and a second surface state image of the distribution of the cluster of cosmetic packaging hoses in the second area, and transmit the first surface state image and the second surface state image to the dust collection controller for image comparison and analysis to generate a power control instruction for the dust collection mechanism, wherein the dust collection controller is deployed above the negative pressure box and is electrically connected to the dust collection mechanism.
[0007] For example, according to the intelligent conveying system for intelligent packaging of cosmetics in an embodiment of the present application, the dust collection controller includes:
[0008] An image feature extraction module, used to extract surface state features from the first surface state image and the second surface state image respectively to obtain a first cosmetic packaging hose cluster distribution surface state feature map and a second cosmetic packaging hose cluster distribution surface state feature map;
[0009] A feature autocorrelation enhancement module is used to perform feature autocorrelation enhancement on the first cosmetic packaging hose cluster distribution surface state feature map and the second cosmetic packaging hose cluster distribution surface state feature map to obtain the first cosmetic packaging hose cluster distribution surface enhancement state feature map and the second cosmetic packaging hose cluster distribution surface enhancement state feature map;
[0010] The dust collection effect analysis module is used to generate the power control instruction based on the characteristic difference between the first cosmetic packaging hose cluster distribution surface enhancement state characteristic diagram and the second cosmetic packaging hose cluster distribution surface enhancement state characteristic diagram.
[0011] For example, according to the intelligent conveying system for intelligent packaging of cosmetics in an embodiment of the present application, the image feature extraction module is used to:
[0012] The first surface state image and the second surface state image are respectively passed through a surface state feature extractor based on a hole convolutional neural network model to obtain the first cosmetic packaging hose cluster distribution surface state feature map and the second cosmetic packaging hose cluster distribution surface state feature map.
[0013] For example, according to the intelligent conveying system for intelligent packaging of cosmetics in an embodiment of the present application, the feature autocorrelation enhancement module is used to:
[0014] The first cosmetic packaging hose cluster distribution surface state characteristic map and the second cosmetic packaging hose cluster distribution surface state characteristic map are respectively input into the feature space structure maintaining self-attention enhancement module based on cross-channel intersection to obtain the first cosmetic packaging hose cluster distribution surface enhancement state characteristic map and the second cosmetic packaging hose cluster distribution surface enhancement state characteristic map.
[0015] For example, according to the intelligent conveying system for intelligent packaging of cosmetics in an embodiment of the present application, the feature autocorrelation enhancement module includes:
[0016] A layer normalization unit, used for performing layer normalization on the first cosmetic packaging hose cluster distribution surface state characteristic map to obtain a normalized first cosmetic packaging hose cluster distribution surface state characteristic map;
[0017] A channel context-related feature extraction unit, used for performing point convolution processing on the normalized first cosmetic packaging hose cluster distribution surface state feature map to obtain a packaging hose cluster distribution surface state channel context-related representation feature map;
[0018] A spatial context-related feature extraction unit, used for performing convolution encoding on the channel context-related representation feature map of the distribution surface state of the packaging hose cluster to obtain a spatial context-related representation feature map of the distribution surface state of the packaging hose cluster;
[0019] A global interactive attention fusion unit is used to perform channel-space global interactive attention fusion on the channel context association representation feature map of the packaging hose cluster distribution surface state and the spatial context association representation feature map of the packaging hose cluster distribution surface state to obtain the first cosmetic packaging hose cluster distribution surface reinforcement state feature map.
[0020] For example, according to the intelligent conveying system for intelligent packaging of cosmetics in an embodiment of the present application, the global interactive attention fusion unit is used to:
[0021] Copying the packaging hose cluster distribution surface state space context association representation feature map to obtain a backup packaging hose cluster distribution surface state space context association representation feature map;
[0022] Reshape the packaging hose cluster distribution surface state channel context association representation feature map, the packaging hose cluster distribution surface state space context association representation feature map and the backup packaging hose cluster distribution surface state space context association representation feature map to obtain a packaging hose cluster distribution surface state channel context association representation feature matrix, a packaging hose cluster distribution surface state space context association representation feature matrix and a backup packaging hose cluster distribution surface state space context association representation feature matrix;
[0023] Calculate the cross-channel cross covariance matrix between the channel context association representation feature matrix of the packaging hose cluster distribution surface state and the spatial context association representation feature matrix of the packaging hose cluster distribution surface state;
[0024] Use the Softmax function to activate the cross-channel cross-covariance matrix to obtain a global interactive attention matrix of the surface state features of the packaging hose cluster distribution;
[0025] Calculate the product of the backup packaging hose cluster distribution surface state space context association representation feature matrix and the packaging hose cluster distribution surface state feature global interactive attention matrix to obtain an attention-enhanced packaging hose cluster distribution surface state feature representation matrix;
[0026] The feature shape of the attention-enhanced packaging hose cluster distribution surface state feature representation matrix is reshaped to obtain the first cosmetic packaging hose cluster distribution surface enhancement state feature map.
[0027] For example, according to the intelligent conveying system for intelligent packaging of cosmetics in an embodiment of the present application, the dust collection effect analysis module includes:
[0028] A difference measurement unit, used for extracting difference features between the first cosmetic packaging hose cluster distribution surface enhancement state feature map and the second cosmetic packaging hose cluster distribution surface enhancement state feature map to obtain a surface state fine-grained differential feature map;
[0029] A control instruction generating unit is used to input the surface state fine-grained differential feature map into a classifier-based adaptive control module to obtain the power control instruction, wherein the power control instruction is used to indicate whether the power of the dust collection mechanism should be increased, decreased or kept unchanged.
