Intelligent closestool production management system and method
Through deep learning technology, the feature vectors of toilet paper are extracted and combined with a classifier have been solved, and the problem of difficulty in accurately judging the quality of toilet paper in the existing technology is solved, automatic detection is realized, and production efficiency is improved.
Patent Information
- Application Number
- CN202411668702.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-21
- Publication Date
- 2025-03-07
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
It is difficult to accurately judge the quality and integrity of toilet paper, and manual inspections have problems of error and low efficiency.
The toilet paper production image is obtained through the camera, and the global correlation feature vector and attention feature vector are extracted using deep learning technology. Combined with a classifier, we can determine whether there are quality problems with the toilet paper.
Automatic quality inspection is realized, reducing the risk of unqualified products entering the market, reducing the workload of manual inspections, and improving production efficiency.
Smart Images

Figure CN119578993A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of smart toilets, and more specifically, to a smart toilet production management system and method. Background Art
[0002] Toilet paper is a product made of toilet paper, usually used on the seat of the toilet in public toilets or at home to provide extra protection and comfort. Its main function is to isolate the body from direct contact with the toilet surface, reducing the risk of bacterial transmission and cross infection.
[0003] Considering that toilet seat covers are usually very thin and light, manual inspection of their quality and integrity becomes relatively difficult. First, it may be difficult for staff to accurately judge the condition of each toilet seat cover during inspection. Second, due to its soft material, it is easy to tear, fold or other shape changes, which increases the difficulty for staff during inspection. In addition, manual inspection is easily affected by fatigue and lack of concentration, which increases the possibility of errors.
[0004] Therefore, a smart toilet production management system and method are desired. Summary of the invention
[0005] In order to solve the above technical problems, this application is proposed. The embodiment of this application provides an intelligent toilet production management system and method, which first obtains the toilet paper production image collected by the camera, then uses deep learning technology to perform feature extraction and association analysis, and finally obtains the classification result through the classifier to determine whether the toilet paper has quality problems, so as to timely discover and deal with quality problems, reduce the risk of unqualified products entering the market, and reduce the workload of manual inspection, efficiently complete quality monitoring, and improve overall production efficiency.
[0006] According to one aspect of the present application, a smart toilet production management system is provided, which includes:
[0007] A toilet paper production image acquisition module, used to acquire toilet paper production images collected by a camera;
[0008] A toilet paper production image extraction module, used to extract a toilet paper global association feature vector and a toilet paper attention feature vector from the toilet paper production image collected by the camera;
[0009] The toilet paper quality judgment module is used to judge whether the toilet paper has quality problems based on the toilet paper global association feature vector and the toilet paper attention feature vector.
[0010] According to another aspect of the present application, a smart toilet production management method is provided, which includes:
[0011] Obtain toilet paper production images captured by a camera;
[0012] Extracting a toilet paper global association feature vector and a toilet paper attention feature vector from the toilet paper production image captured by the camera;
[0013] Based on the toilet paper global association feature vector and the toilet paper attention feature vector, it is determined whether the toilet paper has quality problems.
[0014] Compared with the prior art, the present application provides an intelligent toilet production management system and method, which first obtains the toilet paper production image collected by a camera, then uses deep learning technology to perform feature extraction and association analysis, and finally obtains the classification result through a classifier to determine whether there are quality problems with the toilet paper, thereby promptly discovering and handling quality problems, reducing the risk of unqualified products entering the market, and reducing the workload of manual inspection, efficiently completing quality monitoring, and improving overall production efficiency. BRIEF DESCRIPTION OF THE DRAWINGS
[0015] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.
[0016] Figure 1 It is a block diagram of the smart toilet production management system according to an embodiment of the present application.
[0017] Figure 2 It is a block diagram of the toilet paper production image extraction module in the smart toilet production management system according to an embodiment of the present application.
[0018] Figure 3 It is a block diagram of a toilet paper interesting feature extraction unit in a smart toilet production management system according to an embodiment of the present application.
[0019] Figure 4 This is a block diagram of a toilet paper quality judgment module in an intelligent toilet production management system according to an embodiment of the present application.
[0020] Figure 5 This is a flow chart of a smart toilet production management method according to an embodiment of the present application. DETAILED DESCRIPTION
[0021] Various exemplary embodiments, features and aspects of the present application will be described in detail below with reference to the accompanying drawings. The same reference numerals in the accompanying drawings represent elements with the same or similar functions. Although various aspects of the embodiments are shown in the accompanying drawings, the drawings are not necessarily drawn to scale unless otherwise specified.
[0022] The word “exemplary” is used exclusively herein to mean “serving as an example, example, or illustration.” Any embodiment described herein as “exemplary” is not necessarily to be construed as preferred or advantageous over other embodiments.
