Preparation method of seasoning

Abstract

The invention relates to the field of seasonings, in particular to a preparation method of a seasoning. The method comprises the following steps: crushing and screening rice to obtain rice powder, and crushing edible salt to obtain edible salt powder; crushing and screening the dried mushrooms to obtain mushroom powder, and crushing and screening the dried fruits and vegetables to obtain fruit and vegetable powder; conveying the wheat enzymolysis leavening, the wheat protein zymolyte, the rice powder, the edible salt powder, the mushroom powder, the fruit and vegetable powder, the tricholoma matsutake extract and the yeast extract to a mixing machine, uniformly mixing, granulating by a granulating machine, drying and sterilizing to obtain the tricholoma matsutake seasoning. Furthermore, the mixing state of the raw materials is monitored and judged in real time in an intelligent mode in the preparation process of the seasoning, so that the uniform mixing operation is timely decided and controlled, and the mixing uniformity of the raw materials is ensured.

Description

Technical Field

[0001] The present application relates to the field of seasonings, and in particular to a method for preparing seasonings. Background Art

[0002] Healthy condiments refer to supplementary foods that can enhance the color, aroma and taste of dishes, promote appetite and are beneficial to human health. Its main function is to improve the quality of dishes, meet the sensory needs of consumers, stimulate appetite and improve human health. It not only has a seasoning function, but also has a nutritional effect. It is a high-end compound condiment with nutritional value.

[0003] Chinese patent CN118077871A provides a matsutake seasoning and its preparation method and application, which can use matsutake / Agaricus blazei extract, wheat enzymatic fermentation, wheat protein enzymatic fermentation, yeast extract as the main freshness-enhancing raw materials, plant-type mushrooms and fruits and vegetables as auxiliary materials, and add health-care auxiliary raw materials of different substances. In this way, it is possible to provide a healthy functional seasoning without adding unhealthy raw materials such as monosodium glutamate, white sugar, any flavors, chemical synthetic spices, synthetic pigments, etc., which not only has fresh taste, but also has nutritional functions, and is a nutritious composite seasoning.

[0004] In the process of preparing seasonings, mixing raw materials is a very critical step because it is directly related to the quality and stability of the final product. If the mixing is insufficient or uneven, it may cause uneven distribution of certain ingredients in the seasoning, thereby affecting the taste and nutritional balance of the product. However, in the above-mentioned preparation method of matsutake seasoning, the raw materials are simply introduced into the mixer for mixing, and there is a lack of effective mixing monitoring means, which may cause fluctuations in product quality between batches, thereby affecting the quality and taste of the seasoning; or the mixing operation is continued after mixing, resulting in energy waste and reduced seasoning preparation efficiency.

[0005] Therefore, an optimized seasoning preparation protocol is desired. Summary of the invention

[0006] This application is made in view of the above problems. One object of this application is to provide a method for preparing seasoning.

[0007] The embodiment of the present application provides a method for preparing a seasoning, which comprises the following steps: crushing and sieving rice to obtain rice powder, and crushing edible salt to obtain edible salt powder; crushing and sieving dried mushrooms to obtain mushroom powder, and crushing and sieving dried fruits and vegetables to obtain fruit and vegetable powder; conveying wheat enzymatic fermentation product, wheat protein hydrolysate, the rice powder, the edible salt powder, the mushroom powder, the fruit and vegetable powder, pine mushroom extract and yeast extract to a mixer for mixing, and then granulating, drying and sterilizing in a granulator to obtain pine mushroom seasoning;

[0008] The enzymatic fermentation product of wheat, the protein hydrolyzate of wheat, the rice powder, the edible salt powder, the mushroom powder, the fruit and vegetable powder, the matsutake extract and the yeast extract are transported to a mixer for mixing, comprising:

[0009] Acquire real-time raw material mixing state distribution images collected by a camera;

[0010] Extracting texture features based on grid granularity on the real-time raw material mixing state distribution image to obtain a set of grid granularity real-time raw material mixing state texture feature vectors;

[0011] Performing feature saliency aggregation processing on the set of real-time raw material mixing state texture feature vectors of the grid granularity to obtain a global aggregate representation vector of the raw material mixing state texture;

[0012] Performing a comparative analysis based on texture semantic difference measurement on each grid-size real-time raw material mixed state texture feature vector in the set of the grid-size real-time raw material mixed state texture feature vectors and the raw material mixed state texture global aggregation representation vector to obtain a comparison result;

[0013] Based on the comparison result, it is determined whether to stop the mixing operation.

[0014] For example, according to the method for preparing seasoning of an embodiment of the present application, the real-time raw material mixing state distribution image is subjected to texture feature extraction based on grid granularity to obtain a set of real-time raw material mixing state texture feature vectors of grid granularity, including:

[0015] Grid-cutting the real-time raw material mixing state distribution image to obtain a set of real-time raw material mixing state distribution images with grid granularity;

[0016] Each grid-size real-time raw material mixing state distribution image in the set of grid-size real-time raw material mixing state distribution images is input into a texture feature extractor based on a dilated convolutional layer to obtain a set of grid-size real-time raw material mixing state texture feature vectors.

[0017] For example, according to the method for preparing seasoning of an embodiment of the present application, wherein the set of grid-granularity real-time raw material mixing state texture feature vectors is subjected to feature significance aggregation processing to obtain the raw material mixing state texture global aggregation representation vector, comprising: inputting the set of grid-granularity real-time raw material mixing state texture feature vectors into a feature significance aggregation network based on clustering contribution field analysis to obtain the raw material mixing state texture global aggregation representation vector.

[0018] For example, according to the method for preparing seasoning of an embodiment of the present application, the set of the real-time raw material mixing state texture feature vectors of the grid granularity is input into a feature significant aggregation network based on cluster contribution field analysis to obtain the global aggregation representation vector of the raw material mixing state texture, including:

[0019] Performing cluster analysis on the set of the real-time raw material mixing state texture feature vectors at the grid granularity to obtain a self-supervised cluster representation vector of the real-time raw material mixing state texture feature at the grid granularity;

[0020] Calculating the implicit clustering contribution factor of each grid-size real-time raw material mixed state texture feature vector in the set of the grid-size real-time raw material mixed state texture feature vectors relative to the grid-size real-time raw material mixed state texture feature self-supervised clustering representation vector to obtain a sequence of grid-size real-time raw material mixed state texture feature implicit clustering contribution factors;

[0021] Vectorizing and arranging the sequence of the implicit clustering contribution factors of the real-time raw material mixing state texture characteristics of the grid granularity to obtain the clustering contribution field distribution vector of the real-time raw material mixing state texture characteristics of the grid granularity;

[0022] Input the grid-granularity real-time raw material mixed state texture feature cluster contribution field distribution vector into the cluster contribution field explicit modeling module based on the self-attention mechanism to obtain the grid-granularity real-time raw material mixed state texture feature cluster contribution field modulation weight vector;

[0023] Taking each eigenvalue in the grid-granularity real-time raw material mixed state texture feature clustering contribution field modulation weight vector as weight, the position-weighted sum of the set of the grid-granularity real-time raw material mixed state texture feature vectors is calculated to obtain the raw material mixed state texture global aggregation representation vector.