[0030] For example, according to the intelligent conveying system for intelligent packaging of cosmetics in an embodiment of the present application, the difference measurement unit is used to:
[0031] The position difference between the first cosmetic packaging hose cluster distribution surface enhancement state characteristic map and the second cosmetic packaging hose cluster distribution surface enhancement state characteristic map is calculated to obtain the surface state fine-grained differential characteristic map.
[0032] The embodiment of the present application also provides an intelligent conveying method for intelligent packaging of cosmetics, which includes:
[0033] Acquire a first surface state image of the distribution of the cosmetic packaging hose clusters in the first area and a second surface state image of the distribution of the cosmetic packaging hose clusters in the second area collected by the camera;
[0034] Performing surface state feature extraction on the first surface state image and the second surface state image respectively to obtain a first cosmetic packaging hose cluster distribution surface state feature map and a second cosmetic packaging hose cluster distribution surface state feature map;
[0035] Performing feature autocorrelation enhancement on the first cosmetic packaging hose cluster distribution surface state characteristic map and the second cosmetic packaging hose cluster distribution surface state characteristic map to obtain the first cosmetic packaging hose cluster distribution surface enhanced state characteristic map and the second cosmetic packaging hose cluster distribution surface enhanced state characteristic map;
[0036] The power control instruction is generated based on the characteristic difference between the first cosmetic packaging hose cluster distribution surface enhancement state characteristic diagram and the second cosmetic packaging hose cluster distribution surface enhancement state characteristic diagram.
[0037] For example, according to the intelligent conveying method for intelligent packaging of cosmetics in an embodiment of the present application, surface state feature extraction is performed on the first surface state image and the second surface state image respectively to obtain a first cosmetic packaging hose cluster distribution surface state feature map and a second cosmetic packaging hose cluster distribution surface state feature map, including:
[0038] The first surface state image and the second surface state image are respectively passed through a surface state feature extractor based on a hole convolutional neural network model to obtain the first cosmetic packaging hose cluster distribution surface state feature map and the second cosmetic packaging hose cluster distribution surface state feature map.
[0039] According to the intelligent conveying system and method for intelligent packaging of cosmetics in the embodiment of the present application, it uses computer vision technology based on deep learning to compare and analyze the surface state images of the cosmetic packaging hose before and after dust removal through the negative pressure box, respectively mining the surface state characteristics of the cluster distribution of the cosmetic packaging hose before and after dust removal cleaning, and then characterizing the dust removal effect of the packaging hose based on the characteristic difference between the two, and guiding the power adjustment of the dust collection mechanism based on this. In this way, the intelligent control of the dust collection mechanism can be realized, and its operating power can be dynamically optimized to achieve the optimization of its cleaning effect and energy efficiency ratio. BRIEF DESCRIPTION OF THE DRAWINGS
[0040] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings of the embodiments of the present application are briefly introduced below. Obviously, the drawings described below only relate to some embodiments of the present application, and are not intended to limit the present application.
[0041] Figure 1 A schematic diagram of the structure of a dust collection controller of an intelligent conveying system for intelligent packaging of cosmetics in an embodiment of the present application is shown;
[0042] Figure 2 It shows a structural schematic diagram of a characteristic autocorrelation enhancement module of an intelligent conveying system for intelligent packaging of cosmetics in an embodiment of the present application;
[0043] Figure 3 A schematic diagram showing the structure of a dust collection effect analysis module of an intelligent conveying system for intelligent packaging of cosmetics in an embodiment of the present application is shown;
[0044] Figure 4 A flow chart showing the intelligent conveying method for intelligent packaging of cosmetics in an embodiment of the present application is shown; and
[0045] Figure 5 The application scenario diagram of the intelligent conveying method for intelligent packaging of cosmetics in the embodiment of the present application is shown. DETAILED DESCRIPTION
[0046] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work also fall within the scope of protection of the present application.
[0047] The terms used in this specification are those common terms currently widely used in the art in consideration of the functions of the present application, but these terms may vary according to the intentions of those skilled in the art, precedents, or new technologies in the art. In addition, specific terms may be selected, and in this case, their detailed meanings will be described in the detailed description of the present application. Therefore, the terms used in the specification should not be understood as simple names, but rather as a general description based on the meaning of the terms and the present application.
[0048] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.
[0049] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously as required. At the same time, other operations may also be added to these processes, or a certain step or several steps of operations may be removed from these processes.
[0050] In addition, the application architecture diagram in the embodiment of the present application is intended to more clearly illustrate the technical solution in the embodiment of the present application, and does not constitute a limitation on the technical solution provided in the embodiment of the present application. Of course, for other application architectures and business applications, the technical solution provided in the embodiment of the present application is also applicable to similar problems.
[0051] The following is a non-restrictive description of the intelligent conveying system for intelligent packaging of cosmetics provided according to at least one embodiment of the present application through several examples or embodiments. As described below, different features in these specific examples or embodiments can be combined with each other without conflicting with each other to obtain new examples or embodiments, and these new examples or embodiments also fall within the scope of protection of the present application.