[0023] In addition, in order to better illustrate the present application, numerous specific details are given in the following specific embodiments. It should be understood by those skilled in the art that the present application can also be implemented without certain specific details. In some examples, methods, means, components and circuits well known to those skilled in the art are not described in detail in order to highlight the subject matter of the present application.
[0024] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of this application, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.
[0025] Figure 1 This is a block diagram of the intelligent toilet production management system of the present application embodiment. Figure 1 As shown, according to the embodiment of the present application, the intelligent toilet production management system 100 includes: a toilet paper production image acquisition module 110, which is used to acquire the toilet paper production image collected by a camera; a toilet paper production image extraction module 120, which is used to extract the toilet paper global association feature vector and the toilet paper attention feature vector from the toilet paper production image collected by the camera; a toilet paper quality judgment module 130, which is used to judge whether the toilet paper has a quality problem based on the toilet paper global association feature vector and the toilet paper attention feature vector.
[0026] In the above-mentioned intelligent toilet production management system 100, the toilet pad production image acquisition module 110 is used to obtain the toilet pad production image collected by the camera. It should be understood that toilet pads, as a product made of toilet paper, are often used on the seat cushions of public toilets or home toilets to provide additional protection and comfort. Its core function is to isolate the direct contact between the body and the toilet surface, thereby reducing the risk of bacterial transmission and cross-infection. However, due to the light and thin nature of toilet pads, manual inspection of their quality and integrity is particularly difficult. First, it may be difficult for staff to accurately judge the status of each toilet pad during the inspection process. Secondly, due to its soft material, it is easy to tear, fold or other morphological changes, which further increases the difficulty of inspection. In addition, manual inspection may also be affected by fatigue and distraction, thereby increasing the probability of errors.
[0027] In the technical solution of the present application, the camera can efficiently capture the image of each toilet paper on the production line, and use the advantages of high resolution and fast acquisition to monitor the production status in real time. In this process, the layout of the camera should cover every link of the production line, especially in the key nodes such as the forming, cutting and packaging of the toilet paper. Among them, a high-resolution industrial camera is used to obtain the production image of the toilet paper. Therefore, in the technical solution of the present application, by obtaining the production image of the toilet paper collected by the camera and combining it with deep learning technology, it is determined whether there are quality problems with the toilet paper, and these problems can be discovered and handled in time, thereby reducing the risk of unqualified products entering the market, while reducing the workload of manual inspection, achieving efficient quality monitoring, and thus improving overall production efficiency.
[0028] In the above-mentioned intelligent toilet production management system 100, the toilet paper production image extraction module 120 is used to extract the toilet paper global correlation feature vector and the toilet paper attention feature vector from the toilet paper production image collected by the camera. It should be understood that by combining the toilet paper global correlation feature vector and the toilet paper attention feature vector, the quality of the toilet paper can be more comprehensively evaluated and potential defects can be identified. This not only improves the detection efficiency, but also reduces the errors that may be caused by manual inspection, ensuring the hygiene and safety of each toilet paper during the production process.
[0029] Figure 2 FIG. 1 is a block diagram of a toilet paper production image extraction module in an intelligent toilet production management system according to an embodiment of the present application. Figure 2As shown, in a specific embodiment of the present application, the toilet paper production image extraction module 120 includes: a toilet paper interest feature extraction unit 121, used to extract interest features from the toilet paper production image captured by the camera to obtain the toilet paper global correlation feature vector; a toilet paper attention feature extraction unit 122, used to extract attention features from the toilet paper production image captured by the camera to obtain the toilet paper attention feature vector.
[0030] It should be understood that extracting features of interest from toilet paper production images captured by a camera can comprehensively reflect the overall information of the image, including key features of shape, color, and texture, to help the system identify whether there are defects or non-compliance with standards.
[0031] Furthermore, attention feature extraction of toilet paper production images captured by the camera can more accurately identify areas in the image that are critical to quality judgment. This process can help the inspection system focus on parts that may have defects, such as tears, stains or irregular shapes, which often have a greater impact on product quality.
[0032] Figure 3 FIG. 1 is a block diagram of a toilet paper interesting feature extraction unit in a smart toilet production management system according to an embodiment of the present application. Figure 3 As shown, in a specific embodiment of the present application, the toilet paper interest feature extraction unit 121 includes: a toilet paper interest feature encoding subunit 1211, used to encode the toilet paper production image captured by the camera to obtain a plurality of toilet paper interest feature vectors; a toilet paper convolution encoding subunit 1212, used to pass the plurality of toilet paper interest feature vectors through a toilet paper interest convolution neural network model as a feature encoder to obtain the toilet paper global correlation feature vector.