[0024] For example, according to the method for preparing seasoning of an embodiment of the present application, a cluster analysis is performed on the set of grid-granularity real-time raw material mixing state texture feature vectors to obtain a self-supervised clustering representation vector of grid-granularity real-time raw material mixing state texture features, including: calculating the mean vector of the set of grid-granularity real-time raw material mixing state texture feature vectors to obtain the self-supervised clustering representation vector of grid-granularity real-time raw material mixing state texture features.

[0025] For example, according to the method for preparing seasoning of an embodiment of the present application, the implicit clustering contribution factor of each grid-granularity real-time raw material mixed state texture feature vector in the set of the grid-granularity real-time raw material mixed state texture feature vector relative to the grid-granularity real-time raw material mixed state texture feature self-supervised clustering representation vector is calculated to obtain a sequence of grid-granularity real-time raw material mixed state texture feature implicit clustering contribution factors, including:

[0026] Calculating the absolute value of the positional division between the real-time raw material mixing state texture feature vector of the grid granularity and the real-time raw material mixing state texture feature self-supervised clustering representation vector of the grid granularity to obtain the real-time raw material mixing state texture grid granularity self-supervised clustering interactive correlation feature vector;

[0027] Calculate the logarithmic function value of each position feature value in the real-time raw material mixing state texture grid granularity self-supervised clustering interactive correlation feature vector with base 2 to obtain the real-time raw material mixing state texture grid granularity self-supervised clustering interactive logarithmic representation vector;

[0028] Taking the eigenvalues ​​of each position in the real-time raw material mixing state texture grid granularity self-supervised clustering interactive logarithm representation vector as weighting coefficients, calculating the weighted sum of the eigenvalues ​​of each position in the real-time raw material mixing state texture feature vector of the grid granularity to obtain the grid granularity real-time raw material mixing state texture feature clustering contribution representation factor;

[0029] The exponential function value of the grid granularity real-time raw material mixed state texture feature clustering contribution representation factor is calculated with the natural constant e as the base to obtain the grid granularity real-time raw material mixed state texture feature implicit clustering contribution factor.

[0030] For example, according to the method for preparing seasoning of an embodiment of the present application, the grid-granularity real-time raw material mixed state texture feature cluster contribution field distribution vector is input into the cluster contribution field explicit modeling module based on the self-attention mechanism to obtain the grid-granularity real-time raw material mixed state texture feature cluster contribution field modulation weight vector, including:

[0031] Calculate the matrix multiplication between the grid-granularity real-time raw material mixed state texture feature cluster contribution field distribution vector and the query weight matrix, the key weight matrix and the value weight matrix respectively to obtain the real-time raw material mixed state texture cluster contribution field distribution query vector, the real-time raw material mixed state texture cluster contribution field distribution key vector and the real-time raw material mixed state texture cluster contribution field distribution value vector;

[0032] After calculating the multiplication between the real-time raw material mixed state texture cluster contribution field distribution query vector and the transposed vector of the real-time raw material mixed state texture cluster contribution field distribution key vector, the obtained matrix is ​​divided by the square root of the length of the real-time raw material mixed state texture cluster contribution field distribution key vector by position to obtain a query-key real-time raw material mixed state texture cluster contribution field semantic interaction representation matrix;

[0033] The query-key real-time raw material mixed state texture clustering contribution field semantic interaction representation matrix is ​​normalized by using a Softmax function to obtain a normalized query-key real-time raw material mixed state texture clustering contribution field semantic interaction representation matrix;

[0034] The multiplication between the real-time raw material mixed state texture cluster contribution field distribution value vector and the query-key real-time raw material mixed state texture cluster contribution field semantic interaction representation matrix is ​​calculated to obtain the grid-granularity real-time raw material mixed state texture feature cluster contribution field modulation weight vector.

[0035] For example, according to the method for preparing seasoning of an embodiment of the present application, each grid granularity real-time raw material mixing state texture feature vector in the set of the grid granularity real-time raw material mixing state texture feature vector and the raw material mixing state texture global aggregation representation vector are compared and analyzed based on texture semantic difference measurement to obtain a comparison result, including:

[0036] Calculating a texture semantic difference metric coefficient between any one of the set of the grid-size real-time raw material mixed state texture feature vectors and the raw material mixed state texture global aggregation representation vector to obtain a set of texture semantic difference metric coefficients;

[0037] A maximum value in the set of texture semantic difference measurement coefficients is extracted, and the maximum value is compared with a preset threshold to obtain the comparison result.

[0038] For example, according to the method for preparing seasoning of an embodiment of the present application, the texture semantic difference metric coefficient between any one of the set of the grid granularity real-time raw material mixing state texture feature vectors and the raw material mixing state texture global aggregation representation vector is calculated to obtain a set of texture semantic difference metric coefficients, including:

[0039] Calculating the square of the one-norm of the position difference between the real-time raw material mixed state texture feature vector of the grid granularity and the global aggregation representation vector of the raw material mixed state texture to obtain a one-norm representation factor of the semantic difference of the real-time raw material mixed state texture of the grid granularity;

[0040] After calculating the square of the one-norm of the real-time raw material mixing state texture feature vector of the grid granularity to obtain a texture norm representation value, subtract the texture norm representation value from a constant 1 to obtain a one-norm representation factor of the real-time raw material mixing state texture feature of the grid granularity;

[0041] After calculating the square of the one-norm of the raw material mixed state texture global aggregation representation vector to obtain a texture global aggregation norm representation value, subtract the texture global aggregation norm representation value from a constant 1 to obtain a raw material mixed state texture global aggregation one-norm representation factor;

[0042] After multiplying the grid-size real-time raw material mixed state texture feature-norm representation factor and the raw material mixed state texture global aggregation-norm representation factor to obtain a norm interaction factor, the grid-size real-time raw material mixed state texture semantic difference-norm representation factor is divided by the norm interaction factor to obtain a texture semantic interaction factor;

[0043] The texture semantic interaction factor is multiplied by a constant 2 and then added to a constant 1 to obtain a texture semantic difference coefficient;

[0044] The texture semantic difference coefficient is activated using an arccosh function to obtain a texture semantic difference measurement coefficient.

[0045] According to the method for preparing seasonings of the embodiments of the present application, the mixing state of raw materials is monitored and judged in real time in an intelligent manner during the preparation of the seasonings, so as to make timely decisions and control the mixing operation to ensure the uniformity of the raw material mixing. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] 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.

[0047] Figure 1 A flow chart showing a method for preparing seasoning in an embodiment of the present application is shown;

[0048] Figure 2 A flowchart of sub-step S530 of the method for preparing seasoning in an embodiment of the present application is shown;

[0049] Figure 3 A flow chart showing sub-step S533 of the method for preparing seasoning in an embodiment of the present application is shown;

[0050] Figure 4 A schematic diagram showing the structure of a seasoning preparation system in an embodiment of the present application is shown; and

[0051] Figure 5The application scenario diagram of the seasoning preparation method in the embodiment of the present application is shown. DETAILED DESCRIPTION

[0052] 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.

[0053] 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.

[0054] 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.

[0055] 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.

[0056] 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.

[0057] The following is a non-limiting description of the method for preparing a seasoning 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.