[0052] The invention patent with publication number CN111634610A discloses a negative pressure automatic conveying device for the production of cosmetic packaging hoses, which includes a conveyor frame, one end of the conveyor frame is movably mounted with an active rotating shaft, the other end of the conveyor frame is movably mounted with a driven rotating shaft, the bottom of the conveyor frame is fixedly mounted with a supporting leg, the bottom of the supporting leg is fixedly mounted with a supporting foot pad, the bottom of the supporting foot pad is fixedly mounted with an anti-slip rubber pad, a negative pressure mechanism is provided at the outer middle position of the conveyor frame, the negative pressure mechanism includes a negative pressure box, a blower, a dust collection mechanism, an exhaust branch pipe and an air hood, the dust collection mechanism includes a dust collection cylinder, a dust filter hood, an exhaust main pipe, an air guide pipe, a bag filter and an air filter, a cooling mechanism is provided at the other side of the upper part of the negative pressure box, the cooling mechanism includes a filter box, a first cooling air duct, a second cooling air duct, a first support fixing rod, a second support fixing rod and a filter, and hose flow grooves are provided on both side surfaces of the negative pressure box.
[0053] In the above scheme, by setting up a dust collection mechanism, the surface of the cosmetic packaging hose is cleaned, thereby effectively improving the cleanliness and sanitary safety of the packaging. However, during the operation of the dust collection mechanism of the device, its operating power is one of the important factors affecting the cleaning effect. If the power is too low, the surface impurities cannot be effectively removed and the expected cleaning effect cannot be achieved. If the power is too high, it may lead to energy waste and may cause unnecessary wear on the hose surface. In the existing technology, the power adjustment of the dust collection mechanism mainly relies on manual experience, lacks an intelligent dynamic adjustment mechanism, and cannot optimize the power in real time according to the actual cleaning effect. It is difficult to ensure the stability of the cleaning effect, which easily leads to poor cleaning effect or energy waste.
[0054] In response to the above technical problems, the technical concept of this application is to use computer vision technology based on deep learning to compare and analyze the surface state images of cosmetic packaging hoses before and after dust removal in a negative pressure box, respectively dig out the surface state characteristics of the cluster distribution of cosmetic packaging hoses before and after dust removal, and then characterize the dust removal effect of the packaging hose based on the characteristic differences between the two, and guide the power adjustment of the dust collection mechanism based on this. In this way, intelligent control of the dust collection mechanism can be achieved, and its operating power can be dynamically optimized to optimize its cleaning effect and energy efficiency ratio.
[0055] Based on this, in the technical solution of the present application, further on the basis of the above-mentioned negative pressure automatic conveying device, cameras are arranged above the two side areas of the negative pressure box, wherein the front area of the negative pressure box is the first area, and the rear area of the negative pressure box is the second area, and the camera is used to collect a first surface state image of the distribution of the cluster of cosmetic packaging hoses in the first area, and a second surface state image of the distribution of the cluster of cosmetic packaging hoses in the second area, and transmit the first surface state image and the second surface state image to the dust collection controller for image comparison and analysis to generate a power control instruction for the dust collection mechanism, wherein the dust collection controller is deployed above the negative pressure box and electrically connected to the dust collection mechanism.
[0056] Correspondingly, the intelligent conveying system for intelligent packaging of cosmetics comprises a conveyor frame, one end of the conveyor frame is movably mounted with a driving rotating shaft, the other end of the conveyor frame is movably mounted with a driven rotating shaft, the bottom of the conveyor frame is fixedly mounted with a supporting leg, the bottom of the supporting leg is fixedly mounted with a supporting foot pad, the bottom of the supporting foot pad is fixedly mounted with an anti-slip rubber pad, a negative pressure mechanism is provided at the outer middle position of the conveyor frame, the negative pressure mechanism comprises a negative pressure box, a blower, a dust collection mechanism, an exhaust branch pipe and an air hood, the dust collection mechanism comprises a dust collection barrel, a dust filter hood, an exhaust main pipe, an air duct, a bag filter and an air filter, a cooling mechanism is provided at the other side of the upper part of the negative pressure box, the cooling mechanism comprises a filter box, a first cooling air duct and a second cooling air duct , a first supporting and fixing rod, a second supporting and fixing rod and a filter net, hose flow grooves are provided on both side surfaces of the negative pressure box, and cameras are provided above the two side areas of the negative pressure box, wherein the front area of the negative pressure box is the first area, and the rear area of the negative pressure box is the second area, wherein the camera is used to collect a first surface state image of the distribution of the cluster of cosmetic packaging hoses in the first area, and a second surface state image of the distribution of the cluster of cosmetic packaging hoses in the second area, and transmit the first surface state image and the second surface state image to the dust collection controller 800 for image comparison and analysis to generate a power control instruction for the dust collection mechanism, wherein the dust collection controller 800 is deployed above the negative pressure box and is electrically connected to the dust collection mechanism.
[0057] Furthermore, if Figure 1As shown, the dust collection controller 800 includes: an image feature extraction module 810, which is used to extract surface state features of the first surface state image and the second surface state image respectively to obtain a first cosmetic packaging hose cluster distribution surface state feature map and a second cosmetic packaging hose cluster distribution surface state feature map; a feature autocorrelation enhancement module 820, which is used to perform feature autocorrelation enhancement on the first cosmetic packaging hose cluster distribution surface state feature map and the second cosmetic packaging hose cluster distribution surface state feature map respectively to obtain a first cosmetic packaging hose cluster distribution surface enhancement state feature map and a second cosmetic packaging hose cluster distribution surface enhancement state feature map; a dust collection effect analysis module 830, which is used to generate the power control instruction based on the feature difference between the first cosmetic packaging hose cluster distribution surface enhancement state feature map and the second cosmetic packaging hose cluster distribution surface enhancement state feature map.
