CN115861887A

CN115861887A - Fungus detection method for black garlic

Info

Publication number: CN115861887A
Application number: CN202211598192.1A
Authority: CN
Inventors: 张黎明; 张明永; 沙如意; 张丽颖
Original assignee: Jiangsu Bio Tech Ltd Fu Duomei
Current assignee: Jiangsu Bio Tech Ltd Fu Duomei
Priority date: 2022-12-14
Filing date: 2022-12-14
Publication date: 2023-03-28

Abstract

The fungus detection method for the black garlic can judge the fungus types and the state information in the fermentation liquor based on the state characteristics of the fermentation liquor, capture high-dimensional implicit characteristics among different numbers of fermentation temperature values in different time spans, and obtain classification characteristic expression containing the temperature characteristics of the fermentation liquor and the state change characteristics of the fermentation liquor by utilizing the logical association between the two characteristics. In this way, the temperature of the fermentation liquid is adaptively adjusted based on the fungus detection results obtained by the classification processing so that the fermentation liquid temperature is adapted to the growth of the fungus.

Description

Fungus detection method for black garlic

Technical Field

The application relates to the field of fungus detection, and more particularly relates to a fungus detection method for black garlic.

Background

Garlic is a flavoring spicy vegetable, and has the functions of resisting bacteria, diminishing inflammation, improving immunity, preventing and treating cardiovascular diseases, preventing and treating tumors and the like. The black garlic is prepared by naturally fermenting fresh raw garlic with skin under the environment of high temperature and high humidity without adding any additive, and is a deep-processed product of garlic. Compared with raw garlic, the black garlic has no pungent garlic odor, tastes sour and sweet, has certain content improvement of nutritional functional components such as crude protein, polyphenol and the like, and is popular with consumers.

The conventional black garlic processing technology has long production period and higher cost, the price is high, and the development of the black garlic food is restricted. The prior art related to the optimization of the black garlic process is more, but the prior art related to the application of garlic endophyte as a fermentation inoculant to the processing process of the black garlic is less. In the production process of the black garlic, a plurality of fungi can be involved, some of the different fungi have promotion effect on the preparation of the black garlic, and some of the fungi have no effect or even have negative effect on the preparation of the black garlic.

Therefore, a fungus detection scheme for black garlic is desired.

Disclosure of Invention

The present application is proposed to solve the above-mentioned technical problems. The embodiment of the application provides a fungus detection method for black garlic, which can judge the fungus type and state information in fermentation liquor based on the state characteristics of the fermentation liquor, capture high-dimensional implicit characteristics among different numbers of fermentation temperature values in different time spans, and obtain classification characteristic representation containing fermentation liquor temperature characteristics and state change characteristics of the fermentation liquor by utilizing the logical association between the two characteristics. In this way, the temperature of the fermentation liquid is adaptively adjusted based on the detection result obtained by the classification processing so that the fermentation liquid temperature is adapted to the growth of fungi.

According to one aspect of the present application, there is provided a fungus detection method for black garlic, comprising: acquiring a fermentation liquid monitoring video of a preset time period and fermentation temperature values of a plurality of preset time points in the preset time period; extracting a plurality of fermentation liquor monitoring key frames from the fermentation liquor monitoring video of the preset time period; enabling the plurality of fermentation liquor monitoring key frames to pass through a first convolution neural network model containing a depth fusion module to obtain a plurality of fermentation liquor monitoring feature matrixes; aggregating the plurality of fermentation liquid monitoring feature matrixes into a three-dimensional feature tensor along the sample dimension, and obtaining a fermentation liquid state change feature vector by using a second convolution neural network model of a three-dimensional convolution kernel; arranging the fermentation temperature values of the plurality of preset time points into a fermentation temperature input vector according to a time dimension, and then obtaining a fermentation liquid temperature characteristic vector through a multi-scale neighborhood characteristic extraction module; calculating the responsiveness estimation of the fermentation liquor temperature characteristic vector relative to the fermentation liquor state change characteristic vector to obtain a classification characteristic matrix; and passing the classification characteristic matrix through a classifier to obtain a classification result, wherein the classification result is used for indicating that the temperature of the fermentation liquor at the current time point should be increased or decreased.

In the fungus detection method for black garlic, the extracting a plurality of fermentation liquid monitoring key frames from the fermentation liquid monitoring video of the predetermined time period includes: and extracting the plurality of fermentation liquor monitoring key frames from the fermentation liquor monitoring video of the preset time period at a preset sampling frequency.

In the above fungus detection method for black garlic, the passing of the plurality of fermentation liquid monitoring key frames through a first convolutional neural network model including a depth fusion module to obtain a plurality of fermentation liquid monitoring feature matrices includes: extracting a shallow feature map from an Mth layer of the first convolutional neural network model, wherein M is greater than or equal to 1 and less than or equal to 6; extracting a deep feature map from an Nth layer of the first convolutional neural network model, wherein N/M is greater than or equal to 5 and less than or equal to 10; fusing the shallow feature map and the deep feature map by using a deep and shallow feature fusion module of the first convolution neural network model to obtain a fused feature map; and performing global pooling along the channel dimension on the fused feature map to obtain the fermentation broth monitoring feature matrix.

In the above fungus detection method for black garlic, the aggregating the plurality of fermentation liquid monitoring feature matrices along the sample dimension into a three-dimensional feature tensor, and then obtaining a fermentation liquid state change feature vector by using a second convolution neural network model of a three-dimensional convolution kernel includes: performing three-dimensional convolution coding on the three-dimensional characteristic tensor by using the second convolution neural network model to obtain a fermentation liquid state change characteristic diagram; and performing global mean pooling on each feature matrix of the fermentation broth state change feature map along the channel dimension to obtain the fermentation broth state change feature vector.

In the above method for detecting fungi in black garlic, the performing three-dimensional convolution encoding on the three-dimensional feature tensor by using the second convolutional neural network model to obtain a state change feature map of a fermentation liquid includes: performing, using the second convolutional neural network model using the three-dimensional convolutional kernel, in forward pass of layers, respectively: performing three-dimensional convolution processing on the input data based on the three-dimensional convolution kernel to obtain a convolution characteristic diagram; performing mean pooling processing based on a local feature matrix on the convolution feature map to obtain a pooled feature map; and performing nonlinear activation on the pooled feature map to obtain an activated feature map; wherein, the output of the last layer of the second convolutional neural network model is the fermentation liquor state change characteristic diagram, and the input of the first layer of the second convolutional neural network model is the three-dimensional characteristic tensor.

In the above fungus detection method for black garlic, arranging the fermentation temperature values of the plurality of predetermined time points as a fermentation temperature input vector according to a time dimension, and then obtaining a fermentation liquid temperature feature vector by a multi-scale neighborhood feature extraction module, the method includes: inputting the fermentation temperature input vector into a first convolution layer of the multi-scale neighborhood characteristic extraction module to obtain a first scale fermentation liquid temperature characteristic vector, wherein the first convolution layer has a first one-dimensional convolution kernel with a first length; inputting the fermentation temperature input vector into a second convolution layer of the multi-scale neighborhood feature extraction module to obtain a second scale fermentation liquid temperature feature vector, wherein the second convolution layer has a second one-dimensional convolution kernel with a second length, and the first length is different from the second length; and cascading the first scale fermentation liquor temperature characteristic vector and the second scale fermentation liquor temperature characteristic vector to obtain the fermentation liquor temperature characteristic vector.

In the above fungus detection method for black garlic, the inputting the fermentation temperature input vector into the first convolution layer of the multi-scale neighborhood feature extraction module to obtain a first scale fermentation liquid temperature feature vector includes: performing one-dimensional convolution coding on the fermentation temperature input vector by using a first convolution layer of the multi-scale neighborhood characteristic extraction module according to the following formula to obtain a first scale fermentation liquid temperature characteristic vector; wherein the formula is:

wherein a is the width of the first convolution kernel in the X direction, F (a) is a parameter vector of the first convolution kernel, G (X-a) is a local vector matrix operated with a convolution kernel function, w is the size of the first convolution kernel, and X represents the fermentation temperature input vector; the inputting the fermentation temperature input vector into a second convolution layer of the multi-scale neighborhood characteristic extraction module to obtain a second scale fermentation liquid temperature characteristic vector comprises: performing one-dimensional convolution coding on the fermentation temperature input vector by using a second convolution layer of the multi-scale neighborhood characteristic extraction module according to the following formula to obtain a second scale fermentation liquid temperature characteristic vector; wherein the formula is:

wherein b is the width of the second convolution kernel in the X direction, F (b) is a second convolution kernel parameter vector, G (X-b) is a local vector matrix operated with the convolution kernel function, m is the size of the second convolution kernel, and X represents the fermentation temperature input vector.