[0033] It should be understood that encoding the features of interest in the toilet paper production image captured by the camera can fully capture the information of multiple key areas in the image. Specifically, the toilet paper production image captured by the camera can be denoised and input into the feature extraction model, and the model extracts basic features through multiple convolutional layers. Then, the image is analyzed through a region proposal network or an attention layer to determine the relative importance of each region and generate a series of regions of interest. These regions can be defects, special textures, or other important features that may exist on the toilet paper. Finally, the features of each region of interest are converted into a feature vector of interest through the encoding process, reflecting the unique information of the region. In this way, multiple feature regions can be processed simultaneously, which improves the system's ability to identify potential quality problems.
[0034] Furthermore, processing multiple toilet paper feature vectors of interest through a convolutional neural network model as a feature encoder can integrate information from different feature regions, thereby forming a comprehensive assessment of the overall quality of the product. Specifically, multiple feature vectors of interest obtained through image processing and feature extraction are first input into the feature encoder model. This model is usually based on a convolutional neural network architecture and is specifically designed to capture and fuse features from different regions. The design of the model includes multiple convolutional layers, activation functions, and pooling layers, which gradually extract high-level representations of features through layer-by-layer processing. The input feature vector of interest will undergo a series of convolution operations to capture the relationship between local features, and will be reduced in dimension and aggregated through the pooling layer. Next, the feature encoder will merge the processed feature vectors into a global correlation feature vector through a fully connected layer. This vector comprehensively reflects the feature information of all regions of interest and can provide a comprehensive basis for quality assessment. The final global correlation feature vector not only contains the information of individual features, but also reflects the relationship between them, so that the detection system has a stronger ability to discriminate when judging the quality of toilet paper.
[0035] Specifically, the multiple toilet paper interest feature vectors are passed through a toilet paper interest convolutional neural network model as a feature encoder to obtain the toilet paper global association feature vector, including: using each layer of the toilet paper interest convolutional neural network model as a feature encoder to perform convolution processing, mean pooling processing based on the local feature matrix and nonlinear activation processing on the input data in the forward pass of the layer to output the toilet paper global association feature vector from the last layer of the toilet paper interest convolutional neural network model as a feature encoder, wherein the input of the toilet paper interest convolutional neural network model as a feature encoder is the multiple toilet paper interest feature vectors.
[0036] In a specific embodiment of the present application, the toilet paper interest feature encoding subunit 1211 is used to: pass the toilet paper production image captured by the camera through a toilet paper target detection network to obtain multiple toilet paper interest regions; pass the multiple toilet paper interest regions through a toilet paper interest linear embedding layer to obtain the multiple toilet paper interest feature vectors.
[0037] It should be understood that the toilet paper production image captured by the camera is processed through the toilet paper target detection network to obtain multiple toilet paper interest regions in order to accurately locate and identify key features in the image. In specific operations, the toilet paper production image captured by the camera needs to be input into a designed target detection network, usually based on a deep learning framework (such as YOLO, Faster R-convolutional neural network). These networks are specialized in object detection and can identify specific targets and their locations in the image. The target detection network then extracts image features through multiple convolutional layers and generates feature maps. Subsequently, the network applies a region proposal mechanism to identify possible target regions in the image. These regions are evaluated based on their features, and the model gives a confidence score for each region to determine whether it is an interest region. In the final stage of processing, after the non-maximum suppression (NMS) step, redundant regions are removed, and finally multiple high-confidence toilet paper interest regions are retained. This process not only improves the ability to identify potential defects, but also enables real-time monitoring on the production line, timely detection of problems and measures.
[0038] Accordingly, in a specific example of the present application, the toilet paper target detection network is a Fast R-convolutional neural network, a Faster R-convolutional neural network or a RetinaNet. Specifically, the toilet paper production image captured by the camera is passed through the toilet paper target detection network to obtain multiple toilet paper interest regions, including: using the toilet paper target detection network to process the toilet paper production image captured by the camera using the following target detection formula to obtain the multiple toilet paper interest regions;
[0039] Among them, the target detection formula is:
[0040] ROI=(cls(ψ det ,B),Regr(ψ det ,B))
[0041] Among them, ψ det is the toilet paper production image captured by the camera, B is the anchor frame, ROI is the region of interest of multiple toilet paper, cls(ψ det ,B) indicates classification, Regr(ψ det ,B) indicates regression.
[0042] In a specific embodiment of the present application, the toilet paper target detection network is an anchor window-based target detection network. It should be understood that the toilet paper target detection network is designed as an anchor window-based target detection network in order to improve the accuracy and flexibility of target positioning. Among them, the anchor window (Anchor Boxes) method enhances the detection effect by predefining multiple boxes with different shapes and sizes in the image to better match the actual target in the image. In a specific implementation, the generation of anchor windows usually involves setting multiple anchor boxes at each position of the image. The size and aspect ratio of these anchor boxes can be pre-set according to the characteristics of the target object, so that the network can adapt to different types of targets. When processing the input image, the target detection network will first extract feature maps through the convolution layer, and then combine these feature maps with the predefined anchor window. Each anchor window will be evaluated by the network to generate a corresponding confidence score to determine whether the anchor window contains the target object. At the same time, the network will also predict the coordinate offset of each anchor window to accurately adjust the bounding box of the anchor window so that it better covers the actual target. The mechanism of the anchor window enables the detection network to effectively handle diverse targets while also reducing the false positive rate. By combining region proposals and anchor windows, the network can more accurately capture the features of target objects and improve detection accuracy.