[0058] Specifically, in the data processing process of the dust collection controller, first, when performing the surface state detection of the cluster distribution of the cosmetic packaging hose, it is necessary to consider both the surface micro-details and the overall cluster characteristics of the packaging hose. Therefore, in the technical solution of the present application, a surface state feature extractor based on a dilated convolutional neural network model is used to extract features of the first surface state image and the second surface state image respectively. A person of ordinary skill in the art should know that dilated convolution, also known as dilated convolution, can increase the size of the receptive field without changing the number of convolution kernel parameters and the amount of calculation, so that it can capture more extensive and rich spatial context information, and retain local detail features at the same time, so that the surface micro-impurity features and the overall clean state features of the packaging hose in the image can be retained at the same time, so as to generate a first cosmetic packaging hose cluster distribution surface state feature map and a second cosmetic packaging hose cluster distribution surface state feature map.
[0059] Correspondingly, the image feature extraction module 810 is used to: respectively pass the first surface state image and the second surface state image through a surface state feature extractor based on a void convolutional neural network model to obtain the first cosmetic packaging hose cluster distribution surface state feature map and the second cosmetic packaging hose cluster distribution surface state feature map.
[0060] Next, in order to further enhance the discrimination and stability of feature representation, in the technical solution of the present application, a self-attention enhancement module based on cross-channel cross-feature spatial structure preservation is introduced to perform feature enhancement processing on the first cosmetic packaging hose cluster distribution surface state feature map and the second cosmetic packaging hose cluster distribution surface state feature map. Specifically, the module first performs layer normalization processing on the surface state feature map to eliminate the scale difference between layers and ensure the stability of feature processing. Then, the channel context association and spatial structure information of the feature map are captured through multi-layer convolution operations, and its channel context association information is used as a query, and the spatial structure information is used as the key and value. Based on the self-attention mechanism, the feature cross-channel global association interaction is performed, so that while retaining the spatial structure of the feature map, the channel association information between the feature structures is fused in units of the feature spatial structure, and the expression ability of the feature map is enhanced to generate the first cosmetic packaging hose cluster distribution surface enhancement state feature map and the second cosmetic packaging hose cluster distribution surface enhancement state feature map, providing a more accurate feature representation for the subsequent cleaning effect analysis.
[0061] Correspondingly, the feature autocorrelation enhancement module 820 is used to: input the first cosmetic packaging hose cluster distribution surface state feature map and the second cosmetic packaging hose cluster distribution surface state feature map into the feature space structure maintenance self-attention enhancement module based on cross-channel intersection respectively to obtain the first cosmetic packaging hose cluster distribution surface enhancement state feature map and the second cosmetic packaging hose cluster distribution surface enhancement state feature map.
[0062] Specifically, Figure 2 As shown, the feature autocorrelation enhancement module 820 includes: a layer normalization unit 821, which is used to perform layer normalization on the first cosmetic packaging hose cluster distribution surface state feature map to obtain a normalized first cosmetic packaging hose cluster distribution surface state feature map; a channel context association feature extraction unit 822, which is used to perform point convolution processing on the normalized first cosmetic packaging hose cluster distribution surface state feature map to obtain a packaging hose cluster distribution surface state channel context association representation feature map; a spatial context association feature extraction unit 823, which is used to perform convolution encoding on the packaging hose cluster distribution surface state channel context association representation feature map to obtain a packaging hose cluster distribution surface state space context association representation feature map; a global interactive attention fusion unit 824, which is used to perform channel-space global interactive attention fusion on the packaging hose cluster distribution surface state channel context association representation feature map and the packaging hose cluster distribution surface state space context association representation feature map to obtain the first cosmetic packaging hose cluster distribution surface enhancement state feature map.
[0063] Among them, the global interactive attention fusion unit 824 is used to: copy the packaging hose cluster distribution surface state space context association representation feature map to obtain a backup packaging hose cluster distribution surface state space context association representation feature map; perform feature shape reshaping on the packaging hose cluster distribution surface state channel context association representation feature map, the packaging hose cluster distribution surface state space context association representation feature map and the backup packaging hose cluster distribution surface state space context association representation feature map to obtain a packaging hose cluster distribution surface state channel context association representation feature matrix, a packaging hose cluster distribution surface state space context association representation feature matrix and a backup packaging hose cluster distribution surface state space context association representation feature matrix; calculate the packaging hose cluster distribution surface state space context association representation feature matrix. The cross-channel cross covariance matrix between the cloth surface state channel context association representation feature matrix and the packaging hose cluster distribution surface state space context association representation feature matrix; the cross-channel cross covariance matrix is activated using the Softmax function to obtain the packaging hose cluster distribution surface state feature global interaction attention matrix; the product between the backup packaging hose cluster distribution surface state space context association representation feature matrix and the packaging hose cluster distribution surface state feature global interaction attention matrix is calculated to obtain the attention-enhanced packaging hose cluster distribution surface state feature representation matrix; the attention-enhanced packaging hose cluster distribution surface state feature representation matrix is feature-reshaped to obtain the first cosmetic packaging hose cluster distribution surface enhancement state feature map.