In the above fungus detection method for black garlic, the calculating a responsiveness estimate of the fermentation liquid temperature feature vector with respect to the fermentation liquid state change feature vector to obtain a classification feature matrix includes: respectively carrying out relative angle probability information representation correction on the fermentation broth temperature characteristic vector and the fermentation broth state change characteristic vector to obtain an optimized fermentation broth temperature characteristic vector and an optimized fermentation broth state change characteristic vector; wherein the formula is:

wherein

And &>

Respectively is the ith eigenvalue, V, of the fermentation liquor state change eigenvector and the fermentation liquor temperature eigenvector ₁ And V ₂ Is the characteristic vector of the change in state of the fermentation broth and the characteristic vector of the temperature of the fermentation broth, respectively, and->

And &>

Is the mean value of all characteristic values of the characteristic vector of the state change of the fermentation liquor and the characteristic vector of the temperature of the fermentation liquor respectively>

And &>

Respectively representing the optimized fermentation broth temperature characteristic vector and the optimized fermentation broth state change characteristic vector, wherein log represents logarithm taking 2 as a base; calculating the responsiveness estimation of the optimized fermentation broth temperature characteristic vector relative to the optimized fermentation broth state change characteristic vector by the following formula to obtain a classification characteristic matrix; wherein the formula is:

wherein V _a Representing the temperature characteristic vector, V, of the optimized fermentation broth _b Representing the optimized fermentation broth state change feature vector, M represents the classification feature matrix,

representing a matrix multiplication.

In the above fungus detection method for black garlic, the step of passing the classification feature matrix through a classifier to obtain a classification result, where the classification result is used to indicate that the temperature of the fermentation liquid at the current time point should be increased or decreased, includes: expanding the classification characteristic matrix into classification characteristic vectors according to row vectors or column vectors; performing full-concatenation coding on the classification feature vectors by using a full-concatenation layer of the classifier to obtain coded classification feature vectors; and inputting the encoding classification feature vector into a Softmax classification function of the classifier to obtain the classification result.

According to another aspect of the present application, there is provided a fungus detection system for black garlic, comprising: the data acquisition module is used for acquiring a fermentation liquid monitoring video in a preset time period and fermentation temperature values of a plurality of preset time points in the preset time period; the sampling module is used for extracting a plurality of fermentation liquor monitoring key frames from the fermentation liquor monitoring video of the preset time period; the depth feature coding module is used for enabling the plurality of fermentation liquid monitoring key frames to pass through a first convolution neural network model comprising a depth fusion module so as to obtain a plurality of fermentation liquid monitoring feature matrixes; the three-dimensional convolution coding module is used for aggregating the fermentation liquid monitoring characteristic matrixes into a three-dimensional characteristic tensor along the sample dimension and then obtaining a fermentation liquid state change characteristic vector by using a second convolution neural network model of a three-dimensional convolution kernel; the multi-scale coding module is used for arranging the fermentation temperature values of the plurality of preset time points into a fermentation temperature input vector according to a time dimension and then obtaining a fermentation liquid temperature characteristic vector through the multi-scale neighborhood characteristic extraction module; the responsiveness estimation module is used for calculating the responsiveness estimation of the fermentation liquid temperature characteristic vector relative to the fermentation liquid state change characteristic vector to obtain a classification characteristic matrix; and the detection result generation module is used for enabling the classification characteristic matrix to pass through a classifier to obtain a classification result, and the classification result is used for indicating that the temperature of the fermentation liquor at the current time point should be increased or decreased.

In the fungus detection system for black garlic, the sampling module is configured to extract the plurality of fermentation liquid monitoring key frames from the fermentation liquid monitoring video of the predetermined time period at a predetermined sampling frequency.

In the above fungus detection system for black garlic, the depth feature encoding module is further configured to: extracting a shallow feature map from an Mth layer of the first convolutional neural network model, wherein M is greater than or equal to 1 and less than or equal to 6; extracting a deep feature map from an Nth layer of the first convolutional neural network model, wherein N/M is greater than or equal to 5 and less than or equal to 10; fusing the shallow feature map and the deep feature map by using a deep and shallow feature fusion module of the first convolution neural network model to obtain a fused feature map; and the number of the first and second groups, and performing global pooling along the channel dimension on the fused feature map to obtain the fermentation broth monitoring feature matrix.

In the above fungus detection system for black garlic, the three-dimensional convolutional encoding module includes: the encoding unit is used for carrying out three-dimensional convolution encoding on the three-dimensional characteristic tensor by using the second convolution neural network model so as to obtain a state change characteristic diagram of fermentation liquor; and the dimension reduction unit is used for performing global mean pooling on each feature matrix along the channel dimension of the fermentation liquid state change feature map to obtain the fermentation liquid state change feature vector.

In the above fungus detection system for black garlic, the encoding unit is further configured to: performing, using the second convolutional neural network model using the three-dimensional convolutional kernel, in forward pass of layers, respectively: performing three-dimensional convolution processing on the input data based on the three-dimensional convolution kernel to obtain a convolution characteristic diagram; performing mean pooling processing based on a local feature matrix on the convolution feature map to obtain a pooled feature map; and performing nonlinear activation on the pooled feature map to obtain an activated feature map; wherein, the output of the last layer of the second convolutional neural network model is the fermentation liquor state change characteristic diagram, and the input of the first layer of the second convolutional neural network model is the three-dimensional characteristic tensor.

In the fungus detection system for black garlic, the multi-scale encoding module is further configured to: inputting the fermentation temperature input vector into a first convolution layer of the multi-scale neighborhood characteristic extraction module to obtain a first scale fermentation liquid temperature characteristic vector, wherein the first convolution layer has a first one-dimensional convolution kernel with a first length; inputting the fermentation temperature input vector into a second convolution layer of the multi-scale neighborhood feature extraction module to obtain a second scale fermentation liquid temperature feature vector, wherein the second convolution layer has a second one-dimensional convolution kernel with a second length, and the first length is different from the second length; and cascading the first scale fermentation liquor temperature characteristic vector and the second scale fermentation liquor temperature characteristic vector to obtain the fermentation liquor temperature characteristic vector.

In the above fungus detection system for black garlic, the detection result generation module is further configured to: expanding the classification characteristic matrix into classification characteristic vectors according to row vectors or column vectors; performing full-concatenation coding on the classification feature vectors by using a full-concatenation layer of the classifier to obtain coded classification feature vectors; and inputting the encoding classification feature vector into a Softmax classification function of the classifier to obtain the classification result.

According to still another aspect of the present application, there is provided an electronic apparatus including: a processor; and a memory having stored therein computer program instructions that, when executed by the processor, cause the processor to perform the fungus detection method for black garlic as described above.

According to yet another aspect of the present application, there is provided a computer readable medium having stored thereon computer program instructions which, when executed by a processor, cause the processor to perform the fungus detection method for black garlic as described above.

Compared with the prior art, the fungus detection method for the black garlic can judge the fungus type and the state information in the fermentation liquor based on the state characteristics of the fermentation liquor, capture high-dimensional implicit characteristics among different numbers of fermentation temperature values in different time spans, and obtain the classification characteristic representation containing the temperature characteristics of the fermentation liquor and the state change characteristics of the fermentation liquor by utilizing the logical association between the two characteristics. In this way, the temperature of the fermentation broth is adaptively adjusted based on the detection results obtained by the classification process so that the fermentation broth temperature is adapted to the growth of fungi.

Drawings

The above and other objects, features and advantages of the present application will become more apparent by describing in more detail embodiments of the present application with reference to the attached drawings. The accompanying drawings are included to provide a further understanding of the embodiments of the application and are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description serve to explain the principles of the application. In the drawings, like reference numbers generally indicate like parts or steps.