[0043] Furthermore, processing multiple toilet paper regions of interest through a toilet paper linear embedding layer of interest can integrate and simplify complex information, making subsequent processing and judgment more efficient and accurate. Specifically, the main function of the linear embedding layer is to reduce the dimension and map the input high-dimensional features, and convert multiple toilet paper regions of interest into feature vectors of fixed dimensions. This process is usually achieved by multiplying a weight matrix and adding a bias term. Among them, the feature map of each region of interest is flattened, and then a corresponding plurality of toilet paper feature vectors of interest are generated through a linear transformation. These toilet paper feature vectors of interest are simplified representations of each region of interest, retaining important feature information while removing redundant data, making the information more concentrated and easier to process.
[0044] In a specific embodiment of the present application, the toilet paper attention feature extraction unit 122 is used to: pass the toilet paper production image captured by the camera through a toilet paper convolutional neural network based on the attention mechanism to obtain a toilet paper attention feature map; perform maximum pooling on the toilet paper attention feature map to obtain the toilet paper attention feature vector.
[0045] It should be understood that the toilet paper production image captured by the camera is processed by a convolutional neural network based on the attention mechanism, which can focus on the most relevant parts of the image, thereby improving the efficiency and accuracy of feature extraction. Among them, the introduction of the attention mechanism enables the model to adaptively allocate more computing resources to key areas, enhancing the ability to capture important features. In the specific operation, the toilet paper production image captured by the camera is input into the convolutional neural network based on the attention mechanism. The first few layers of the network usually contain convolutional layers and pooling layers, which are designed to extract the basic features of the image. At this stage, the traditional convolution operation is enhanced, and the attention module is introduced to enable the model to assign different importance to different features. The attention mechanism generates a weight map that reflects the importance of each area in the image. Specifically, the network calculates the relationship between the feature vector of each position and the global feature through self-attention or convolutional attention, thereby determining which areas need to be focused on. These weights will be applied to the feature map to strengthen the information of important feature areas while suppressing irrelevant or redundant information. Through this process, the final toilet paper attention feature map will highlight the key parts of the image, such as possible defects or important structural features. This feature map not only effectively reduces the impact of noise, but also improves the performance of subsequent tasks (such as classification and detection), making the system more accurate and efficient in practical applications.
[0046] Specifically, the toilet paper production image captured by the camera is passed through a toilet paper convolutional neural network based on an attention mechanism to obtain a toilet paper attention feature map, including: using the convolutional coding part of the toilet paper convolutional neural network based on the attention mechanism to perform deep convolutional coding on the toilet paper production image captured by the camera to obtain an initial convolutional feature map; inputting the initial convolutional feature map into the spatial attention part of the toilet paper convolutional neural network based on the attention mechanism to obtain a spatial attention map; passing the spatial attention map through a Softmax activation function to obtain a spatial attention feature map; calculating the positional point multiplication of the spatial attention feature map and the initial convolutional feature map to obtain the toilet paper attention feature map. More specifically, the initial convolution feature map is input into the spatial attention part of the toilet paper convolution neural network based on the attention mechanism to obtain a spatial attention map, including: performing average pooling and maximum pooling on the initial convolution feature map along the channel dimension to obtain an average feature matrix and a maximum feature matrix; cascading and channel-adjusting the average feature matrix and the maximum feature matrix to obtain a channel feature matrix; and using the convolution layer of the spatial attention feature map to convolutionally encode the channel feature matrix to obtain the spatial attention map.
[0047] Furthermore, the maximum value pooling is performed on the toilet paper attention feature map, which can extract the most significant features from the high-dimensional feature map, simplify the data, and keep the most important information. This process helps to reduce the dimension of the feature, reduce the computational complexity, and enhance the generalization ability of the model. In specific operation, maximum value pooling is implemented by sliding a fixed-size window (such as 2x2 or 3x3) on the feature map. Each time the window moves, the pooling operation takes the maximum value in the area covered by the window to form a new feature map. In this way, the pooling layer can retain the strongest activation information in the feature map and suppress weaker features, thereby effectively extracting the most important features of each area. When processing the attention feature map of the toilet paper, maximum value pooling can help highlight the key features related to the quality and structure of the toilet paper, such as edges, textures, and shapes. Finally, the pooled feature map will be flattened into a one-dimensional feature vector, which contains the most important information emphasized by the attention mechanism and can be used for subsequent classification, detection, and quality assessment tasks.