[0058] In view of the above technical problems, in the technical solution of the present application, a method for preparing seasoning is proposed, such as Figure 1 As shown, it includes the steps of: S510, crushing and sieving rice to obtain rice powder, and crushing edible salt to obtain edible salt powder; S520, crushing and sieving dried mushrooms to obtain mushroom powder, and crushing and sieving dried fruits and vegetables to obtain fruit and vegetable powder; S530, conveying wheat enzymatic fermentation product, wheat protein hydrolysate, the rice powder, the edible salt powder, the mushroom powder, the fruit and vegetable powder, matsutake extract and yeast extract to a mixer for mixing, and then granulating, drying and sterilizing in a granulator to obtain matsutake seasoning.

[0059] It should be understood that the matsutake seasoning provided by the present application does not add monosodium glutamate, sucrose, flavors and spices, artificial colors, preservatives, and the salt content is only about 20%. In addition, due to the addition of matsutake and hericium, the mushroom flavor is prominent, and various fruits, vegetables and seafood raw materials that are beneficial to children's growth are added. Specifically, the addition ratio of mushroom powder is about 22-25%, including matsutake, hericium, shiitake and pleurotus eryngii. Among them, matsutake can improve immunity, anti-tumor, anti-oxidation, protect the liver, promote intestinal health, lower blood sugar, beautify and nourish the skin, and supplement nutrition; among them, hericium can protect the stomach, enhance immunity, lower blood lipids, anti-tumor, anti-aging, promote digestion, calm the nerves, and protect the liver. The addition ratio of vegetables is about 22-25%, including apples, carrots, spinach, tomatoes, and pumpkins. These vegetables are rich in vitamins, minerals, lower cholesterol, control blood sugar, anti-oxidation, promote brain development, and prevent constipation. The addition ratio of rice flour is about 18-21%, which specifically includes rice flour. Rice flour is rich in nutrients and easier for children to absorb. The addition ratio of sea salt is about 16-20%. Sea salt does not contain iodine and is rich in minerals. The addition ratio of yeast extract is about 5-7%. The addition ratio of fish sauce is about 3-5%, which can enhance the flavor and supplement nutrition. The addition ratio of shrimp powder is about 1-3%. Shrimp powder can supplement calcium, promote children's growth and development, and improve appetite. The addition ratio of Chinese yam is about 2-3%. Chinese yam is rich in microorganisms, minerals, and dietary fiber, which can enhance immunity. The addition ratio of walnuts is about 1-3%. Walnuts can improve intelligence and brain health and protect cardiovascular and cerebrovascular systems. The addition ratio of monk fruit is about 0.1-1%. Monk fruit can clear heat and moisten the lungs, improve immunity, and regulate the stomach.

[0060] In particular, in the process of preparing the above-mentioned seasoning, in the process of conveying the wheat enzymatic fermentation product, the wheat protein hydrolysate, the rice powder, the edible salt powder, the mushroom powder, the fruit and vegetable powder, the matsutake extract and the yeast extract to the mixer for mixing, the real-time raw material mixing state distribution image is collected by real-time monitoring through the camera, and the image processing and analysis algorithm based on artificial intelligence and machine vision is introduced at the back end to analyze the real-time raw material mixing state distribution image, so as to capture the local distribution characteristics and global aggregation representation characteristics of the real-time state of the raw material mixing in the mixing process, and the semantic difference between the local regional characteristics of each mixing state and the global aggregation representation characteristics is measured to determine whether the raw material is mixed, so as to determine whether to stop the mixing operation. In this way, the raw material mixing state can be monitored and judged in real time in an intelligent way during the preparation of the seasoning, so as to make timely decisions to control the mixing operation, ensure the uniformity of the raw material mixing, and ensure the quality stability and taste of the final seasoning product, and reduce energy waste.

[0061] Accordingly, if Figure 2 As shown, in step S530, the wheat enzymatic fermentation product, the wheat protein hydrolysate, the rice powder, the edible salt powder, the mushroom powder, the fruit and vegetable powder, the matsutake extract and the yeast extract are conveyed to a mixer for mixing, including: S531, obtaining a real-time raw material mixing state distribution image collected by a camera; S532, performing a grid-granularity-based texture feature extraction on the real-time raw material mixing state distribution image to obtain a set of grid-granularity real-time raw material mixing state texture feature vectors; S533, performing feature significance aggregation processing on the set of grid-granularity real-time raw material mixing state texture feature vectors to obtain a raw material mixing state texture global aggregation representation vector; S534, performing a comparative analysis based on texture semantic difference measurement on each grid-granularity real-time raw material mixing state texture feature vector in the set of grid-granularity real-time raw material mixing state texture feature vectors and the raw material mixing state texture global aggregation representation vector to obtain a comparison result; S535, based on the comparison result, determining whether to stop the mixing operation.

[0062] Specifically, in the technical solution of the present application, in the process of conveying the wheat enzymatic fermentation product, the wheat protein hydrolysate, the rice powder, the edible salt powder, the mushroom powder, the fruit and vegetable powder, the matsutake extract and the yeast extract to the mixer for mixing, first, a real-time raw material mixing state distribution image captured by a camera is obtained. Then, in order to more finely monitor and analyze the mixing state of the raw materials, the real-time raw material mixing state distribution image is further grid-cut to obtain a set of real-time raw material mixing state distribution images of grid granularity. By dividing the entire real-time raw material mixing state distribution image into multiple small grid areas, the local regional distribution of different raw materials during the mixing process can be captured more accurately. This segmentation method can help identify which areas have more even mixing of raw materials and which areas may have uneven mixing.

[0063] Next, each grid-size real-time raw material mixing state distribution image in the set of grid-size real-time raw material mixing state distribution images is input into a texture feature extractor based on a dilated convolutional layer for feature mining, so as to respectively extract local area texture feature information about the raw material mixing state in each grid-size real-time raw material mixing state distribution image, thereby obtaining a set of grid-size real-time raw material mixing state texture feature vectors.

[0064] Correspondingly, in step S532, the real-time raw material mixing state distribution image is subjected to texture feature extraction based on grid granularity to obtain a set of real-time raw material mixing state texture feature vectors at grid granularity, including: grid-cutting the real-time raw material mixing state distribution image to obtain a set of real-time raw material mixing state distribution images at grid granularity; and inputting each grid granularity real-time raw material mixing state distribution image in the set of real-time raw material mixing state distribution images at grid granularity into a texture feature extractor based on a dilated convolutional layer to obtain a set of real-time raw material mixing state texture feature vectors at grid granularity.

[0065] It should be understood that since each grid-size real-time raw material mixing state texture feature vector in the set of the grid-size real-time raw material mixing state texture feature vectors respectively contains information about the mixing state texture feature of the local area during the raw material mixing process, but these features are not all important for the subsequent global area raw material mixing state detection and mixing operation control tasks. Therefore, in order to be able to identify which local area texture features are more significant in the entire mixing state monitoring and mixing operation control tasks, that is, which features are most important for describing the overall mixing state, thereby facilitating a comprehensive analysis of the overall texture state features of the entire mixture, so as to more effectively analyze and evaluate the mixing state of the raw materials during the mixing process, in the technical solution of the present application, the set of the grid-size real-time raw material mixing state texture feature vectors is further input into a feature significant aggregation network based on cluster contribution field analysis to obtain a global aggregation representation vector of the raw material mixing state texture. It is worth mentioning that the feature significant aggregation network based on clustering contribution field analysis is a feature aggregation technology based on simulated field characteristic analysis, which aims to use the simulated field analysis theory to perform field-based explicit modeling of the clustering contribution of the real-time raw material mixing state texture feature vectors of each grid granularity to improve the accuracy of feature aggregation analysis, thereby improving the performance and generalization ability of the model.