Fig. 1 is an application scene diagram of a fungus detection method for black garlic according to an embodiment of the application.

FIG. 2 is a flowchart of a fungus detection method for black garlic according to an embodiment of the present application.

FIG. 3 is a schematic diagram illustrating a fungus detection method for black garlic according to an embodiment of the present invention.

Fig. 4 is a flowchart illustrating that the plurality of fermentation liquid monitoring key frames pass through a first convolutional neural network model including a depth fusion module to obtain a plurality of fermentation liquid monitoring feature matrices according to the fungus detection method for black garlic in the embodiment of the present application.

FIG. 5 is a block diagram of a fungus detection system for black garlic according to an embodiment of the present application.

Fig. 6 is a block diagram of an electronic device according to an embodiment of the application.

Detailed Description

Hereinafter, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. It should be understood that the described embodiments are only some embodiments of the present application and not all embodiments of the present application, and that the present application is not limited by the example embodiments described herein.

Application overview: as mentioned in the background art, there are many existing technologies related to the optimization of black garlic process, but there are few existing technologies related to the application of garlic endophytes as fermentation inoculants in the processing of black garlic. In the production process of the black garlic, a plurality of fungi are involved, wherein some fungi in different fungi have a promoting effect on the preparation of the black garlic, and some fungi have no effect or even have negative effect on the preparation of the black garlic. Therefore, a fungus detection scheme for black garlic is desired, which can determine the fungus type and status information in a fermentation broth based on status characteristics of the fermentation broth, and adaptively adjust the temperature of the fermentation broth based on the detection result so that the fermentation broth temperature is adapted to the growth of the fungus.

As will be appreciated by those of ordinary skill in the art, an endophytic bacteria is a bacteria that can colonize a living plant without causing significant changes in the host plant tissue structure and, under normal circumstances, without causing disease symptoms in the host plant. The endophytic bacteria have stable living space in the plant body, are not easily influenced by environmental conditions, and have a harmonious symbiotic relationship with host plants. The endophyte can produce a plurality of secondary metabolites which have a plurality of biological activities, and the functions enable the endophyte to be used as a plant probiotic bacterium to play roles in a plurality of fields such as biological pesticide, yield increasing microbial inoculum and the likePlays an important role. Found through research, S ₈ The nyzx-1 strain has excellent performance in promoting the fermentation of the black garlic, but utilizes S ₈ When the nyzx-1 strain is used for optimizing the preparation process of the black garlic, S is required ₈ The nyzx-1 strain provides suitable growth and reproduction conditions, and therefore, a fungus detection scheme for black garlic capable of determining S in a fermentation broth of black garlic based on the state of the fermentation broth is desired ₈ Growth status of nyzx-1 strain, and then adjusting the temperature of the fermentation broth based on the detection result to adapt the fermentation broth temperature to S ₈ Growth and reproduction of the nyzx-1 fungus.

Specifically, in the technical scheme of the application, a fermentation liquid monitoring video of a predetermined time period and fermentation temperature values of a plurality of predetermined time points in the predetermined time period are obtained first. It should be understood that the species in the fermentation broth may affect the state characteristics, particularly the color characteristics, of the fermentation broth, and thus, the species detection may be performed on the fermentation broth by monitoring the change in the state characteristics of the fermentation broth in the video. In particular, in the solution of the present application, with S ₈ The growth state of the nyzx-1 strain in the fermentation liquor is more and more vigorous, the number of the nyzx-1 strain in the fermentation liquor is more and more, and the color of the fermentation liquor can slowly turn red, so that the change characteristic of the state characteristic of the fermentation liquor in the fermentation liquor monitoring video and the strain in the fermentation liquor have a complex nonlinear relation.

In order to capture and utilize the implicit relationship, in the technical scheme of the application, the fermentation liquor monitoring video is processed by using a deep neural network model based on deep learning to obtain a fermentation liquor state change feature vector. In particular, it is considered that a plurality of image frames in all image frame sequences of the fermentation broth monitoring video are highly similar or even repeated, which causes information redundancy and interferes with feature extraction. Therefore, before feature extraction, in the technical solution of the present application, sampling processing is performed on the fermentation broth monitoring video, and in a specific example, a plurality of fermentation broth monitoring key frames are extracted from the fermentation broth monitoring video in the predetermined time period at a predetermined sampling frequency.

And then, enabling the plurality of fermentation liquor monitoring key frames to pass through a first convolutional neural network model comprising a depth fusion module to obtain a plurality of fermentation liquor monitoring characteristic matrixes. That is, a convolutional neural network model having excellent performance in the field of image feature extraction is used as a feature extractor to capture high-dimensional local image implicit features of each of the plurality of fermentation broth monitoring key frames. It is considered that as the depth of convolutional encoding of the convolutional neural network model increases, the extracted image features are more abstract and more reflect the essence of the object, specifically, the shallow features thereof more represent appearance, lines, textures and colors, and the deep features thereof more represent object types, object features and the like. In view of the expectation that the color change of the fermentation liquid is more focused in fungus detection in the technical solution of the present application, the structure of the convolutional neural network model is adjusted to integrate a depth feature fusion mechanism into a feature extraction mechanism of the convolutional neural network model.

And then, aggregating the plurality of fermentation liquor monitoring characteristic matrixes into a three-dimensional characteristic tensor along the sample dimension, and obtaining a fermentation liquor state change characteristic vector by using a second convolution neural network model of a three-dimensional convolution kernel. That is, in the high-dimensional feature space, the plurality of fermentation liquid monitoring feature matrices are subjected to information aggregation along the sample dimension to obtain a three-dimensional feature tensor, and a convolution neural network model using a three-dimensional convolution kernel is used as a feature extractor to capture fermentation liquid state change features. The second convolutional neural network model performs three-dimensional convolutional encoding using a three-dimensional convolutional kernel compared to a conventional convolutional neural network model, wherein the three-dimensional convolutional kernel has three dimensions: the width dimension, the height dimension and the channel dimension correspond to local spaces of the image frames, and the channel dimension corresponds to the time dimension of the three-dimensional feature tensor, so that the variation feature of the state feature of the fermentation liquid in the space dimension in the time dimension can be extracted in the process of carrying out three-dimensional convolution coding.

According to the technical scheme, the fermentation temperature values of the preset time points are arranged into a fermentation temperature input vector according to the time dimension, and then the fermentation temperature input vector is obtained through a multi-scale neighborhood characteristic extraction module. That is, the fermentation temperature values at the plurality of predetermined time points are first subjected to vectorization processing to obtain a fermentation temperature input vector, i.e., a time-series distribution of the fermentation temperature values. And then, carrying out multi-scale one-dimensional convolution coding on the fermentation temperature input vector by using a multi-scale neighborhood feature extraction module comprising a plurality of parallel one-dimensional convolution layers so as to capture high-dimensional implicit features among different numbers of fermentation temperature values in different time spans, and carrying out feature fusion on the associated features of different scales to obtain the fermentation liquid temperature feature vector.

In the technical scheme of the application, the fermentation liquor temperature is a cause of state change of the fermentation liquor method, that is, the fermentation liquor temperature and the state change of the fermentation liquor have a correlation in a logic level, and a classification characteristic expression comprising a fermentation liquor temperature characteristic and a state change characteristic of the fermentation liquor is obtained by utilizing the logic correlation between the fermentation liquor temperature and the state change of the fermentation liquor. Specifically, calculating the responsiveness estimation of the fermentation liquor temperature characteristic vector relative to the fermentation liquor state change characteristic vector to obtain a classification characteristic matrix. And after the classification characteristic matrix is obtained, the classification characteristic matrix is passed through a classifier to obtain a classification result, wherein the classification result is used for indicating that the temperature of the fermentation liquor at the current time point should be increased or decreased.

Thus, the S in the fermentation liquor of the black garlic is judged based on the state of the fermentation liquor ₈ Growth state of nyzx-1 strain, and then adjusting the temperature of the fermentation broth based on the detection result to make the fermentation broth temperature adapted to S ₈ Growth and reproduction of the nyzx-1 fungus.