[0048] In a specific embodiment of the present application, the toilet paper pad attention feature map is subjected to maximum pooling to obtain the toilet paper pad attention feature vector, which is used to: import necessary Python libraries, including the TensorFlow deep learning framework and the datetime date formatting library; create a max_pooling function to perform a maximum pooling operation, which is used to receive the toilet paper pad attention feature map as input and return the toilet paper pad attention feature vector; use the DateFormat mechanism to record and format the timestamp of the deployment process, which is used to track and record the execution time of each step; call the defined max_pooling function to perform a pooling operation on the toilet paper pad attention feature map, and record the start and end time of the operation; save the pooled toilet paper pad attention feature vector to the database, and record the saved timestamp.
[0049] Part of the deployment code is shown below.
[0050]
[0051]
[0052] Among them, tensorflow is used as a deep learning framework to process image data and perform pooling operations. The datetime library is used to record the timestamp of the operation, which is convenient for tracking and recording the execution time of each step. The max_pooling function is used to receive the attention feature map and the pooling window size as parameters and return the pooled feature vector. This function implements the maximum pooling operation through tf.nn.max_pool to ensure that the most important features are retained during the pooling process. The get_current_time function is used to return a formatted string of the current time, which is convenient for recording the start and end time of the operation. Use get_current_time to record the start time of the maximum pooling operation to ensure that the timestamp of each operation is accurately recorded. Call the max_pooling function to perform a pooling operation on the attention feature map. Use get_current_time to record the end time of the maximum pooling operation to ensure that the timestamp of the operation is accurately recorded. Save the pooled feature vector to the specified file for subsequent analysis and use. In this way, the maximum value pooling operation of the toilet paper attention feature map can be realized, and the timestamp of each step is recorded, so that the key features in the toilet paper production image can be automatically extracted through deep learning technology, reducing the need for manual intervention and improving production efficiency. In addition, the timestamp of each operation is recorded through the DateFormat mechanism, which is easy to deploy and expand. Among them, the pooling window size and feature extraction method can be adjusted according to actual needs to adapt to different production environments and needs. In this way, high-precision quality detection of toilet paper production images can be achieved, improving production efficiency and product quality. At the same time, through real-time monitoring and recording, the traceability and optimizability of the production process are ensured.
[0053] In the above-mentioned intelligent toilet production management system 100, the toilet paper quality judgment module 130 is used to judge whether the toilet paper has quality problems based on the toilet paper global association feature vector and the toilet paper attention feature vector. It should be understood that the global association feature vector is obtained by summarizing the features of the entire image, and usually contains general information about the product, such as size, shape, and color. The attention feature vector focuses on key areas in the image that may indicate potential quality defects, such as foreign matter, blemishes, or irregular edges. By combining these two feature vectors, the model can understand local features in a larger context, thereby more effectively identifying quality problems and reducing the risk of missed detection.
[0054] Figure 4 FIG. 1 is a block diagram of a toilet paper quality judgment module in an intelligent toilet production management system according to an embodiment of the present application. Figure 4As shown, in a specific embodiment of the present application, the toilet paper quality judgment module 130 includes: a toilet paper feature fusion unit 131, which is used to perform heterogeneous feature fusion on the toilet paper global association feature vector and the toilet paper attention feature vector based on the spatial distribution of data points to obtain a toilet paper quality classification feature vector; a toilet paper quality problem classification judgment unit 132, which is used to pass the toilet paper quality classification feature vector through a classifier to obtain a classification result, and the classification result is used to determine whether the toilet paper has a quality problem.
[0055] It should be understood that the global correlation feature vector provides the overall information of the product, while the attention feature vector emphasizes the most critical local features in the image. By combining these two features, the model can fully understand the quality of the toilet paper and accurately identify potential defects.
[0056] In particular, in the technical solution of the present application, the global correlation feature vector of toilet paper is obtained by feature encoding of the area of interest of toilet paper, which reflects the overall characteristics of multiple toilet paper samples, covers shape, texture and color features, has strong globality and comprehensiveness, and is suitable for characterizing the overall quality level of toilet paper. In contrast, the attention feature vector of toilet paper is extracted based on the attention mechanism, aiming to capture more detailed and important local features in the production image. These features tend to describe the details of a specific area, such as flaws, defects or abnormal phenomena in the production process, and thus have stronger locality and pertinence. Due to the differences in the generation mechanism and feature dimension of these two feature vectors, they may be in different manifold structures in the feature space. This heterogeneity may cause the mismatch of the feature manifold, thereby affecting the effect of feature fusion. The mismatch of the feature manifold means that in the multidimensional space, the relationship between the global correlation feature vector of the toilet paper and the attention feature vector of the toilet paper is not smooth enough, so that the fused toilet paper quality classification feature vector lacks smoothness and consistency. The smoothness of the feature manifold is an important indicator for evaluating the quality of feature representation. If the feature manifold is not smooth, the model may be affected when making quality judgments, resulting in uncertainty in the judgment results, and may even misjudge the quality of the toilet paper. Therefore, in the technical solution of the present application, the toilet paper global correlation feature vector and the toilet paper attention feature vector are fused based on the heterogeneous features of the spatial distribution of data points to obtain the toilet paper quality classification feature vector.