[0066] Specifically, the feature significant aggregation network based on cluster contribution field analysis first uses an unsupervised learning method to perform cluster analysis on the set of grid-size real-time raw material mixed state texture feature vectors, and automatically discovers the grid-size real-time raw material mixed state texture context-related feature patterns and structures between each local area in the set to generate a grid-size real-time raw material mixed state texture feature self-supervised cluster representation vector. It should be understood that the inherent structure of the grid-size context-related features of each real-time raw material mixed state texture feature can be revealed through self-supervised clustering, which helps the model learn meaningful feature representations without explicit labels, and enhances the model's understanding of the grid-size context-related feature distribution of the real-time raw material mixed state texture feature. Furthermore, by calculating the implicit clustering contribution factor between the grid-size real-time raw material mixed state texture feature vector and the grid-size real-time raw material mixed state texture feature self-supervised cluster representation vector, it is used to reflect the contribution of each grid-size real-time raw material mixed state texture feature vector to its cluster. Accordingly, the calculation of implicit clustering contribution factors helps to identify and quantify the role of each grid-size real-time raw material mixed state texture feature vector in the clustering process, providing a basis for subsequent feature selection and weight allocation, so that the model can pay more attention to the features that contribute significantly to clustering. That is, it helps to identify which grid-size mixed state texture features are most significant in the entire mixture state representation, that is, which features are most important for describing the overall mixed state. Next, the sequence of implicit clustering contribution factors of the real-time raw material mixed state texture features of the grid-size is converted into a distribution vector of the clustering contribution field of the real-time raw material mixed state texture features of the grid-size, that is, the sequence of implicit clustering contribution factors of the real-time raw material mixed state texture features of the grid-size is integrated into the vector field by data integration to construct the clustering contribution field. Then, the self-attention mechanism is used to explicitly model the distribution vector of the clustering contribution field of the real-time raw material mixed state texture features of the grid-size to generate a modulation weight vector of the clustering contribution field of the real-time raw material mixed state texture features of the grid-size. That is, the self-attention mechanism is used to explicitly model the clustering contribution of the implicit clustering contribution factors of the real-time raw material mixing state texture features of each grid granularity in the clustering contribution field based on the global analysis of the field autocorrelation. This can strengthen the attention on the feature vectors with large clustering contributions, while suppressing the local texture feature expression of the raw material mixing state in unimportant grid areas, thereby improving the quality of feature representation and the accuracy of subsequent mixing detection tasks.Finally, the set of the grid-granularity real-time raw material mixing state texture feature vectors is weightedly fused using the grid-granularity real-time raw material mixing state texture feature clustering contribution field modulation weight vector to generate a spatial domain real-time raw material mixing state texture significant aggregation representation vector. The spatial domain real-time raw material mixing state texture significant aggregation representation vector provides a comprehensive feature representation, which not only contains the grid-granularity context-associated feature information of the original real-time raw material mixing state texture information, but also integrates the knowledge learned in the clustering process, so that the model can more accurately capture the key grid areas and texture features of the real-time raw material mixing state when processing complex tasks, so as to more accurately obtain the global key mixing state feature representation of the entire mixing state, thereby improving the performance and generalization ability of the model.

[0067] Correspondingly, in step S533, the set of grid-sized real-time raw material mixture state texture feature vectors is subjected to feature significance aggregation processing to obtain a global aggregate representation vector of the raw material mixture state texture, including: inputting the set of grid-sized real-time raw material mixture state texture feature vectors into a feature significance aggregation network based on clustering contribution field analysis to obtain a global aggregate representation vector of the raw material mixture state texture.

[0068] Specifically, Figure 3 As shown, the set of the grid-granularity real-time raw material mixed state texture feature vectors is input into a feature significant aggregation network based on cluster contribution field analysis to obtain the global aggregation representation vector of the raw material mixed state texture, including: S5331, clustering analysis is performed on the set of the grid-granularity real-time raw material mixed state texture feature vectors to obtain a grid-granularity real-time raw material mixed state texture feature self-supervised cluster representation vector; S5332, the implicit cluster contribution factor of each grid-granularity real-time raw material mixed state texture feature vector in the set of the grid-granularity real-time raw material mixed state texture feature vectors relative to the grid-granularity real-time raw material mixed state texture feature self-supervised cluster representation vector is calculated to obtain a sequence of grid-granularity real-time raw material mixed state texture feature implicit cluster contribution factors; S5333, vectorize the sequence of implicit clustering contribution factors of the real-time raw material mixed state texture features at the grid granularity to obtain the clustering contribution field distribution vector of the real-time raw material mixed state texture features at the grid granularity; S5334, input the clustering contribution field distribution vector of the real-time raw material mixed state texture features at the grid granularity into the clustering contribution field explicit modeling module based on the self-attention mechanism to obtain the grid granularity real-time raw material mixed state texture features clustering contribution field modulation weight vector; S5335, use each eigenvalue in the clustering contribution field modulation weight vector of the real-time raw material mixed state texture features at the grid granularity as a weight, calculate the position-weighted sum of the set of the real-time raw material mixed state texture feature vectors at the grid granularity to obtain the global aggregation representation vector of the raw material mixed state texture.

[0069] Among them, in step S5331, a cluster analysis is performed on the set of grid-granularity real-time raw material mixing state texture feature vectors to obtain a self-supervised clustering representation vector of grid-granularity real-time raw material mixing state texture features, including: calculating the mean vector of the set of grid-granularity real-time raw material mixing state texture feature vectors to obtain a self-supervised clustering representation vector of grid-granularity real-time raw material mixing state texture features.

[0070] Wherein, in step S5332, the implicit clustering contribution factor of each grid-granularity real-time raw material mixed state texture feature vector in the set of the grid-granularity real-time raw material mixed state texture feature vectors relative to the grid-granularity real-time raw material mixed state texture feature self-supervised clustering representation vector is calculated to obtain a sequence of grid-granularity real-time raw material mixed state texture feature implicit clustering contribution factors, including: calculating the absolute value of the positional division between the grid-granularity real-time raw material mixed state texture feature vector and the grid-granularity real-time raw material mixed state texture feature self-supervised clustering representation vector to obtain the real-time raw material mixed state texture grid-granularity self-supervised clustering interactive correlation feature vector; calculating the real-time raw material mixed state texture with base 2 as the real-time raw material mixed state texture feature vector; The logarithmic function value of each position feature value in the grid granularity self-supervised clustering interactive correlation feature vector is calculated to obtain the real-time raw material mixing state texture grid granularity self-supervised clustering interactive logarithmic representation vector; the weighted sum of the feature values ​​at each position in the grid granularity real-time raw material mixing state texture feature vector is calculated using the feature values ​​at each position in the real-time raw material mixing state texture grid granularity self-supervised clustering interactive logarithmic representation vector as the weighting coefficient to obtain the grid granularity real-time raw material mixing state texture feature clustering contribution representation factor; the exponential function value of the grid granularity real-time raw material mixing state texture feature clustering contribution representation factor is calculated with the natural constant e as the base to obtain the grid granularity real-time raw material mixing state texture feature implicit clustering contribution factor.