In particular, when the classification feature vector is obtained by calculating the responsiveness estimation of the fermentation liquid temperature feature vector relative to the fermentation liquid state change feature vector, since the fermentation liquid temperature feature vector expresses a time-series multi-scale correlation feature of a fermentation temperature value, and the fermentation liquid state change feature vector expresses a channel dimension correlation feature of monitoring image semantics in time-series arrangement, a spatial position error exists in a time-series direction in a feature distribution of the fermentation liquid temperature feature vector and the fermentation liquid state change feature vector, so that the accuracy of the classification feature vector obtained by calculating the position-by-position responsiveness estimation of the fermentation liquid temperature feature vector relative to the fermentation liquid state change feature vector is influenced.

The applicant of the present application considers that the fermentation liquid temperature characteristic vector and the fermentation liquid state change characteristic vector are both obtained from time-series arranged data, and therefore are substantially in-phase dimension, and thus the fermentation liquid temperature characteristic vector and the fermentation liquid state change characteristic vector have a certain correspondence on characteristic distribution as in-phase characteristic expression, and therefore, the fermentation liquid temperature characteristic vector and the fermentation liquid state change characteristic vector can be respectively subjected to relative class angle probability information representation correction, which is expressed as:

wherein

And &>

Respectively is the state change characteristic vector V of the fermentation liquor ₁ And the temperature eigenvector V of the fermentation liquor ₂ And/or an ith characteristic value of>

And &>

Is the fermentation liquor state change characteristic vector V ₁ And the temperature eigenvector V of the fermentation liquor ₂ Log represents the base 2 logarithm of the mean of all characteristic values of (a).

Here, the relative angle-like probability information indicates a characteristic vector V for correcting the state change of the fermentation liquid ₁ And the temperature eigenvector V of the fermentation liquor ₂ Relative angle probability information representation between the two is carried out to carry out the fermentation liquor state change characteristic vector V ₁ And the temperature eigenvector V of the fermentation liquor ₂ Geometric dilution of spatial position error of feature distribution in high-dimensional feature space, thereby changing feature vector V in state of fermentation liquid ₁ And the temperature eigenvector V of the fermentation liquor ₂ Under the condition of certain correspondence, based on the fermentation liquor state change characteristic vector V ₁ And the temperature eigenvector V of the fermentation liquor ₂ The implicit context correspondence correction of the features is performed by point-by-point regression of the positions, as compared with the integral distribution constraint of the respective feature value distributions of the respective positions, thereby improving the state change feature vector V of the fermentation liquid ₁ And the fermentation broth temperature eigenvector V ₂ The position-by-position correspondence between the fermentation liquor temperature characteristic vectors is improved ₂ Relative to the state change eigenvector V of the fermentation broth ₁ The computational accuracy of the responsiveness estimate.

Based on the above, the present application provides a fungus detection method for black garlic, which includes: acquiring a fermentation liquid monitoring video of a preset time period and fermentation temperature values of a plurality of preset time points in the preset time period; extracting a plurality of fermentation liquor monitoring key frames from the fermentation liquor monitoring video of the preset time period; enabling the plurality of fermentation liquor monitoring key frames to pass through a first convolution neural network model containing a depth fusion module to obtain a plurality of fermentation liquor monitoring feature matrixes; aggregating the plurality of fermentation liquid monitoring characteristic matrixes into a three-dimensional characteristic tensor along the sample dimension, and obtaining a fermentation liquid state change characteristic vector by using a second convolution neural network model of a three-dimensional convolution kernel; arranging the fermentation temperature values of the plurality of preset time points into a fermentation temperature input vector according to a time dimension, and then obtaining a fermentation liquid temperature characteristic vector through a multi-scale neighborhood characteristic extraction module; calculating the responsiveness estimation of the fermentation liquor temperature characteristic vector relative to the fermentation liquor state change characteristic vector to obtain a classification characteristic matrix; and passing the classification characteristic matrix through a classifier to obtain a classification result, wherein the classification result is used for indicating that the temperature of the fermentation liquor at the current time point should be increased or decreased.

Fig. 1 is an application scenario diagram of a fungus detection method for black garlic according to an embodiment of the present application. As shown in fig. 1, in this application scenario, a camera (e.g., C as illustrated in fig. 1) first obtains a fermentation broth monitoring video for a predetermined period of time, and a temperature sensor (e.g., se as illustrated in fig. 1) obtains fermentation temperature values at a plurality of predetermined time points within the predetermined period of time. Further, the fermentation broth monitoring video of the predetermined time period and the fermentation temperature values at a plurality of predetermined time points in the predetermined time period are input into a server (for example, S as shown in fig. 1) deployed with a fungus detection algorithm for black garlic, wherein the server can process the fermentation broth monitoring video of the predetermined time period and the fermentation temperature values at a plurality of predetermined time points in the predetermined time period based on the fungus detection algorithm for black garlic to obtain a classification result indicating that the fermentation broth temperature at the current time point should be increased or decreased.

Having described the basic principles of the present application, various non-limiting embodiments of the present application will now be described with reference to the accompanying drawings.

An exemplary method: fig. 2 is a flowchart of a fungus detection method for black garlic according to an embodiment of the present application. As shown in fig. 2, the fungus detection method for black garlic according to the embodiment of the present application includes: s110, acquiring a fermentation liquid monitoring video in a preset time period and fermentation temperature values of a plurality of preset time points in the preset time period; s120, extracting a plurality of fermentation liquor monitoring key frames from the fermentation liquor monitoring video in the preset time period; s130, enabling the plurality of fermentation liquor monitoring key frames to pass through a first convolution neural network model comprising a depth fusion module to obtain a plurality of fermentation liquor monitoring characteristic matrixes; s140, aggregating the plurality of fermentation liquor monitoring characteristic matrixes into a three-dimensional characteristic tensor along the sample dimension, and obtaining a fermentation liquor state change characteristic vector by using a second convolution neural network model of a three-dimensional convolution kernel; s150, arranging the fermentation temperature values of the plurality of preset time points into a fermentation temperature input vector according to a time dimension, and then obtaining a fermentation liquid temperature characteristic vector through a multi-scale neighborhood characteristic extraction module; s160, calculating the responsiveness estimation of the temperature characteristic vector of the fermentation liquor relative to the state change characteristic vector of the fermentation liquor to obtain a classification characteristic matrix; and S170, passing the classification characteristic matrix through a classifier to obtain a classification result, wherein the classification result is used for indicating that the temperature of the fermentation liquor at the current time point should be increased or decreased.

FIG. 3 is an architecture diagram of a fungus detection method for black garlic according to an embodiment of the present application. As shown in fig. 3, in this framework, first, a fermentation broth monitoring video of a predetermined time period, fermentation temperature values of a plurality of predetermined time points within the predetermined time period are obtained, and a plurality of fermentation broth monitoring key frames are extracted from the fermentation broth monitoring video of the predetermined time period. And then, passing the fermentation liquid monitoring key frames through a first convolution neural network model containing a depth fusion module to obtain a plurality of fermentation liquid monitoring feature matrixes. Then, after the plurality of fermentation liquid monitoring characteristic matrixes are aggregated into a three-dimensional characteristic tensor along the sample dimension, a fermentation liquid state change characteristic vector is obtained by using a second convolution neural network model of a three-dimensional convolution kernel, and meanwhile, fermentation temperature values of the plurality of preset time points are arranged into a fermentation temperature input vector according to the time dimension and then are obtained by a multi-scale neighborhood characteristic extraction module. Then, calculating the responsiveness estimation of the fermentation liquor temperature characteristic vector relative to the fermentation liquor state change characteristic vector to obtain a classification characteristic matrix. And further, the classification characteristic matrix is passed through a classifier to obtain a classification result, and the classification result is used for indicating that the temperature of the fermentation liquor at the current time point should be increased or decreased.