[0064] In a specific example, the feature autocorrelation enhancement module 820 is used to process the first cosmetic packaging hose cluster distribution surface state feature map using the following autocorrelation attention enhancement formula to obtain the first cosmetic packaging hose cluster distribution surface enhancement state feature map, wherein the autocorrelation attention enhancement formula is:
[0065] F ln =Layer Normalization(F i )
[0066] F Q =Conv 1×1 (F ln )
[0067] F K =Conv 3×3 (Conv 1×1 (F ln ))
[0068] F V =Copy(F K )
[0069] MQ =Reshape(F Q )
[0070] M K =Reshape(F K )
[0071] M V =Reshape(F V )
[0072]
[0073] Among them, F i represents the surface state feature map of the first cosmetic packaging tube cluster distribution, LayerNormalization(·) represents the layer normalization operation, F ln represents the normalized surface state feature map of the first cosmetic packaging tube cluster distribution, Conv 1×1 represents point convolution, F Q Represents the distribution surface state channel context association feature map of the packaging hose cluster, Conv 3×3 represents the convolution processing based on 3×3 convolution kernel, F K represents the packaging hose cluster distribution surface state space context association representation feature map, Copy(·) represents the copy operation, F V represents the backup packaging hose cluster distribution surface state space context association represents the feature map, reshape(·) represents the feature shape reshaping, M Q 、M K and M V They represent the channel context association representation feature matrix of the distribution surface state of the packaging hose cluster, the spatial context association representation feature matrix of the distribution surface state of the packaging hose cluster, and the backup packaging hose cluster distribution surface state spatial context association representation feature matrix, M a represents the cross-channel cross covariance matrix, θ is the scaling factor, softmax represents the normalized exponential function, Represents matrix multiplication operation, F a A characteristic diagram showing the surface enhancement state of the first cosmetic packaging hose cluster distribution.
[0074] The first cosmetic packaging hose cluster distribution surface enhancement state characteristic diagram and the second cosmetic packaging hose cluster distribution surface enhancement state characteristic diagram respectively represent the surface state information of the cosmetic packaging hose before and after the negative pressure dust removal treatment. In order to quantitatively analyze the dust removal effect of the dust collection mechanism, the present application further calculates the positional difference between the two to reveal the degree of change in the surface state of the packaging hose and generate a fine-grained differential characteristic diagram of the surface state.
[0075] Furthermore, the surface state fine-grained differential feature map is input into the classifier-based adaptive control module to obtain a control instruction, and the control instruction is used to indicate whether the power of the dust collection mechanism should be increased, decreased, or kept unchanged. It should be understood that after receiving the surface state fine-grained differential feature map, the classifier-based adaptive control module can effectively identify whether the current dust removal effect meets the expected standard by performing feature learning on the difference information in the surface state fine-grained differential feature map, thereby judging whether the power of the dust collection mechanism should be increased, decreased, or kept unchanged.
[0076] Accordingly, if Figure 3 As shown, the dust collection effect analysis module 830 includes: a difference measurement unit 831, used to extract the difference characteristics between the first cosmetic packaging hose cluster distribution surface enhancement state feature map and the second cosmetic packaging hose cluster distribution surface enhancement state feature map to obtain a surface state fine-grained differential feature map; a control instruction generation unit 832, used to input the surface state fine-grained differential feature map into a classifier-based adaptive control module to obtain the power control instruction, and the power control instruction is used to indicate that the power of the dust collection mechanism should be increased, decreased or kept unchanged.
[0077] Wherein, the difference measurement unit 831 is used to calculate the positional difference between the first cosmetic packaging hose cluster distribution surface enhancement state feature map and the second cosmetic packaging hose cluster distribution surface enhancement state feature map to obtain the surface state fine-grained differential feature map.
[0078] It should be understood that the role of the classifier is to use the given categories and known training data to learn classification rules and classifiers, and then classify (or predict) unknown data. Logistic regression and SVM are often used to solve binary classification problems. For multi-class classification problems, logistic regression or SVM can also be used, but multiple binary classifications are required to form a multi-classification. However, this is prone to errors and is not efficient. Commonly used multi-classification methods include the Softmax classification function.
[0079] In the embodiment of the present application, it is considered that the first cosmetic packaging hose cluster distribution surface reinforcement state feature map and the second cosmetic packaging hose cluster distribution surface reinforcement state feature map represent the image semantic reinforcement features of the cosmetic packaging hose cluster distribution in the first area and the cosmetic packaging hose cluster distribution in the second area respectively. However, when the first cosmetic packaging hose cluster distribution surface state feature map and the second cosmetic packaging hose cluster distribution surface state feature map are respectively input into the feature space structure preserving self-attention reinforcement module based on cross-channel intersection for feature reinforcement, the cross-channel cross self-attention reinforcement mechanism of the feature space structure preserving self-attention reinforcement module based on cross-channel intersection will aggravate the fine-grained correspondence deviation of the feature distribution between the first cosmetic packaging hose cluster distribution surface reinforcement state feature map and the second cosmetic packaging hose cluster distribution surface reinforcement state feature map. In this way, when calculating the positional difference between the first cosmetic packaging hose cluster distribution surface enhancement state feature map and the second cosmetic packaging hose cluster distribution surface enhancement state feature map, the obtained surface state fine-grained differential feature map will also cause insufficient differential coding coverage due to the fine-grained correspondence offset of the feature distribution, thereby causing outlier class inference mapping deviation, affecting the accuracy of the control instructions obtained by the surface state fine-grained differential feature map through the classifier-based adaptive control module.
[0080] In a preferred embodiment, the control instruction generating unit 832 includes: an optimization subunit, used to optimize the surface state fine-grained differential feature map to obtain an optimized surface state fine-grained differential feature map; and a classification subunit, used to input the optimized surface state fine-grained differential feature map into the classifier-based adaptive control module to obtain the control instruction.