[0057] Among them, the global association feature vector of toilet paper and the toilet paper attention feature vector are fused with heterogeneous features based on the spatial distribution of data points to obtain the toilet paper quality classification feature vector, including: constructing a toilet paper distance topological matrix between the toilet paper global association feature vector and the toilet paper attention feature vector; constructing a toilet paper eigenvalue granularity association matrix between the toilet paper global association feature vector and the toilet paper attention feature vector; performing topological association modulation on the toilet paper eigenvalue granularity association matrix based on the toilet paper distance topological matrix to obtain a toilet paper topological modulation association matrix; using the toilet paper topological modulation association matrix as the perspective co-projection space, projecting the toilet paper global association feature vector and the toilet paper attention feature vector to the perspective co-projection space to obtain the perspective modulated toilet paper global association feature vector and the perspective modulated toilet paper attention feature vector; fusing the perspective modulated toilet paper global association feature vector and the perspective modulated toilet paper attention feature vector to obtain the toilet paper quality classification feature vector.
[0058] The fusion step is specifically expressed as:
[0059] M t =D(V1,V2)
[0060] D(V1, V2) is the toilet paper distance topological matrix between the toilet paper global association feature vector and the toilet paper attention feature vector, that is, d i,j =d(v 1i ,v 2j ), and V1 and V2 are both column vectors.
[0061]
[0062] V c =α·V1′+β·V2′
[0063] Wherein, V1 represents the global association feature vector of the toilet paper, V2 represents the attention feature vector of the toilet paper, and M t represents the toilet paper distance topology matrix, T represents the transpose of the vector, Represents matrix multiplication, M v represents the toilet paper eigenvalue granularity association matrix, M c represents the toilet paper topology modulation association matrix, V1' represents the viewpoint modulation toilet paper global association feature vector, V2' represents the viewpoint modulation toilet paper attention feature vector, α and β represent weighted hyperparameters, V c Represents the toilet seat paper quality classification feature vector.
[0064] In the technical solution of the present application, the toilet paper global association feature vector and the toilet paper attention feature vector have relatively significant feature heterogeneity, which results in that when the feature fusion of the toilet paper global association feature vector and the toilet paper attention feature vector is performed, the feature manifold and feature space mismatch caused by the heterogeneity will cause the toilet paper quality classification feature vector to fail to have feature manifold smoothness, thereby affecting the certainty of its feature expression.
[0065] Based on this, in the technical solution of the present application, the toilet paper global correlation feature vector and the toilet paper attention feature vector are subjected to heterogeneous feature fusion based on the spatial distribution of data points. First, a toilet paper distance topological matrix between the toilet paper global correlation feature vector and the toilet paper attention feature vector is constructed. This step utilizes the idea of graph theory, taking the eigenvalues of each position in the toilet paper global correlation feature vector and the toilet paper attention feature vector as nodes, and using the distance metric function to determine the low-dimensional embedding expression of the edges between nodes, so as to construct a feature expression that can quantify the topological correlation relationship between the toilet paper global correlation feature vector and the toilet paper attention feature vector. This step is crucial to understanding the spatial distribution of data points and their mutual connections, and also provides basic structural information for subsequent steps.
[0066] Next, a toilet pad eigenvalue granularity association matrix between the toilet pad global association feature vector and the toilet pad attention feature vector is constructed. This step focuses on exploring the correlation between the internal attributes of the feature vector. In a specific example, the product between the transposed vector of the toilet pad global association feature vector and the toilet pad attention feature vector is calculated to obtain the toilet pad eigenvalue granularity association matrix. Furthermore, the toilet pad eigenvalue granularity association matrix is topologically modulated based on the toilet pad distance topology matrix to obtain the toilet pad topology modulation association matrix. That is, the eigenvalue granularity association information of the toilet pad global association feature vector and the toilet pad attention feature vector is driven to wander on the distance topology map to construct a low-dimensional modulated perspective co-projection space domain for mapping the toilet pad global association feature vector and the toilet pad attention feature vector to a continuous high-dimensional regression space attribute.
[0067] Next, the toilet pad topology modulation association matrix is used as the perspective co-projection space, and the toilet pad global association feature vector and the toilet pad attention feature vector are projected into the perspective co-projection space to obtain the perspective modulated toilet pad global association feature vector and the perspective modulated toilet pad attention feature vector. That is, the key to this technical solution is to find an implicit third space with modulation capability, in which as much information of the original high-dimensional data as possible is retained, so that the converted new feature vector not only reduces the dimension, but also maintains the relative position relationship between the original features.