In step S110, a fermentation broth monitoring video of a predetermined time period and fermentation temperature values of a plurality of predetermined time points within the predetermined time period are obtained. As mentioned in the background art, there are many existing technologies related to the optimization of black garlic process, but there are few existing technologies related to the application of garlic endophytes as fermentation inoculants to the processing of black garlic. In the production process of the black garlic, a plurality of fungi are involved, wherein some fungi in different fungi have a promoting effect on the preparation of the black garlic, and some fungi have no effect or even have negative effect on the preparation of the black garlic. Therefore, a fungus detection scheme for black garlic is expected, which can determine the type and state information of fungi in a fermentation liquid based on the state characteristics of the fermentation liquid, and adaptively adjust the temperature of the fermentation liquid based on the detection result so that the temperature of the fermentation liquid is adapted to the growth of fungi.

As will be appreciated by those of ordinary skill in the art, a plant-derived endophytic bacterium is one that is capable of colonizing a living plant without causing significant changes in the tissue structure of the host plant and, under normal conditions, without causing disease symptoms in the host plant. The endophytic bacteria have stable living space in the plant body, are not easily influenced by environmental conditions, and have a harmonious symbiotic relationship with host plants. The endophyte can produce a plurality of secondary metabolites which have a plurality of biological activities, and the functions ensure that the endophyte plays an important role as a plant probiotic in a plurality of fields such as biological pesticide, yield increasing microbial inoculum and the like. Through research, S ₈ The nyzx-1 strain has excellent performance in promoting the fermentation of the black garlic, but utilizes S ₈ When the nyzx-1 strain is used for optimizing the preparation process of the black garlic, S is required ₈ The nyzx-1 strain provides suitable growth and reproduction conditions, and therefore, a fungus detection scheme for black garlic capable of determining S in a fermentation broth of black garlic based on the state of the fermentation broth is desired ₈ Growth status of nyzx-1 strain, and then adjusting the temperature of the fermentation broth based on the detection result to adapt the fermentation broth temperature to S ₈ Growth and reproduction of nyzx-1 fungus.

In particular toIn the technical scheme of the application, firstly, a fermentation liquid monitoring video of a preset time period and fermentation temperature values of a plurality of preset time points in the preset time period are obtained. The fermentation liquid monitoring video can be acquired by a camera, and the fermentation temperature values of the plurality of preset time points can be acquired by a temperature sensor. It should be understood that the species in the fermentation broth may affect the state characteristics, particularly the color characteristics, of the fermentation broth, and thus, the species detection may be performed on the fermentation broth by monitoring the change in the state characteristics of the fermentation broth in the video. In particular, in the solution of the present application, with S ₈ The growth state of the nyzx-1 strain in the fermentation liquor is more and more vigorous, the number of the nyzx-1 strain in the fermentation liquor is more and more, and the color of the fermentation liquor can slowly turn red, so that the change characteristic of the state characteristic of the fermentation liquor in the fermentation liquor monitoring video and the strain in the fermentation liquor have a complex nonlinear relation.

In step S120, a plurality of fermentation broth monitoring key frames are extracted from the fermentation broth monitoring video of the predetermined time period. In order to capture and utilize the complex nonlinear implicit association, in the technical scheme of the application, the fermentation liquid monitoring video is processed by using a deep neural network model based on deep learning to obtain a fermentation liquid state change feature vector. In particular, it is considered that a plurality of image frames in all image frame sequences of the fermentation broth monitoring video are highly similar or even repeated, which causes information redundancy and interferes with feature extraction. Therefore, before feature extraction, in the technical solution of the present application, sampling processing is performed on the fermentation broth monitoring video, and in a specific example, a plurality of fermentation broth monitoring key frames are extracted from the fermentation broth monitoring video in the predetermined time period at a predetermined sampling frequency.

In step S130, the plurality of fermentation broth monitoring key frames are passed through a first convolutional neural network model including a depth fusion module to obtain a plurality of fermentation broth monitoring feature matrices. That is, a convolutional neural network model having excellent performance in the field of image feature extraction is used as a feature extractor to capture high-dimensional local image implicit features of each of the plurality of fermentation broth monitoring key frames. It is considered that as the depth of convolutional encoding of the convolutional neural network model increases, the extracted image features are more abstract and more reflect the essence of the object, specifically, the shallow features thereof more represent appearance, lines, textures and colors, and the deep features thereof more represent object types, object features and the like. In view of the expectation that the color change of the fermentation liquid is more focused in fungus detection in the technical solution of the present application, the structure of the convolutional neural network model is adjusted to integrate a depth feature fusion mechanism into a feature extraction mechanism of the convolutional neural network model.

In one particular example, the first convolutional neural network model includes a plurality of neural network layers cascaded with one another, where each neural network layer includes a convolutional layer, a pooling layer, and an activation layer. In the encoding process of the first convolutional neural network model, each layer of the first convolutional neural network model performs convolution processing based on a convolution kernel on input data by using the convolutional layer in the forward transmission process of the layer, performs pooling processing on a convolution feature map output by the convolutional layer by using the pooling layer, and performs activation processing on the pooling feature map output by using the activation layer, wherein the input data of the first layer of the first convolutional neural network model is used for each fermentation liquid monitoring key frame in the plurality of fermentation liquid monitoring key frames. Here, each layer of the first convolutional neural network model may output a feature map.

Fig. 4 is a flowchart illustrating that the plurality of fermentation liquid monitoring key frames pass through a first convolutional neural network model including a depth fusion module to obtain a plurality of fermentation liquid monitoring feature matrices according to the fungus detection method for black garlic in the embodiment of the present application. As shown in fig. 4, the passing the plurality of fermentation broth monitoring key frames through a first convolutional neural network model including a depth fusion module to obtain a plurality of fermentation broth monitoring feature matrices includes: s210, extracting a shallow feature map from the Mth layer of the first convolution neural network model, wherein M is more than or equal to 1 and less than or equal to 6; s220, extracting a deep feature map from the Nth layer of the first convolution neural network model, wherein N/M is more than or equal to 5 and less than or equal to 10; s230, fusing the shallow feature map and the deep feature map by using a deep and shallow feature fusion module of the first convolution neural network model to obtain a fusion feature map; and S240, performing global pooling along the channel dimension on the fused feature map to obtain the fermentation broth monitoring feature matrix.

In step S140, the plurality of fermentation broth monitoring feature matrices are aggregated into a three-dimensional feature tensor along the sample dimension, and then a fermentation broth state change feature vector is obtained by using a second convolution neural network model of a three-dimensional convolution kernel. That is, in the high-dimensional feature space, the plurality of fermentation liquid monitoring feature matrices are subjected to information aggregation along the sample dimension to obtain a three-dimensional feature tensor, and a convolution neural network model using a three-dimensional convolution kernel is used as a feature extractor to capture fermentation liquid state change features. The second convolutional neural network model performs three-dimensional convolutional encoding using a three-dimensional convolutional kernel compared to a conventional convolutional neural network model, wherein the three-dimensional convolutional kernel has three dimensions: the method comprises the following steps of obtaining a state feature of a fermentation liquid, wherein the state feature of the fermentation liquid is a state feature of the fermentation liquid in a spatial dimension, and the state feature of the fermentation liquid in a time dimension can be extracted in the process of carrying out three-dimensional convolutional coding.

Specifically, in this embodiment of the present application, after aggregating the plurality of fermentation broth monitoring feature matrices into a three-dimensional feature tensor along the sample dimension, obtaining a fermentation broth state change feature vector by using a second convolutional neural network model of a three-dimensional convolution kernel includes: performing three-dimensional convolution coding on the three-dimensional characteristic tensor by using the second convolution neural network model to obtain a fermentation liquid state change characteristic diagram; and performing global mean pooling on each feature matrix of the fermentation broth state change feature map along the channel dimension to obtain the fermentation broth state change feature vector.

More specifically, in this embodiment of the present application, the three-dimensional convolution encoding the three-dimensional feature tensor by using the second convolutional neural network model to obtain a state change feature map of a fermentation broth includes: performing, using the second convolutional neural network model using the three-dimensional convolutional kernel, in forward pass of layers, respectively: performing three-dimensional convolution processing on the input data based on the three-dimensional convolution kernel to obtain a convolution characteristic diagram; performing mean pooling processing based on a local feature matrix on the convolution feature map to obtain a pooled feature map; and performing nonlinear activation on the pooled feature map to obtain an activated feature map; wherein, the output of the last layer of the second convolutional neural network model is the fermentation liquor state change characteristic diagram, and the input of the first layer of the second convolutional neural network model is the three-dimensional characteristic tensor.