[0071] Wherein, in step S5334, the grid-granularity real-time raw material mixed state texture feature cluster contribution field distribution vector is input into the cluster contribution field explicit modeling module based on the self-attention mechanism to obtain the grid-granularity real-time raw material mixed state texture feature cluster contribution field modulation weight vector, including: calculating the matrix multiplication between the grid-granularity real-time raw material mixed state texture feature cluster contribution field distribution vector and the query weight matrix, the key weight matrix and the value weight matrix respectively to obtain the real-time raw material mixed state texture cluster contribution field distribution query vector, the real-time raw material mixed state texture cluster contribution field distribution key vector and the real-time raw material mixed state texture cluster contribution field distribution value vector; calculating the real-time raw material mixed state texture cluster contribution field distribution query vector and the real-time raw material mixed state texture cluster contribution field distribution key vector. After multiplying the transposed vectors of the field distribution key vector, the obtained matrix is ​​divided by the square root of the length of the real-time raw material mixed state texture clustering contribution field distribution key vector by position to obtain the query-key real-time raw material mixed state texture clustering contribution field semantic interaction representation matrix; the query-key real-time raw material mixed state texture clustering contribution field semantic interaction representation matrix is ​​normalized by using the Softmax function to obtain the normalized query-key real-time raw material mixed state texture clustering contribution field semantic interaction representation matrix; the multiplication between the real-time raw material mixed state texture clustering contribution field distribution value vector and the query-key real-time raw material mixed state texture clustering contribution field semantic interaction representation matrix is ​​calculated to obtain the grid-granularity real-time raw material mixed state texture feature clustering contribution field modulation weight vector.

[0072] In a specific example, the set of the real-time raw material mixed state texture feature vectors of the grid granularity is input into a feature significant aggregation network based on cluster contribution field analysis to obtain the global aggregate representation vector of the raw material mixed state texture, including: inputting the set of the real-time raw material mixed state texture feature vectors of the grid granularity into the feature significant aggregation network based on cluster contribution field analysis and processing them with the following feature significant aggregation formula to obtain the global aggregate representation vector of the raw material mixed state texture; wherein the feature significant aggregation formula is:

[0073] X={x 1 ,x 2 ,...,x k ,...,x n}

[0074]

[0075]

[0076] v d ={d 1 ;d 2 ;...;d j;...;d n}

[0077]

[0078] v q =W q v d

[0079] v k =W k v d

[0080] v v =W b v d

[0081]

[0082] Wherein, X is the set of texture feature vectors of the real-time raw material mixing state of the grid granularity, x 1 , x 2 , x k , x n are the first, second, kth and nth grid-size real-time raw material mixing state texture feature vectors in the set of the grid-size real-time raw material mixing state texture feature vectors, respectively, x i is the ith grid-size real-time raw material mixed state texture feature vector in the set of grid-size real-time raw material mixed state texture feature vectors, N is the number of feature vectors in the set of grid-size real-time raw material mixed state texture feature vectors, x c is the grid-scale real-time raw material mixing state texture feature self-supervised clustering representation vector, x j is the jth grid-size real-time raw material mixed state texture feature vector in the set of grid-size real-time raw material mixed state texture feature vectors, is the feature value of the kth position in the texture feature vector of the real-time raw material mixing state at the jth grid granularity, is the eigenvalue of the kth position in the self-supervised clustering representation vector of the real-time raw material mixing state texture feature of the grid granularity, log represents the logarithmic function operation with base 2, exp represents the natural exponential function value, d j is the implicit clustering contribution factor of the real-time raw material mixing state texture feature of the grid granularity corresponding to the real-time raw material mixing state texture feature vector of the j-th grid granularity, v d is the distribution vector of the clustering contribution field of the real-time raw material mixing state texture feature of the grid granularity obtained by quantizing the implicit clustering contribution factors of the real-time raw material mixing state texture feature of multiple grid granularity, W q , W k and W bare the query weight matrix, key weight matrix and value weight matrix respectively, v q 、v k and v v are respectively a real-time raw material mixed state texture clustering contribution field distribution query vector, a real-time raw material mixed state texture clustering contribution field distribution key vector and a real-time raw material mixed state texture clustering contribution field distribution value vector, L is the length of the real-time raw material mixed state texture clustering contribution field distribution key vector, Softmax(·) is the Softmax function, is vector multiplication, v w The contribution field modulation weight vector for real-time raw material mixing state texture feature clustering at the grid granularity, x f is a global aggregate representation vector of the raw material mixed state texture.

[0083] Further, in order to quantitatively evaluate the uniformity of raw material mixing and thus perform mixing operation control, in the technical solution of the present application, the texture semantic difference measurement coefficient between any one of the grid granularity real-time raw material mixing state texture feature vectors in the set of the grid granularity real-time raw material mixing state texture feature vectors and the global aggregate representation vector of the raw material mixing state texture is further calculated to obtain a set of texture semantic difference measurement coefficients. It should be understood that since each of the grid granularity real-time raw material mixing state texture feature vectors reflects the raw material mixing distribution and texture feature information in the grid area, and the global aggregate representation vector of the raw material mixing state texture reflects the global aggregate representation information of the grid granularity raw material mixing state texture features under the overall mixing state. By comparing the difference between the real-time raw material mixing state texture features of each grid area and the global aggregate representation of the raw material mixing state texture features, the degree of difference between the raw material mixing distribution and the overall mixing state in the corresponding grid area can be measured. If the difference between the mixing state of a certain grid area and the global mixed state aggregate representation is large, it means that the mixing state of the area may not be ideal, and it may take longer to mix to improve its uniformity.

[0084] In particular, in the technical solution of the present application, the reason for using the hyperbolic space distance metric rather than the traditional Euclidean distance or other metric methods is that the hyperbolic space can more naturally represent the situation with hierarchical structure or high-dimensional sparse data. Since the hyperbolic space grows faster as it moves away from the origin, this metric can better maintain the hierarchical structure of the data and is more suitable for processing data distributions with exponential growth characteristics, which is crucial for capturing the relationship between features, especially when dealing with complex or nonlinear relationships. Specifically, when calculating the texture semantic difference metric coefficient between any one of the grid granularity real-time raw material mixing state texture feature vectors in the set of the grid granularity real-time raw material mixing state texture feature vectors and the raw material mixing state texture global aggregation representation vector, the hyperbolic space distance metric can help more accurately measure the difference between the raw material distribution in different grid areas and the overall mixing state. This metric can not only effectively reflect the similarity or difference between the feature vectors of each grid area and the global aggregation representation vector, but also better adapt to the hierarchical structure and distribution characteristics of the data set, thereby more accurately evaluating the uniformity of raw material mixing. In this way, the quality control of the mixing process can be ensured, thereby ensuring the quality of the final product.

[0085] Correspondingly, in step S534, a comparative analysis based on texture semantic difference measurement is performed on each grid-granularity real-time raw material mixed state texture feature vector in the set of grid-granularity real-time raw material mixed state texture feature vectors and the raw material mixed state texture global aggregation representation vector to obtain a comparison result, including: calculating the texture semantic difference measurement coefficient between any grid-granularity real-time raw material mixed state texture feature vector in the set of grid-granularity real-time raw material mixed state texture feature vectors and the raw material mixed state texture global aggregation representation vector to obtain a set of texture semantic difference measurement coefficients; extracting the maximum value in the set of texture semantic difference measurement coefficients, and comparing the maximum value with a preset threshold to obtain the comparison result.