[0068] Finally, the perspective modulated toilet seat paper global association feature vector and the perspective modulated toilet seat paper attention feature vector are fused to obtain the toilet seat paper quality classification feature vector. Fusion can be achieved in various forms, ranging from simple arithmetic average to complex ensemble learning algorithms, depending on the application scenario and the goals pursued. The final toilet seat paper quality classification feature vector combines the main features of the two sets of input data and provides a more comprehensive data representation, which is not only conducive to improving the performance of machine learning models, but also promotes the ability of cross-modal or multi-source data analysis, thereby significantly enhancing the understanding and utilization efficiency of complex data structures.
[0069] Furthermore, the toilet paper quality classification feature vector is passed through the classifier to obtain the classification result, mainly to realize automated quality inspection and timely identify potential quality problems. Among them, the quality classification feature vector combines the global correlation feature and the attention feature, which can effectively reflect the quality characteristics of the toilet paper. By processing the feature vector through the classifier, the system can judge whether the product meets the quality standard according to the pre-learned pattern. In the specific implementation process, the quality classification feature vector is first input into the trained classifier. Then, through the training stage, the classifier learns the relationship between different features and quality problems with a large amount of labeled data to form a set of decision rules. Then, in actual application, the input toilet paper quality classification feature vector will be processed by the classifier and a classification result will be output. This result is a binary classification (such as "qualified" or "unqualified") label. The decision-making process of the classifier includes calculating the similarity between the feature vector and each category, and selecting the most likely category based on the similarity. Through this process, the system can quickly determine whether there is a quality problem with the toilet paper. This automated quality inspection method not only improves the inspection efficiency, but also reduces the errors caused by human factors, ensuring the stability of product quality. Ultimately, this method provides manufacturers with an efficient means of quality monitoring, which helps to timely discover and correct problems in the production process and improve the market competitiveness of products. Overall, the quality classification feature vector is passed through the classifier to obtain the classification result, which provides a scientific and systematic solution for the quality management of toilet paper.
[0070] In a specific embodiment of the present application, the toilet seat cushion fixing paper sheet is composed of multiple paper sheets stacked and bonded in sequence, wherein each paper sheet is provided with an easy-tear strip on the side, and the tear strip head is fixed by a paginator, and the tear strip can be conveniently separated by pressing a button. Then, after tearing, the toilet paper sheet can be directly put into the toilet, and an automatic tearing and throwing function can be optionally configured, and it has a real-time monitoring function to display the remaining paper sheet amount and whether it is damaged. It is worth mentioning that the adhesive surface of the paper sheet has a certain viscosity and can be quickly decomposed after contacting water, while the side in contact with people has a certain toughness and waterproof performance. In addition, automatic sensing flushing, mosquito disinfecting and deodorization functions have been added to enhance the user experience and hygiene.
[0071] In summary, the embodiment of the present application first obtains the toilet paper production image collected by the camera, and then uses deep learning technology to perform feature extraction and association analysis on it, and finally obtains the classification result through the classifier to determine whether there are quality problems with the toilet paper, thereby timely discovering and handling quality problems, reducing the risk of unqualified products entering the market, and reducing the workload of manual inspection, efficiently completing quality monitoring, and improving overall production efficiency.
[0072] As described above, the smart toilet production management system 100 according to the embodiment of the present application can be implemented in various terminal devices. In one example, the smart toilet production management system 100 can be integrated into the terminal device as a software module and / or a hardware module. For example, the smart toilet production management system 100 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 smart toilet production management system 100 can also be one of the many hardware modules of the terminal device.
[0073] Alternatively, in another example, the smart toilet production management system 100 and the terminal device may also be separate devices, and the smart toilet production management system 100 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.
[0074] Figure 5 Flow chart of the intelligent toilet production management method according to the embodiment of the present application. Figure 5 As shown, according to the intelligent toilet production management method of the embodiment of the present application, it includes: S110, acquiring a toilet paper production image captured by a camera; S120, extracting a toilet paper global association feature vector and a toilet paper attention feature vector from the toilet paper production image captured by the camera; S130, judging whether the toilet paper has a quality problem based on the toilet paper global association feature vector and the toilet paper attention feature vector.
[0075] Here, those skilled in the art can understand that the specific operations of each step in the above-mentioned intelligent toilet production management method have been referred to above. Figures 1 to 4 The description of the smart toilet production management system has been introduced in detail, and therefore, its repeated description will be omitted.
[0076] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of the modules is only a logical function division, and there may be other division methods in actual implementation.
[0077] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0078] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional modules.
[0079] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.
[0080] Therefore, no matter from which point of view, the embodiments should be regarded as illustrative and non-restrictive, and the scope of the present invention is limited by the appended claims rather than the above description, so it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention. Any attached figure mark in the claims should not be regarded as limiting the claims involved.