In step S150, the fermentation temperature values at the plurality of predetermined time points are arranged as a fermentation temperature input vector according to a time dimension, and then a multi-scale neighborhood feature extraction module is used to obtain a fermentation liquid temperature feature vector. Here, for the fermentation temperature values at the plurality of predetermined time points, in the technical scheme of the application, the fermentation temperature values at the plurality of predetermined time points are arranged as a fermentation temperature input vector according to a time dimension, and then a fermentation liquid temperature feature vector is obtained through a multi-scale neighborhood feature extraction module. That is, the fermentation temperature values at the plurality of predetermined time points are first subjected to vectorization processing to obtain a fermentation temperature input vector, i.e., a time-series distribution of the fermentation temperature values. And then, carrying out multi-scale one-dimensional convolution coding on the fermentation temperature input vector by using a multi-scale neighborhood feature extraction module comprising a plurality of parallel one-dimensional convolution layers so as to capture high-dimensional implicit features among different numbers of fermentation temperature values in different time spans, and carrying out feature fusion on the associated features of different scales to obtain the fermentation liquid temperature feature vector.

Specifically, in this embodiment of the present application, the arranging the fermentation temperature values of the plurality of predetermined time points into a fermentation temperature input vector according to a time dimension and then obtaining a fermentation liquid temperature feature vector through a multi-scale neighborhood feature extraction module includes: inputting the fermentation temperature input vector into a first convolution layer of the multi-scale neighborhood characteristic extraction module to obtain a first scale fermentation liquid temperature characteristic vector, wherein the first convolution layer has a first one-dimensional convolution kernel with a first length; inputting the fermentation temperature input vector into a second convolution layer of the multi-scale neighborhood feature extraction module to obtain a second scale fermentation liquid temperature feature vector, wherein the second convolution layer has a second one-dimensional convolution kernel with a second length, and the first length is different from the second length; and cascading the first scale fermentation liquor temperature characteristic vector and the second scale fermentation liquor temperature characteristic vector to obtain the fermentation liquor temperature characteristic vector.

Specifically, in this embodiment of the present application, the inputting the fermentation temperature input vector into the first convolution layer of the multi-scale neighborhood region feature extraction module to obtain a first-scale fermentation broth temperature feature vector includes: performing one-dimensional convolution coding on the fermentation temperature input vector by using a first convolution layer of the multi-scale neighborhood characteristic extraction module according to the following formula to obtain a first scale fermentation liquid temperature characteristic vector; wherein the formula is:

In step S160, a responsiveness estimation of the fermentation broth temperature eigenvector relative to the fermentation broth state change eigenvector is calculated to obtain a classification feature matrix. In the technical scheme of the application, the fermentation liquor temperature is a cause of state change of a fermentation liquor method, that is, the fermentation liquor temperature and the state change of the fermentation liquor have a correlation on a logic level, and a classification characteristic representation comprising a fermentation liquor temperature characteristic and a fermentation liquor state change characteristic is obtained by utilizing logic correlation between the fermentation liquor temperature and the fermentation liquor state change.

In particular, when the classification feature vector is obtained by calculating the responsiveness estimation of the fermentation liquid temperature feature vector relative to the fermentation liquid state change feature vector, since the fermentation liquid temperature feature vector expresses a time-series multi-scale correlation feature of a fermentation temperature value, and the fermentation liquid state change feature vector expresses a channel dimension correlation feature of monitoring image semantics in time-series arrangement, a spatial position error exists in a time-series direction in a feature distribution of the fermentation liquid temperature feature vector and the fermentation liquid state change feature vector, so that the accuracy of the classification feature vector obtained by calculating the position-by-position responsiveness estimation of the fermentation liquid temperature feature vector relative to the fermentation liquid state change feature vector is influenced. The applicant of the present application considers that the fermentation liquid temperature characteristic vector and the fermentation liquid state change characteristic vector are both obtained from time-series arranged data and are therefore substantially in-phase dimensions, and therefore, the fermentation liquid temperature characteristic vector and the fermentation liquid state change characteristic vector have a certain correspondence on characteristic distribution as in-phase characteristic expressions, and therefore, the fermentation liquid temperature characteristic vector and the fermentation liquid state change characteristic vector can be respectively subjected to relative class angle probability information representation correction.

Specifically, in this embodiment of the present application, the calculating a responsiveness estimation of the fermentation liquid temperature feature vector with respect to the fermentation liquid state change feature vector to obtain a classification feature matrix includes: respectively carrying out relative angle probability information representation correction on the fermentation liquor temperature characteristic vector and the fermentation liquor state change characteristic vector to obtain an optimized fermentation liquor temperature characteristic vector and an optimized fermentation liquor state change characteristic vector; wherein the formula is:

wherein

And &>

And &>

Respectively are the average values of all the characteristic values of the fermentation liquor state change characteristic vector and the fermentation liquor temperature characteristic vector, device for selecting or keeping>

And &>

representing a matrix multiplication.

In step S170, the classification feature matrix is passed through a classifier to obtain a classification result, which is used to indicate that the temperature of the fermentation broth at the current time point should be increased or decreased. Namely, the classifier is used for carrying out class boundary division and determination on the classification characteristic matrix so as to obtain the classification result. Thus, the S in the fermentation liquor of the black garlic is judged based on the state of the fermentation liquor ₈ Growth status of nyzx-1 strain, and then adjusting the temperature of the fermentation broth based on the detection result to adapt the fermentation broth temperature to S ₈ Growth and reproduction of the nyzx-1 fungus.

Specifically, in this embodiment of the present application, the passing the classification feature matrix through a classifier to obtain a classification result, where the classification result is used to indicate that the fermentation broth temperature at the current time point should be increased or decreased, includes: expanding the classification characteristic matrix into classification characteristic vectors according to row vectors or column vectors; performing full-concatenation coding on the classification feature vectors by using a full-concatenation layer of the classifier to obtain coded classification feature vectors; and inputting the encoding classification feature vector into a Softmax classification function of the classifier to obtain the classification result.

In summary, the fungus detection method for black garlic according to the embodiments of the present application is elucidated, which is capable of determining the fungus type and state information in a fermentation liquid based on the state characteristics of the fermentation liquid, and capturing high-dimensional implicit characteristics between different numbers of fermentation temperature values in different time spans, so as to obtain a classification characteristic representation including the temperature characteristics of the fermentation liquid and the state change characteristics of the fermentation liquid by using the logical association between the two. In this way, the temperature of the fermentation liquid is adaptively adjusted based on the detection result obtained by the classification processing so that the fermentation liquid temperature is adapted to the growth of fungi.

An exemplary system: FIG. 5 is a block diagram of a fungus detection system for black garlic according to an embodiment of the present application. As shown in fig. 5, the fungus detection system 100 for black garlic according to the embodiment of the present application includes: the data acquisition module 110 is configured to acquire a fermentation broth monitoring video in a predetermined time period and fermentation temperature values at a plurality of predetermined time points in the predetermined time period; a sampling module 120, configured to extract a plurality of fermentation broth monitoring key frames from the fermentation broth monitoring video of the predetermined time period; the depth feature encoding module 130 is configured to pass the plurality of fermentation broth monitoring key frames through a first convolutional neural network model including a depth fusion module to obtain a plurality of fermentation broth monitoring feature matrices; the three-dimensional convolution coding module 140 is configured to aggregate the plurality of fermentation liquid monitoring feature matrices into a three-dimensional feature tensor along the sample dimension, and then obtain a fermentation liquid state change feature vector through a second convolution neural network model using a three-dimensional convolution kernel; the multi-scale coding module 150 is used for arranging the fermentation temperature values of the plurality of preset time points into a fermentation temperature input vector according to a time dimension and then obtaining a fermentation liquid temperature characteristic vector through the multi-scale neighborhood characteristic extraction module; a responsiveness estimation module 160, configured to calculate a responsiveness estimation of the fermentation liquid temperature eigenvector relative to the fermentation liquid state change eigenvector to obtain a classification feature matrix; and a detection result generating module 170, configured to pass the classification feature matrix through a classifier to obtain a classification result, where the classification result is used to indicate that the temperature of the fermentation liquid at the current time point should be increased or decreased.