[0081] The optimization subunit is used to: calculate the characteristic mean of the surface state fine-grained differential feature map, and divide the characteristic mean by the difference between the maximum characteristic value and the minimum characteristic value of the surface state fine-grained differential feature map to obtain a surface state fine-grained differential distribution characterization value; subtract the surface state fine-grained differential distribution characterization value and divide it by the surface state fine-grained differential distribution characterization value to obtain a surface state fine-grained differential distribution modulation value; activate the surface state fine-grained differential feature map through a probabilistic function to obtain a probabilistic surface state fine-grained differential feature map; perform point subtraction between the probabilistic surface state fine-grained differential feature map and the surface state fine-grained differential distribution modulation value, take the absolute value and calculate Calculate the negative of the logarithmic value with base 2 to obtain a probabilistic surface state fine-grained differential distribution modulation information feature map; divide the surface state fine-grained differential distribution characterization value by one minus the difference of each eigenvalue of the probabilistic surface state fine-grained differential feature map, sum all eigenvalues of the probabilistic surface state fine-grained differential feature map, and divide by the scale of the surface state fine-grained differential feature map to obtain a probabilistic surface state fine-grained differential distribution modulation bias value; perform point addition on the probabilistic surface state fine-grained differential distribution modulation information feature map and the product of the probabilistic surface state fine-grained differential distribution modulation bias value and the weight as a hyperparameter to obtain an optimized surface state fine-grained differential feature map.
[0082] The calculation process of the optimized surface state fine-grained differential feature map is expressed as follows:
[0083]
[0084] And among them:
[0085]
[0086] in, represents the feature mean of the fine-grained differential feature map of the surface state, f max and f min represents the maximum eigenvalue and the minimum eigenvalue in the surface state fine-grained differential feature map, respectively; p represents the surface state fine-grained differential distribution representation value; F represents the probabilistic surface state fine-grained differential feature map obtained after the surface state fine-grained differential feature map is activated by a probabilistic function; f irepresents the i-th eigenvalue of the probabilistic surface state fine-grained differential feature map, log represents the logarithmic function value with base 2, ε is the weight as a hyperparameter, S is the scale of the surface state fine-grained differential feature map, that is, the width of the feature matrix of the surface state fine-grained differential feature map multiplied by the height and then multiplied by the number of channels of the surface state fine-grained differential feature map, and F represents the optimized surface state fine-grained differential feature map.
[0087] That is, in the preferred example described above, the eigenvalue-based probability information distribution planning of the surface state fine-grained differential feature map is performed by using the Bernoulli probability modulation distribution of the surface state fine-grained differential feature map relative to the eigenvalue distribution, and the probability reverse mapping of the probability characteristics of the surface state fine-grained differential feature map as a whole is used as an extended coverage of the set mapping space of the surface state fine-grained differential feature map to independently understand the interaction path between the intuitive probability information distribution and the abstract probability space mapping of the surface state fine-grained differential feature map, so as to improve the accuracy of the control instructions obtained by the optimized surface state fine-grained differential representation vector through the classifier-based adaptive control module by avoiding the counterfactual reasoning mapping of the outlier feature distribution of the surface state fine-grained differential feature map to the class regression probability.
[0088] Based on the above embodiments, see Figure 4 As shown in FIG. 1 , it is a flow chart of an intelligent conveying method for intelligent packaging of cosmetics in an embodiment of the present application. Figure 4 As shown, according to the embodiment of the present application, the intelligent conveying method for intelligent packaging of cosmetics includes the steps of: S510, acquiring a first surface state image of the distribution of the cluster of cosmetic packaging hoses in the first area, and a second surface state image of the distribution of the cluster of cosmetic packaging hoses in the second area, collected by a camera; S520, performing surface state feature extraction on the first surface state image and the second surface state image respectively to obtain a first cosmetic packaging hose cluster distribution surface state feature map and a second cosmetic packaging hose cluster distribution surface state feature map; S530, performing feature autocorrelation enhancement on the first cosmetic packaging hose cluster distribution surface state feature map and the second cosmetic packaging hose cluster distribution surface state feature map respectively to obtain a first cosmetic packaging hose cluster distribution surface enhancement state feature map and a second cosmetic packaging hose cluster distribution surface enhancement state feature map; S540, generating the power control instruction based on the feature difference between the first cosmetic packaging hose cluster distribution surface enhancement state feature map and the second cosmetic packaging hose cluster distribution surface enhancement state feature map.
[0089] In one example, surface state feature extraction is performed on the first surface state image and the second surface state image respectively to obtain a first cosmetic packaging hose cluster distribution surface state feature map and a second cosmetic packaging hose cluster distribution surface state feature map, including: passing the first surface state image and the second surface state image respectively through a surface state feature extractor based on a void convolutional neural network model to obtain the first cosmetic packaging hose cluster distribution surface state feature map and the second cosmetic packaging hose cluster distribution surface state feature map.
[0090] Here, those skilled in the art can understand that the specific operations of each step in the above-mentioned intelligent delivery method for intelligent packaging of cosmetics have been referred to above. Figures 1 to 3 The dust collection controller 800 has been introduced in detail in the description of the intelligent conveying system for intelligent packaging of cosmetics, and therefore, its repeated description will be omitted.
[0091] Figure 5 FIG. 1 is an application scenario diagram of the intelligent conveying method for intelligent packaging of cosmetics according to an embodiment of the present application. Figure 5 As shown, in this application scenario, first, a first surface state image (for example, Figure 5 D1 as shown in the figure) and a second surface state image of the distribution of the cosmetic packaging hose clusters in the second area (for example, Figure 5 Then, the first surface state image and the second surface state image are input to a server (for example, Figure 5 As shown in S), the server is capable of using the intelligent conveying algorithm for the cosmetics intelligent packaging to process the first surface state image and the second surface state image to obtain a power control instruction indicating that the power of the dust collection mechanism should be increased, decreased or maintained unchanged.
[0092] Based on the above embodiments, another exemplary embodiment of an electronic device is also provided in the embodiments of the present application. In some possible implementations, the electronic device in the embodiments of the present application may include a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor can implement the steps of the intelligent delivery method for intelligent packaging of cosmetics in the above embodiments when executing the program.