[0086] Specifically, the texture semantic difference measurement coefficient between any one of the set of the grid-granularity real-time raw material mixed state texture feature vectors and the raw material mixed state texture global aggregation representation vector is calculated to obtain a set of texture semantic difference measurement coefficients, including: calculating the square of the one-norm of the position difference between the grid-granularity real-time raw material mixed state texture feature vector and the raw material mixed state texture global aggregation representation vector to obtain the grid-granularity real-time raw material mixed state texture semantic difference one-norm representation factor; calculating the square of the one-norm of the grid-granularity real-time raw material mixed state texture feature vector to obtain the texture norm representation value, and then subtracting the texture norm representation value from the constant 1 to obtain the grid-granularity real-time raw material mixed state texture feature one-norm representation factor; calculating the The square of the norm of the raw material mixed state texture global aggregation representation vector is obtained to obtain the texture global aggregation norm representation value, and then the texture global aggregation norm representation value is subtracted from the constant 1 to obtain the raw material mixed state texture global aggregation norm representation factor; the grid granularity real-time raw material mixed state texture feature norm representation factor and the raw material mixed state texture global aggregation norm representation factor are multiplied to obtain the norm interaction factor, and then the grid granularity real-time raw material mixed state texture semantic difference norm representation factor is divided by the norm interaction factor to obtain the texture semantic interaction factor; the texture semantic interaction factor is multiplied by the constant 2 and then added to the constant 1 to obtain the texture semantic difference coefficient; the texture semantic difference coefficient is activated using the arccosh function to obtain the texture semantic difference measurement coefficient.

[0087] In a specific example, calculating the texture semantic difference measurement coefficient between any one of the set of the grid-size real-time raw material mixed state texture feature vectors and the global aggregate representation vector of the raw material mixed state texture to obtain a set of texture semantic difference measurement coefficients, including: calculating the texture semantic difference measurement coefficient between any one of the set of the grid-size real-time raw material mixed state texture feature vectors and the global aggregate representation vector of the raw material mixed state texture using the following semantic difference measurement formula to obtain the set of texture semantic difference measurement coefficients; wherein the semantic difference measurement formula is:

[0088]

[0089] Among them, v 1i is the i-th grid-size real-time raw material mixed state texture feature vector in the set of the grid-size real-time raw material mixed state texture feature vectors, v 2is the global aggregate representation vector of the raw material mixed state texture, · is the vector norm, arccosh(·) is the arccosh function, d P (v 1i ,v 2 ) are each texture semantic difference metric coefficient in the set of texture semantic difference metric coefficients.

[0090] Furthermore, in order to find the most uneven grid area in the mixing process, so as to judge whether the raw materials are mixed evenly, in the technical solution of the present application, the maximum value in the set of texture semantic difference measurement coefficients is further extracted, and the maximum value is compared with the preset threshold to obtain a comparison result, and based on the comparison result, it is determined whether to stop the mixing operation. It should be understood that even if most areas have been mixed evenly, as long as one area does not meet the standard, it means that the overall mixing is not complete. By comparing this maximum value with the preset threshold, it can be determined whether to continue the mixing operation. If the maximum value exceeds the preset threshold, it means that the mixing is not uniform enough, and it is necessary to continue mixing until the mixing degree of all areas is within the allowable range; on the contrary, if the maximum value is less than or equal to the preset threshold, it means that the mixing is uniform enough and the mixing operation can be stopped. In this way, the mixing state of the raw materials can be monitored and judged in real time in an intelligent manner during the preparation of the seasoning, so as to make timely decisions to control the mixing operation, ensure the uniformity of the raw material mixing, and ensure the quality stability and taste of the final seasoning product, and reduce energy waste.

[0091] In the technical solution of the present application, each grid-size real-time raw material mixed state texture feature vector in the set of grid-size real-time raw material mixed state texture feature vectors respectively represents the texture features of the local area determined by the grid-size segmentation of the real-time raw material mixed state distribution image. However, considering that the image content distribution of each local area in the real-time raw material mixed state distribution image has other interference components in addition to texture differences, when the set of grid-size real-time raw material mixed state texture feature vectors is input into the feature significant aggregation network based on cluster contribution field analysis for texture clustering analysis, the obtained raw material mixed state texture global aggregation representation vector has a diversified set expression distribution of image semantic aggregation features. This will result in that when calculating the texture semantic difference measurement coefficient between any grid-size real-time raw material mixed state texture feature vector in the set of grid-size real-time raw material mixed state texture feature vectors and the raw material mixed state texture global aggregation representation vector, the calculation accuracy of the texture semantic difference measurement coefficient is affected due to the difference in feature distribution diversity between the grid-size real-time raw material mixed state texture feature vector and the raw material mixed state texture global aggregation representation vector.

[0092] Therefore, before calculating the texture semantic difference measurement coefficient, the feature distribution of the raw mixed state texture global aggregation representation vector is adjusted, and the process includes: determining the raw mixed state texture global aggregation representation maximum eigenvalue and the raw mixed state texture global aggregation representation minimum eigenvalue of the raw mixed state texture global aggregation representation vector; calculating the raw mixed state texture global aggregation representation mean and the raw mixed state texture global aggregation representation standard deviation of the feature set of the raw mixed state texture global aggregation representation vector, and calculating the quotient of the raw mixed state texture global aggregation representation mean and the raw mixed state texture global aggregation representation standard deviation to obtain the raw mixed state texture global aggregation representation statistical normalization value; calculating the inverse of each eigenvalue of the raw mixed state texture global aggregation representation vector, and After multiplying the difference between the maximum eigenvalue of the global aggregation representation of the raw mixed state texture and the minimum eigenvalue of the global aggregation representation of the raw mixed state texture, perform point subtraction from the statistical normalized value of the global aggregation representation of the raw mixed state texture to obtain an approximate vector of the global aggregation representation of the raw mixed state texture; calculate an exponential function with the natural constant as the base, with each eigenvalue of the approximate vector of the global aggregation representation of the raw mixed state texture as the exponent to obtain an approximate vector of the global aggregation representation of the raw mixed state texture; perform point addition on the approximate vector of the global aggregation representation of the raw mixed state texture and the statistical normalized value of the global aggregation representation of the raw mixed state texture, and calculate the base-two logarithm of the absolute value of each eigenvalue of the point addition vector to obtain an optimized global aggregation representation vector of the raw mixed state texture.

[0093] Among them, the feature distribution diversity modulation process of the raw material mixed state texture global aggregation representation vector, denoted as V, is expressed by the following optimization formula:

[0094]

[0095]

[0096]

[0097] δ=v max -v min

[0098] Among them, V is the global aggregate representation vector of the raw material mixed state texture, v max The maximum eigenvalue of the raw material mixed state texture global aggregate representation vector, v m i nrepresents the minimum eigenvalue of the global aggregate representation of the raw material mixed state texture of the global aggregate representation vector of the raw material mixed state texture, and δ is the difference between the maximum eigenvalue of the global aggregate representation of the raw material mixed state texture and the minimum eigenvalue of the global aggregate representation of the raw material mixed state texture, μ is the mean of the global aggregate representation of the raw material mixed state texture of the feature set of the global aggregate representation vector of the raw material mixed state texture, σ is the standard deviation of the global aggregate representation of the raw material mixed state texture of the feature set of the global aggregate representation vector of the raw material mixed state texture, η is the statistical normalization value of the global aggregate representation of the raw material mixed state texture, ⊙ represents the dot product, Indicates point reduction, Indicates point plus, V ⊙-1 represents the calculation of the inverse of each eigenvalue of the global aggregate representation vector of the raw material mixing state texture, exp(·) represents the calculation of the exponential function operation with the natural constant as the base and each eigenvalue of the eigenvector as the exponent, V e is the approximate vector of the global aggregate representation distribution of the raw material mixture state texture, log|·| represents the logarithm operation with base 2, and V′ represents the optimized global aggregate representation vector of the raw material mixture state texture.