In an example, in the fungus detection system 100 for black garlic, the sampling module 120 is configured to extract the plurality of fermentation broth monitoring key frames from the fermentation broth monitoring video of the predetermined time period at a predetermined sampling frequency.

In an example, in the fungus detection system 100 for black garlic, the depth feature encoding module 130 is further configured to: extracting a shallow feature map from an Mth layer of the first convolutional neural network model, wherein M is greater than or equal to 1 and less than or equal to 6; extracting a deep feature map from an Nth layer of the first convolutional neural network model, wherein N/M is greater than or equal to 5 and less than or equal to 10; fusing the shallow feature map and the deep feature map by using a deep and shallow feature fusion module of the first convolution neural network model to obtain a fused feature map; and performing global pooling along the channel dimension on the fused feature map to obtain the fermentation broth monitoring feature matrix.

In one example, in the above fungus detection system 100 for black garlic, the three-dimensional convolutional encoding module 140 includes: the encoding unit is used for carrying out three-dimensional convolution encoding on the three-dimensional characteristic tensor by using the second convolution neural network model so as to obtain a state change characteristic diagram of fermentation liquor; and the dimension reduction unit is used for performing global mean pooling on each feature matrix of the fermentation liquid state change feature map along the channel dimension to obtain the fermentation liquid state change feature vector.

In an example, in the fungus detection system 100 for black garlic, the encoding unit is further configured to: respectively performing, in forward pass of layers, input data using the second convolutional neural network model using the three-dimensional convolutional kernel: performing three-dimensional convolution processing on the input data based on the three-dimensional convolution kernel to obtain a convolution characteristic diagram; performing mean pooling processing based on a local feature matrix on the convolution feature map to obtain a pooled feature map; and performing nonlinear activation on the pooled feature map to obtain an activated feature map; wherein the output of the last layer of the second convolutional neural network model is the fermentation liquor state change characteristic diagram, and the input of the first layer of the second convolutional neural network model is the three-dimensional characteristic tensor.

In one example, in the above fungus detection system 100 for black garlic, the multi-scale encoding module 150 is further configured to: inputting the fermentation temperature input vector into a first convolution layer of the multi-scale neighborhood characteristic extraction module to obtain a first scale fermentation liquid temperature characteristic vector, wherein the first convolution layer has a first one-dimensional convolution kernel with a first length; inputting the fermentation temperature input vector into a second convolution layer of the multi-scale neighborhood feature extraction module to obtain a second scale fermentation liquid temperature feature vector, wherein the second convolution layer has a second one-dimensional convolution kernel with a second length, and the first length is different from the second length; and cascading the first scale fermentation liquor temperature characteristic vector and the second scale fermentation liquor temperature characteristic vector to obtain the fermentation liquor temperature characteristic vector.

In one example, in the fungus detection system 100 for black garlic, the responsiveness estimation module 160 is further configured to: respectively carrying out relative angle probability information representation correction on the fermentation broth temperature characteristic vector and the fermentation broth state change characteristic vector to obtain an optimized fermentation broth temperature characteristic vector and an optimized fermentation broth state change characteristic vector; wherein the formula is:

wherein

And &>

And &>

Is the mean value of all the characteristic values of the fermentation liquor state change characteristic vector and the fermentation liquor temperature characteristic vector respectively, and is/are>

And &>

representing a matrix multiplication.

In an example, in the fungus detection system 100 for black garlic, the detection result generation module 170 is further configured to: expanding the classification characteristic matrix into classification characteristic vectors according to row vectors or column vectors; performing full-concatenation coding on the classification feature vectors by using a full-concatenation layer of the classifier to obtain coded classification feature vectors; and inputting the encoding classification feature vector into a Softmax classification function of the classifier to obtain the classification result.

Here, it can be understood by those skilled in the art that the specific functions and operations of the respective units and modules in the above-described fungus detection system 100 for black garlic have been described in detail in the above description of the fungus detection method for black garlic with reference to fig. 1 to 4, and thus, a repetitive description thereof will be omitted.

As described above, the fungus detection system 100 for black garlic according to the embodiment of the present application can be implemented in various terminal devices, for example, a server or the like for fungus detection of black garlic. In one example, the fungus detection system 100 for black garlic according to an embodiment of the present application may be integrated into a terminal device as one software module and/or hardware module. For example, the fungus detection system 100 for black garlic 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 fungus detection system 100 for black garlic can also be one of many hardware modules of the terminal device.

Alternatively, in another example, the fungus detection system 100 for black garlic and the terminal device may be separate devices, and the fungus detection system 100 for black garlic may be connected to the terminal device through a wired and/or wireless network and transmit the interaction information according to an agreed data format.

An exemplary electronic device: next, an electronic apparatus according to an embodiment of the present application is described with reference to fig. 6. Fig. 6 is a block diagram of an electronic device according to an embodiment of the application. As shown in fig. 6, the electronic device 10 includes one or more processors 11 and memory 12.

The processor 11 may be a Central Processing Unit (CPU) or other form of processing unit having data processing capabilities and/or instruction execution capabilities, and may control other components in the electronic device 10 to perform desired functions.

Memory 12 may include one or more computer program products that may include various forms of computer-readable storage media, such as volatile memory and/or non-volatile memory. The volatile memory may include, for example, random Access Memory (RAM), cache memory (cache), and/or the like. The non-volatile memory may include, for example, read Only Memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium and executed by the processor 11 to implement the functions of the fungus detection method for black garlic of the various embodiments of the present application described above and/or other desired functions. Various contents such as fermentation broth monitoring video, fermentation temperature value, etc. may also be stored in the computer readable storage medium.

In one example, the electronic device 10 may further include: an input device 13 and an output device 14, which are interconnected by a bus system and/or other form of connection mechanism (not shown).

The input device 13 may include, for example, a keyboard, a mouse, and the like.

The output device 14 can output various information including the classification result to the outside. The output devices 14 may include, for example, a display, speakers, a printer, and a communication network and its connected remote output devices, among others.

Of course, for simplicity, only some of the components of the electronic device 10 relevant to the present application are shown in fig. 6, and components such as buses, input/output interfaces, and the like are omitted. In addition, the electronic device 10 may include any other suitable components depending on the particular application.

Exemplary computer program products and computer-readable storage media: in addition to the above-described methods and apparatus, embodiments of the present application may also be a computer program product comprising computer program instructions that, when executed by a processor, cause the processor to perform the steps in the functions of the fungus detection method for black garlic according to various embodiments of the present application described in the "exemplary methods" section above of the present specification.

The computer program product may be written with program code for performing the operations of embodiments of the present application in any combination of one or more programming languages, including an object oriented programming language such as Java, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device, or entirely on the remote computing device or server.

Furthermore, embodiments of the present application may also be a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, cause the processor to perform the steps in the functions in the fungus detection method for black garlic according to various embodiments of the present application described in the "exemplary methods" section above in this specification.

The computer-readable storage medium may take any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may include, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples (a non-exhaustive list) of the readable storage medium include: an electrical connection having one or more wires, a portable disk, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

The foregoing describes the general principles of the present application in conjunction with specific embodiments, however, it is noted that the advantages, effects, etc. mentioned in the present application are merely examples and are not limiting, and they should not be considered essential to the various embodiments of the present application. Furthermore, the foregoing disclosure of specific details is for the purpose of illustration and description and is not intended to be limiting, since the foregoing disclosure is not intended to be exhaustive or to limit the disclosure to the precise details disclosed.

The block diagrams of devices, apparatuses, systems referred to in this application are only given as illustrative examples and are not intended to require or imply that the connections, arrangements, configurations, etc. must be made in the manner shown in the block diagrams. These devices, apparatuses, devices, systems may be connected, arranged, configured in any manner, as will be appreciated by those skilled in the art. Words such as "including," "comprising," "having," and the like are open-ended words that mean "including, but not limited to," and are used interchangeably therewith. The words "or" and "as used herein mean, and are used interchangeably with, the word" and/or, "unless the context clearly dictates otherwise. The word "such as" is used herein to mean, and is used interchangeably with, the phrase "such as but not limited to".