[0081] In addition, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices stated in the system claim can also be implemented by one unit or device through software or hardware. The second and other words are used to indicate names, but not to indicate any particular order.
[0082] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit of the technical solution of the present invention.
Claims
1. A smart toilet production management system, characterized in that: include: A toilet paper production image acquisition module, used to acquire toilet paper production images collected by a camera; A toilet paper production image extraction module, used to extract a toilet paper global association feature vector and a toilet paper attention feature vector from the toilet paper production image collected by the camera; The toilet paper quality judgment module is used to judge whether the toilet paper has quality problems based on the toilet paper global association feature vector and the toilet paper attention feature vector.
2. The intelligent toilet production management system according to claim 1, characterized in that: The toilet paper production image extraction module comprises: A toilet paper feature extraction unit, configured to extract features of interest from the toilet paper production image captured by the camera to obtain a global correlation feature vector of the toilet paper; The toilet paper attention feature extraction unit is used to extract the attention features of the toilet paper production image collected by the camera to obtain the toilet paper attention feature vector.
3. The intelligent toilet production management system according to claim 2 is characterized in that: The toilet paper interesting feature extraction unit comprises: A toilet paper feature of interest encoding subunit, used for encoding the toilet paper production image captured by the camera to obtain a plurality of toilet paper feature of interest vectors; The toilet paper convolutional encoding subunit is used to obtain the toilet paper global correlation feature vector by passing the multiple toilet paper interest feature vectors through a toilet paper interest convolutional neural network model as a feature encoder.
4. The intelligent toilet production management system according to claim 3, characterized in that: The toilet paper feature of interest encoding subunit is used for: Passing the toilet paper production image captured by the camera through a toilet paper object detection network to obtain a plurality of toilet paper interest regions; The multiple toilet paper interest regions are passed through a toilet paper interest linear embedding layer to obtain the multiple toilet paper interest feature vectors.
5. The intelligent toilet production management system according to claim 4, characterized in that: The toilet paper object detection network is an anchor window-based object detection network.
6. The intelligent toilet production management system according to claim 5, characterized in that: The toilet paper attention feature extraction unit is used to: Passing the toilet paper production image captured by the camera through a toilet paper convolutional neural network based on an attention mechanism to obtain a toilet paper attention feature map; Maximum pooling is performed on the toilet paper attention feature map to obtain the toilet paper attention feature vector.
7. The intelligent toilet production management system according to claim 6, characterized in that: The toilet paper attention feature map is subjected to maximum pooling to obtain the toilet paper attention feature vector, which is used for: Import necessary Python libraries, including the TensorFlow deep learning framework and the datetime date formatting library; Create a max_pooling function to perform the maximum pooling operation, which receives the toilet paper attention feature map as input and returns the toilet paper attention feature vector; Use the DateFormat mechanism to record and format the timestamps during the deployment process to track and record the execution time of each step; Call the defined max_pooling function to perform pooling on the toilet paper attention feature map and record the start and end time of the operation; The pooled toilet paper attention feature vector is saved in the database, and the saving timestamp is recorded.
8. The intelligent toilet production management system according to claim 7, characterized in that: The toilet paper quality judgment module comprises: A toilet paper feature fusion unit is used to fuse the toilet paper global correlation feature vector and the toilet paper attention feature vector based on the spatial distribution of data points to obtain a toilet paper quality classification feature vector; The toilet paper quality problem classification and judgment unit is used to pass the toilet paper quality classification feature vector through a classifier to obtain a classification result, and the classification result is used to judge whether there is a quality problem with the toilet paper.
9. The intelligent toilet production management system according to claim 8, characterized in that: The toilet paper feature fusion unit is used to: Constructing a toilet seat paper distance topology matrix between the toilet seat paper global association feature vector and the toilet seat paper attention feature vector; Constructing a toilet seat paper feature value granularity association matrix between the toilet seat paper global association feature vector and the toilet seat paper attention feature vector; Performing topological association modulation on the toilet paper eigenvalue granularity association matrix based on the toilet paper distance topological matrix to obtain a toilet paper topological modulation association matrix; Taking the toilet paper topology modulation association matrix as the perspective co-projection space, projecting the toilet paper global association feature vector and the toilet paper attention feature vector into the perspective co-projection space to obtain the perspective modulated toilet paper global association feature vector and the perspective modulated toilet paper attention feature vector; The toilet seat paper quality classification feature vector is obtained by fusing the perspective-modulated toilet seat paper global association feature vector and the perspective-modulated toilet seat paper attention feature vector.
10. A method for production management of intelligent toilets, characterized in that: include: Obtain toilet paper production images captured by a camera; Extracting a toilet paper global association feature vector and a toilet paper attention feature vector from the toilet paper production image captured by the camera; Based on the toilet paper global association feature vector and the toilet paper attention feature vector, it is determined whether the toilet paper has quality problems.