[0099] Based on this, by performing random statistical normalization on the diversified feature set of the global aggregate representation vector of the raw material mixed state texture, the confidence space constructed based on the overall eigenvalue of the global aggregate representation vector of the raw material mixed state texture is approximated by the response hypothesis test of each eigenvalue of the global aggregate representation vector of the raw material mixed state texture, thereby establishing the target reachability of the diversified feature distribution of the global aggregate representation vector of the raw material mixed state texture to the unified regression target, so as to achieve the balanced executableness between the mapping accuracy and mapping completeness in the texture semantic difference metric coefficient metric based on the diversified feature distribution of the global aggregate representation vector of the raw material mixed state texture, thereby improving the calculation accuracy of the texture semantic difference metric coefficient between any grid granularity real-time raw material mixed state texture feature vector in the set of the grid granularity real-time raw material mixed state texture feature vector and the global aggregate representation vector of the raw material mixed state texture. In this way, it is possible to more accurately judge whether the raw materials are mixed, thereby determining whether to stop the mixing operation. In this way, the raw material mixing state can be monitored and judged in real time in an intelligent manner during the preparation process of the seasoning, so as to make timely decisions to control the mixing operation and ensure the uniformity of the raw material mixing.

[0100] Based on the above embodiments, see Figure 4The figure is a schematic diagram of a system 800 for preparing a seasoning in an embodiment of the present application. The seasoning preparation system 800 includes: a primary crushing and screening module 810, which is used to crush and screen rice to obtain rice powder, and crush edible salt to obtain edible salt powder; a secondary crushing and screening module 820, which is used to crush and screen dried mushrooms to obtain mushroom powder, and crush and screen dried fruits and vegetables to obtain fruit and vegetable powder; a mixing and granulating module 830, which is used to convey wheat enzymatic fermentation, wheat protein hydrolysate, the rice powder, the edible salt powder, the mushroom powder, the fruit and vegetable powder, pine mushroom extract and yeast extract to a mixer for mixing, and then granulate, dry and sterilize in a granulator to obtain pine mushroom seasoning.

[0101] In one example, in the above-mentioned seasoning preparation system 800, the mixing and granulation module 830 is used to: obtain a real-time raw material mixing state distribution image collected by a camera; perform grid-granularity-based texture feature extraction on the real-time raw material mixing state distribution image to obtain a set of grid-granularity real-time raw material mixing state texture feature vectors; perform feature significance aggregation processing on the set of grid-granularity real-time raw material mixing state texture feature vectors to obtain a raw material mixing state texture global aggregation representation vector; perform a comparative analysis based on texture semantic difference measurement on each grid-granularity real-time raw material mixing state texture feature vector in the set of grid-granularity real-time raw material mixing state texture feature vectors and the raw material mixing state texture global aggregation representation vector to obtain a comparison result; based on the comparison result, determine whether to stop the mixing operation.

[0102] Here, those skilled in the art will appreciate that the specific functions and operations of the various modules in the seasoning preparation system 800 have been described in detail above. Figures 1 to 3 The preparation method of the seasoning has been described in detail in the description of the present invention, and therefore, its repeated description will be omitted.

[0103] Figure 5 FIG. 1 is an application scenario diagram of the method for preparing seasoning according to an embodiment of the present application. Figure 5 As shown, in this application scenario, first, a real-time raw material mixing state distribution image (for example, Figure 5 D), then, the real-time raw material mixing state distribution image is input to a server (e.g., Figure 5 In S) shown in , the server is able to use the seasoning preparation algorithm to process the real-time raw material mixing state distribution image to determine whether to stop the mixing operation.

[0104] Based on the above embodiment, 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 steps of the method for preparing the seasoning in the above embodiment can be implemented when the processor executes the program.

[0105] 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 method for preparing seasonings 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.

[0106] 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 method for preparing the seasoning according to the embodiments of the present application.

[0107] 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.

[0108] 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.

[0109] 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.

[0110] 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.

[0111] 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. A method for preparing a seasoning, comprising the steps of: crushing and sieving rice to obtain rice powder, and crushing edible salt to obtain edible salt powder; crushing and sieving dried mushrooms to obtain mushroom powder, and crushing and sieving dried fruits and vegetables to obtain fruit and vegetable powder; conveying wheat enzymatic fermentation product, wheat protein hydrolysate, the rice powder, the edible salt powder, the mushroom powder, the fruit and vegetable powder, pine mushroom extract and yeast extract to a mixer for mixing, and then granulating, drying and sterilizing in a granulator to obtain pine mushroom seasoning; It is characterized in that The enzymatic fermentation product of wheat, the protein hydrolyzate of wheat, the rice powder, the edible salt powder, the mushroom powder, the fruit and vegetable powder, the pine mushroom extract and the yeast extract are conveyed to a mixer for mixing, including: Acquire real-time raw material mixing state distribution images collected by a camera; Extracting texture features based on grid granularity on the real-time raw material mixing state distribution image to obtain a set of grid granularity real-time raw material mixing state texture feature vectors; Performing feature saliency aggregation processing on the set of real-time raw material mixing state texture feature vectors of the grid granularity to obtain a global aggregate representation vector of the raw material mixing state texture; Performing a comparative analysis based on texture semantic difference measurement on each grid-size real-time raw material mixed state texture feature vector in the set of the grid-size real-time raw material mixed state texture feature vectors and the raw material mixed state texture global aggregation representation vector to obtain a comparison result; Based on the comparison result, it is determined whether to stop the mixing operation.

2. The method for preparing seasoning according to claim 1, characterized in that: The real-time raw material mixing state distribution image is subjected to texture feature extraction based on grid granularity to obtain a set of real-time raw material mixing state texture feature vectors of grid granularity, including: Grid-cutting the real-time raw material mixing state distribution image to obtain a set of real-time raw material mixing state distribution images with grid granularity; Each grid-size real-time raw material mixing state distribution image in the set of grid-size real-time raw material mixing state distribution images is input into a texture feature extractor based on a dilated convolutional layer to obtain a set of grid-size real-time raw material mixing state texture feature vectors.

3. The method for preparing seasoning according to claim 2, characterized in that: The set of real-time raw material mixing state texture feature vectors at the grid granularity is subjected to feature significance aggregation processing to obtain a global aggregate representation vector of the raw material mixing state texture, including: inputting the set of real-time raw material mixing state texture feature vectors at the grid granularity into a feature significance aggregation network based on cluster contribution field analysis to obtain a global aggregate representation vector of the raw material mixing state texture.