[0093] The embodiment of the present application also provides a computer-readable storage medium, on which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, the intelligent delivery method for intelligent packaging of cosmetics according to the embodiment of the present application described with reference to the above figures can be executed. The computer-readable storage medium includes, but is not limited to, for example, volatile memory and / or non-volatile memory. The volatile memory may, for example, include a random access memory (RAM) and / or a cache memory (cache), etc. The non-volatile memory may, for example, include a read-only memory (ROM), a hard disk, a flash memory, etc.
[0094] The embodiments of the present application also provide a computer program product or a computer program, which includes computer executable instructions, and the computer executable instructions are stored in a computer readable storage medium. The processor of the computer device reads the computer executable instructions from the computer readable storage medium, and the processor executes the computer executable instructions, so that the computer device executes the intelligent delivery method for intelligent packaging of cosmetics according to the embodiments of the present application.
[0095] Those skilled in the art will appreciate that the contents disclosed in this application may be subject to various variations and improvements. For example, the various devices or components described above may be implemented by hardware, or by software, firmware, or a combination of some or all of the three.
[0096] In addition, although the present application makes various references to certain units in the system according to embodiments of the present application, any number of different units can be used and run on the client and / or server. The units are only illustrative, and different aspects of the system and method can use different units.
[0097] Those skilled in the art will appreciate that all or part of the steps in the above method can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium, such as a read-only memory, a disk or an optical disk. Optionally, all or part of the steps in the above embodiment can also be implemented using one or more integrated circuits. Accordingly, each module / unit in the above embodiment can be implemented in the form of hardware or in the form of software function modules. The present application is not limited to any particular form of combination of hardware and software.
[0098] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those of ordinary skill in the art to which this application belongs. It should also be understood that terms such as those defined in common dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an idealized or extremely formal sense, unless explicitly defined as such herein.
[0099] The above is an explanation of the present application and should not be considered as limiting thereof. Although several exemplary embodiments of the present application have been described, those skilled in the art will readily appreciate that many modifications may be made to the exemplary embodiments without departing from the novel teachings and advantages of the present application.
Claims
1. An intelligent conveying system for intelligent packaging of cosmetics, comprising a conveyor frame, one end of the conveyor frame is movably mounted with an active rotating shaft, the other end of the conveyor frame is movably mounted with a driven rotating shaft, the bottom of the conveyor frame is fixedly mounted with a support leg, the bottom of the support leg is fixedly mounted with a support foot pad, the bottom of the support foot pad is fixedly mounted with an anti-slip rubber pad, a negative pressure mechanism is provided at the outer middle position of the conveyor frame, the negative pressure mechanism comprises a negative pressure box, a blower, a dust collection mechanism, an exhaust branch pipe and an air hood, the dust collection mechanism comprises a dust collection cylinder, a dust filter cover, an exhaust main pipe, an air guide pipe, a bag filter and an air filter, a cooling mechanism is provided at the other side of the upper part of the negative pressure box, the cooling mechanism comprises a filter box, a first cooling air duct, a second cooling air duct, a first supporting fixing rod, a second supporting fixing rod and a filter, both side surfaces of the negative pressure box are provided with hose flow grooves, and both side areas of the negative pressure box are provided with cameras above, wherein, The front area of the negative pressure box is the first area, and the rear area of the negative pressure box is the second area, and it is characterized in that the camera is used to collect a first surface state image of the distribution of the cluster of cosmetic packaging hoses in the first area, and a second surface state image of the distribution of the cluster of cosmetic packaging hoses in the second area, and transmit the first surface state image and the second surface state image to the dust collection controller for image comparison and analysis to generate a power control instruction for the dust collection mechanism, wherein the dust collection controller is deployed above the negative pressure box and electrically connected to the dust collection mechanism.
2. The intelligent conveying system for intelligent packaging of cosmetics according to claim 1, characterized in that: The dust collection controller comprises: An image feature extraction module, used to extract surface state features from the first surface state image and the second surface state image respectively to obtain a first cosmetic packaging hose cluster distribution surface state feature map and a second cosmetic packaging hose cluster distribution surface state feature map; A feature autocorrelation enhancement module is used to perform feature autocorrelation enhancement on the first cosmetic packaging hose cluster distribution surface state feature map and the second cosmetic packaging hose cluster distribution surface state feature map to obtain the first cosmetic packaging hose cluster distribution surface enhancement state feature map and the second cosmetic packaging hose cluster distribution surface enhancement state feature map; The dust collection effect analysis module is used to generate the power control instruction based on the characteristic difference between the first cosmetic packaging hose cluster distribution surface enhancement state characteristic diagram and the second cosmetic packaging hose cluster distribution surface enhancement state characteristic diagram.
3. The intelligent conveying system for intelligent packaging of cosmetics according to claim 2, characterized in that: The image feature extraction module is used to: The first surface state image and the second surface state image are respectively passed through a surface state feature extractor based on a hole convolutional neural network model to obtain the first cosmetic packaging hose cluster distribution surface state feature map and the second cosmetic packaging hose cluster distribution surface state feature map.
4. The intelligent conveying system for intelligent packaging of cosmetics according to claim 3, characterized in that: The feature autocorrelation enhancement module is used to: The first cosmetic packaging hose cluster distribution surface state characteristic map and the second cosmetic packaging hose cluster distribution surface state characteristic map are respectively input into the feature space structure maintaining self-attention enhancement module based on cross-channel intersection to obtain the first cosmetic packaging hose cluster distribution surface enhancement state characteristic map and the second cosmetic packaging hose cluster distribution surface enhancement state characteristic map.