It should also be noted that in the devices, apparatuses, and methods of the present application, the components or steps may be decomposed and/or recombined. These decompositions and/or recombinations are to be considered as equivalents of the present application.

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

The foregoing description has been presented for purposes of illustration and description. Furthermore, the description is not intended to limit embodiments of the application to the form disclosed herein. While a number of example aspects and embodiments have been discussed above, those of skill in the art will recognize certain variations, modifications, alterations, additions and sub-combinations thereof.

Claims

1. A fungus detection method for black garlic is characterized by comprising the following steps: acquiring a fermentation liquid monitoring video of a preset time period and fermentation temperature values of a plurality of preset time points in the preset time period; extracting a plurality of fermentation liquor monitoring key frames from the fermentation liquor monitoring video of the preset time period; enabling the plurality of fermentation liquor monitoring key frames to pass through a first convolution neural network model containing a depth fusion module to obtain a plurality of fermentation liquor monitoring feature matrixes; aggregating the plurality of fermentation liquid monitoring characteristic matrixes into a three-dimensional characteristic tensor along the sample dimension, and obtaining a fermentation liquid state change characteristic vector by using a second convolution neural network model of a three-dimensional convolution kernel; arranging the fermentation temperature values of the plurality of preset time points into a fermentation temperature input vector according to a time dimension, and then obtaining a fermentation liquid temperature characteristic vector through a multi-scale neighborhood characteristic extraction module; calculating the responsiveness estimation of the fermentation liquor temperature characteristic vector relative to the fermentation liquor state change characteristic vector to obtain a classification characteristic matrix; and passing the classification characteristic matrix through a classifier to obtain a classification result, wherein the classification result is used for indicating that the temperature of the fermentation liquor at the current time point should be increased or decreased.

2. The fungus detection method for black garlic according to claim 1, wherein the extracting a plurality of fermentation broth monitoring key frames from the fermentation broth monitoring video of the predetermined period of time comprises: and extracting the plurality of fermentation liquor monitoring key frames from the fermentation liquor monitoring video of the preset time period at a preset sampling frequency.

3. The fungus detection method for black garlic according to claim 2, wherein the passing the plurality of fermentation broth monitoring key frames through a first convolutional neural network model comprising a depth fusion module to obtain a plurality of fermentation broth monitoring feature matrices comprises: extracting a shallow feature map from an Mth layer of the first convolutional neural network model, wherein M is greater than or equal to 1 and less than or equal to 6; extracting a deep feature map from an Nth layer of the first convolutional neural network model, wherein N/M is greater than or equal to 5 and less than or equal to 10; fusing the shallow feature map and the deep feature map by using a deep and shallow feature fusion module of the first convolution neural network model to obtain a fused feature map; and performing global pooling along the channel dimension on the fusion feature map to obtain the fermentation broth monitoring feature matrix.

4. The fungus detection method for black garlic according to claim 3, wherein the step of aggregating the plurality of fermentation broth monitoring feature matrices along the sample dimension into a three-dimensional feature tensor and obtaining a fermentation broth state change feature vector by using a second convolutional neural network model of a three-dimensional convolutional kernel comprises: performing three-dimensional convolution coding on the three-dimensional characteristic tensor by using the second convolution neural network model to obtain a fermentation liquid state change characteristic diagram; and performing global mean pooling on each feature matrix of the fermentation broth state change feature map along the channel dimension to obtain the fermentation broth state change feature vector.

5. The fungus detection method for black garlic according to claim 4, wherein the three-dimensional convolution encoding the three-dimensional feature tensor using the second convolutional neural network model to obtain a fermentation broth state change feature map comprises: performing, using the second convolutional neural network model using the three-dimensional convolutional kernel, in forward pass of layers, respectively: performing three-dimensional convolution processing on the input data based on the three-dimensional convolution kernel to obtain a convolution characteristic diagram; performing mean pooling processing based on a local feature matrix on the convolution feature map to obtain a pooled feature map; and carrying out nonlinear activation on the pooling feature map to obtain an activation feature map; wherein, the output of the last layer of the second convolutional neural network model is the fermentation liquor state change characteristic diagram, and the input of the first layer of the second convolutional neural network model is the three-dimensional characteristic tensor.

6. The fungus detection method for black garlic according to claim 5, wherein the step of arranging the fermentation temperature values of the plurality of predetermined time points into a fermentation temperature input vector according to a time dimension and then obtaining a fermentation liquid temperature feature vector through a multi-scale neighborhood feature extraction module comprises the steps of: inputting the fermentation temperature input vector into a first convolution layer of the multi-scale neighborhood characteristic extraction module to obtain a first scale fermentation liquid temperature characteristic vector, wherein the first convolution layer has a first one-dimensional convolution kernel with a first length; inputting the fermentation temperature input vector into a second convolution layer of the multi-scale neighborhood feature extraction module to obtain a second scale fermentation liquid temperature feature vector, wherein the second convolution layer has a second one-dimensional convolution kernel with a second length, and the first length is different from the second length; and

and cascading the first scale fermentation liquor temperature characteristic vector and the second scale fermentation liquor temperature characteristic vector to obtain the fermentation liquor temperature characteristic vector.

7. The fungus detection method for black garlic according to claim 6, wherein the inputting the fermentation temperature input vector into the first convolution layer of the multi-scale neighborhood feature extraction module to obtain a first-scale fermentation broth temperature feature vector comprises: performing one-dimensional convolution coding on the fermentation temperature input vector by using a first convolution layer of the multi-scale neighborhood characteristic extraction module according to the following formula to obtain a first scale fermentation liquid temperature characteristic vector; wherein the formula is:

wherein a is the width of the first convolution kernel in the X direction, F (a) is a parameter vector of the first convolution kernel, G (X-a) is a local vector matrix operated with a convolution kernel function, w is the size of the first convolution kernel, and X represents the fermentation temperature input vector; the inputting the fermentation temperature input vector into the second convolution layer of the multi-scale neighborhood characteristic extraction module to obtain a second scale fermentation liquid temperature characteristic vector comprises: performing one-dimensional convolution coding on the fermentation temperature input vector by using a second convolution layer of the multi-scale neighborhood characteristic extraction module according to the following formula to obtain a second scale fermentation liquid temperature characteristic vector;

wherein the formula is:

8. The fungus detection method for black garlic according to claim 7, wherein the calculating of the responsiveness estimate of the fermentation liquid temperature eigenvector with respect to the fermentation liquid state change eigenvector to obtain a classification feature matrix comprises: respectively carrying out relative angle probability information representation correction on the fermentation broth temperature characteristic vector and the fermentation broth state change characteristic vector to obtain an optimized fermentation broth temperature characteristic vector and an optimized fermentation broth state change characteristic vector;

wherein the formula is:

wherein

And &>

Respectively is the ith eigenvalue, V, of the fermentation liquor state change eigenvector and the fermentation liquor temperature eigenvector ₁ And V ₂ Respectively are the fermentation liquor state change characteristic vector and the fermentation liquor temperature characteristic vector, and

and &>

And &>

Respectively representing the optimized fermentation broth temperature characteristic vector and the optimized fermentation broth state change characteristic vector, wherein log represents logarithm taking 2 as a base; calculating the responsiveness estimation of the optimized fermentation broth temperature characteristic vector relative to the optimized fermentation broth state change characteristic vector by the following formula to obtain a classification characteristic matrix;

wherein the formula is:

wherein V _a Expressing the temperature characteristic vector, V, of the optimized fermentation broth _b Representing the optimized fermentation broth state change feature vector, M represents the classification feature matrix,

representing a matrix multiplication.

9. The fungus detection method for black garlic according to claim 8, wherein the step of passing the classification feature matrix through a classifier to obtain a classification result indicating that the fermentation broth temperature should be increased or decreased at the current time point comprises: expanding the classification characteristic matrix into classification characteristic vectors according to row vectors or column vectors; performing full-concatenation coding on the classification feature vectors by using a full-concatenation layer of the classifier to obtain coded classification feature vectors; and inputting the encoding classification feature vector into a Softmax classification function of the classifier to obtain the classification result.