4. The method for preparing seasoning according to claim 3, characterized in that: Inputting the set of the real-time raw material mixing state texture feature vectors of the grid granularity into the feature significant aggregation network based on cluster contribution field analysis to obtain the global aggregation representation vector of the raw material mixing state texture, including: Performing cluster analysis on the set of the real-time raw material mixing state texture feature vectors at the grid granularity to obtain a self-supervised cluster representation vector of the real-time raw material mixing state texture feature at the grid granularity; Calculating the implicit clustering contribution factor of each grid-size real-time raw material mixed state texture feature vector in the set of the grid-size real-time raw material mixed state texture feature vectors relative to the grid-size real-time raw material mixed state texture feature self-supervised clustering representation vector to obtain a sequence of grid-size real-time raw material mixed state texture feature implicit clustering contribution factors; Vectorizing and arranging the sequence of the implicit clustering contribution factors of the real-time raw material mixing state texture characteristics of the grid granularity to obtain the clustering contribution field distribution vector of the real-time raw material mixing state texture characteristics of the grid granularity; Input the grid-granularity real-time raw material mixed state texture feature cluster contribution field distribution vector into the cluster contribution field explicit modeling module based on the self-attention mechanism to obtain the grid-granularity real-time raw material mixed state texture feature cluster contribution field modulation weight vector; Taking each eigenvalue in the grid-granularity real-time raw material mixed state texture feature clustering contribution field modulation weight vector as weight, the position-weighted sum of the set of the grid-granularity real-time raw material mixed state texture feature vectors is calculated to obtain the raw material mixed state texture global aggregation representation vector.

5. The method for preparing seasoning according to claim 4, characterized in that: Cluster analysis is performed on the set of real-time raw material mixing state texture feature vectors at the grid granularity to obtain a self-supervised clustering representation vector of the real-time raw material mixing state texture feature at the grid granularity, including: calculating the mean vector of the set of real-time raw material mixing state texture feature vectors at the grid granularity to obtain the self-supervised clustering representation vector of the real-time raw material mixing state texture feature at the grid granularity.

6. The method for preparing seasoning according to claim 5, characterized in that: Calculating the implicit clustering contribution factor of each grid-size real-time raw material mixed state texture feature vector in the set of the grid-size real-time raw material mixed state texture feature vectors relative to the grid-size real-time raw material mixed state texture feature self-supervised clustering representation vector to obtain a sequence of grid-size real-time raw material mixed state texture feature implicit clustering contribution factors, including: Calculating the absolute value of the positional division between the real-time raw material mixing state texture feature vector of the grid granularity and the real-time raw material mixing state texture feature self-supervised clustering representation vector of the grid granularity to obtain the real-time raw material mixing state texture grid granularity self-supervised clustering interactive correlation feature vector; Calculate the logarithmic function value of each position feature value in the real-time raw material mixing state texture grid granularity self-supervised clustering interactive correlation feature vector with base 2 to obtain the real-time raw material mixing state texture grid granularity self-supervised clustering interactive logarithmic representation vector; Taking the eigenvalues ​​of each position in the real-time raw material mixing state texture grid granularity self-supervised clustering interactive logarithm representation vector as weighting coefficients, calculating the weighted sum of the eigenvalues ​​of each position in the real-time raw material mixing state texture feature vector of the grid granularity to obtain the grid granularity real-time raw material mixing state texture feature clustering contribution representation factor; The exponential function value of the grid granularity real-time raw material mixed state texture feature clustering contribution representation factor is calculated with the natural constant e as the base to obtain the grid granularity real-time raw material mixed state texture feature implicit clustering contribution factor.

7. The method for preparing seasoning according to claim 6, characterized in that: The grid-granularity real-time raw material mixed state texture feature cluster contribution field distribution vector is input into the cluster contribution field explicit modeling module based on the self-attention mechanism to obtain the grid-granularity real-time raw material mixed state texture feature cluster contribution field modulation weight vector, including: Calculate the matrix multiplication between the grid-granularity real-time raw material mixed state texture feature cluster contribution field distribution vector and the query weight matrix, the key weight matrix and the value weight matrix respectively to obtain the real-time raw material mixed state texture cluster contribution field distribution query vector, the real-time raw material mixed state texture cluster contribution field distribution key vector and the real-time raw material mixed state texture cluster contribution field distribution value vector; After calculating the multiplication between the real-time raw material mixed state texture cluster contribution field distribution query vector and the transposed vector of the real-time raw material mixed state texture cluster contribution field distribution key vector, the obtained matrix is ​​divided by the square root of the length of the real-time raw material mixed state texture cluster contribution field distribution key vector by position to obtain a query-key real-time raw material mixed state texture cluster contribution field semantic interaction representation matrix; The query-key real-time raw material mixed state texture clustering contribution field semantic interaction representation matrix is ​​normalized by using a Softmax function to obtain a normalized query-key real-time raw material mixed state texture clustering contribution field semantic interaction representation matrix; The multiplication between the real-time raw material mixed state texture cluster contribution field distribution value vector and the query-key real-time raw material mixed state texture cluster contribution field semantic interaction representation matrix is ​​calculated to obtain the grid-granularity real-time raw material mixed state texture feature cluster contribution field modulation weight vector.

8. The method for preparing seasoning according to claim 7, characterized in that: A comparison analysis based on texture semantic difference measurement is performed on each grid-size real-time raw material mixed state texture feature vector in the set of the grid-size real-time raw material mixed state texture feature vectors and the raw material mixed state texture global aggregation representation vector to obtain a comparison result, including: Calculating a texture semantic difference metric coefficient between any one of the set of the grid-size real-time raw material mixed state texture feature vectors and the raw material mixed state texture global aggregation representation vector to obtain a set of texture semantic difference metric coefficients; A maximum value in the set of texture semantic difference measurement coefficients is extracted, and the maximum value is compared with a preset threshold to obtain the comparison result.

9. The method for preparing seasoning according to claim 8, characterized in that: Calculating the texture semantic difference metric coefficient between any one of the set of the grid-size real-time raw material mixed state texture feature vectors and the raw material mixed state texture global aggregation representation vector to obtain a set of texture semantic difference metric coefficients, including: Calculating the square of the one-norm of the position difference between the real-time raw material mixed state texture feature vector of the grid granularity and the global aggregation representation vector of the raw material mixed state texture to obtain a one-norm representation factor of the semantic difference of the real-time raw material mixed state texture of the grid granularity; After calculating the square of the one-norm of the real-time raw material mixing state texture feature vector of the grid granularity to obtain a texture norm representation value, subtract the texture norm representation value from a constant 1 to obtain a one-norm representation factor of the real-time raw material mixing state texture feature of the grid granularity; After calculating the square of the one-norm of the raw material mixed state texture global aggregation representation vector to obtain a texture global aggregation norm representation value, subtract the texture global aggregation norm representation value from a constant 1 to obtain a raw material mixed state texture global aggregation one-norm representation factor; After multiplying the grid-size real-time raw material mixed state texture feature-norm representation factor and the raw material mixed state texture global aggregation-norm representation factor to obtain a norm interaction factor, the grid-size real-time raw material mixed state texture semantic difference-norm representation factor is divided by the norm interaction factor to obtain a texture semantic interaction factor; The texture semantic interaction factor is multiplied by a constant 2 and then added to a constant 1 to obtain a texture semantic difference coefficient; The texture semantic difference coefficient is activated using an arccosh function to obtain a texture semantic difference measurement coefficient.

Images