5. The intelligent conveying system for intelligent packaging of cosmetics according to claim 4, characterized in that: The feature autocorrelation enhancement module includes: A layer normalization unit, used for performing layer normalization on the first cosmetic packaging hose cluster distribution surface state characteristic map to obtain a normalized first cosmetic packaging hose cluster distribution surface state characteristic map; A channel context-related feature extraction unit, used for performing point convolution processing on the normalized first cosmetic packaging hose cluster distribution surface state feature map to obtain a packaging hose cluster distribution surface state channel context-related representation feature map; A spatial context-related feature extraction unit, used for performing convolution encoding on the channel context-related representation feature map of the distribution surface state of the packaging hose cluster to obtain a spatial context-related representation feature map of the distribution surface state of the packaging hose cluster; A global interactive attention fusion unit is used to perform channel-space global interactive attention fusion on the channel context association representation feature map of the packaging hose cluster distribution surface state and the spatial context association representation feature map of the packaging hose cluster distribution surface state to obtain the first cosmetic packaging hose cluster distribution surface reinforcement state feature map.
6. The intelligent conveying system for intelligent packaging of cosmetics according to claim 5, characterized in that: The global interactive attention fusion unit is used to: Copying the packaging hose cluster distribution surface state space context association representation feature map to obtain a backup packaging hose cluster distribution surface state space context association representation feature map; Reshape the packaging hose cluster distribution surface state channel context association representation feature map, the packaging hose cluster distribution surface state space context association representation feature map and the backup packaging hose cluster distribution surface state space context association representation feature map to obtain a packaging hose cluster distribution surface state channel context association representation feature matrix, a packaging hose cluster distribution surface state space context association representation feature matrix and a backup packaging hose cluster distribution surface state space context association representation feature matrix; Calculate the cross-channel cross covariance matrix between the channel context association representation feature matrix of the packaging hose cluster distribution surface state and the spatial context association representation feature matrix of the packaging hose cluster distribution surface state; Use the Softmax function to activate the cross-channel cross-covariance matrix to obtain a global interactive attention matrix of the surface state features of the packaging hose cluster distribution; Calculate the product of the backup packaging hose cluster distribution surface state space context association representation feature matrix and the packaging hose cluster distribution surface state feature global interactive attention matrix to obtain an attention-enhanced packaging hose cluster distribution surface state feature representation matrix; The feature shape of the attention-enhanced packaging hose cluster distribution surface state feature representation matrix is reshaped to obtain the first cosmetic packaging hose cluster distribution surface enhancement state feature map.
7. The intelligent conveying system for intelligent packaging of cosmetics according to claim 6, characterized in that: The dust collection effect analysis module comprises: A difference measurement unit, used for extracting difference features between the first cosmetic packaging hose cluster distribution surface enhancement state feature map and the second cosmetic packaging hose cluster distribution surface enhancement state feature map to obtain a surface state fine-grained differential feature map; A control instruction generating unit is used to input the surface state fine-grained differential feature map into a classifier-based adaptive control module to obtain the power control instruction, wherein the power control instruction is used to indicate whether the power of the dust collection mechanism should be increased, decreased or kept unchanged.
8. The intelligent conveying system for intelligent packaging of cosmetics according to claim 7, characterized in that: The difference measurement unit is used for: The positional difference between the first cosmetic packaging hose cluster distribution surface enhancement state characteristic map and the second cosmetic packaging hose cluster distribution surface enhancement state characteristic map is calculated to obtain the surface state fine-grained differential characteristic map.
9. An intelligent conveying method for intelligent packaging of cosmetics, characterized in that: include: Acquire a first surface state image of the distribution of the cosmetic packaging hose clusters in the first area and a second surface state image of the distribution of the cosmetic packaging hose clusters in the second area collected by the camera; Performing surface state feature extraction on the first surface state image and the second surface state image respectively to obtain a first cosmetic packaging hose cluster distribution surface state feature map and a second cosmetic packaging hose cluster distribution surface state feature map; Performing feature autocorrelation enhancement on the first cosmetic packaging hose cluster distribution surface state characteristic map and the second cosmetic packaging hose cluster distribution surface state characteristic map to obtain the first cosmetic packaging hose cluster distribution surface enhanced state characteristic map and the second cosmetic packaging hose cluster distribution surface enhanced state characteristic map; The power control instruction is generated based on the characteristic difference between the first cosmetic packaging hose cluster distribution surface enhancement state characteristic diagram and the second cosmetic packaging hose cluster distribution surface enhancement state characteristic diagram.
10. The intelligent conveying method for intelligent packaging of cosmetics according to claim 9, characterized in that: Surface state feature extraction is performed on the first surface state image and the second surface state image respectively to obtain a first cosmetic packaging hose cluster distribution surface state feature map and a second cosmetic packaging hose cluster distribution surface state feature map, including: The first surface state image and the second surface state image are respectively passed through a surface state feature extractor based on a hole convolutional neural network model to obtain the first cosmetic packaging hose cluster distribution surface state feature map and the second cosmetic packaging hose cluster distribution surface state feature map.
Citation Information
Patent Citations
Intelligent call quality detection method and device and storage medium
CN118646823A
Hotel installation project pipeline arrangement optimization system and method based on BIM technology
CN118709341A
Charging and discharging energy storage system capable of automatically adjusting and compensating overload imbalance
CN118868202A
Printing method of carbon oil printed circuit board (PCB)
CN118890800A
Code scanning group entering method and device and electronic equipment
CN119094486A
