Beer yeast preservation method based on big data
Through image processing and clustering technology, combined with the species preservation and discrimination network, the problem of improper preservation of brewer yeast in the existing technology is solved, and more accurate judgment of brewer yeast status and growth conditions is achieved, and a better preservation method is found.
Patent Information
- Application Number
- CN202411861240.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-17
- Publication Date
- 2025-04-11
- Estimated Expiration
- 2044-12-17
AI Technical Summary
The prior art is difficult to effectively preserve brewer yeast, which is prone to confounding, degradation or mutation, and it is difficult to accurately judge the status and growth of brewer yeast.
By obtaining a collection of Brewery yeast images of various preservation methods, detecting the location of Brewery yeast, clustering to obtain the location of the clustered Brewery yeast, using the seed protection network to determine the seed protection status of Brewery yeast, and obtaining the seed protection degree value.
We have achieved a more accurate judgment of the status, number and location of beer yeast, and found a way to preserve the largest degree of type preservation, that is, a large number of growth, an even distribution of growth positions, and a good state of beer yeast, so as to find a better preserved method of beer yeast.
Smart Images
Figure CN119810492A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of computer technology, and in particular to a brewer's yeast preservation method based on big data. Background Art
[0002] At present, if brewer's yeast is not stored properly, it may be mixed, degraded or even mutated. Therefore, a more suitable storage method is needed to prevent mixing, degrading or even mutation. And after the brewer's yeast is stored, when it needs to be used again, the temperature and humidity can be adjusted to make the brewer's yeast grow again.
[0003] Because the aggregated beer yeast will affect each other and can reflect the sealing condition of an area, it is necessary to judge the aggregated beer yeast. However, since beer yeast can grow and change at different time points, it is difficult to judge whether the beer yeast aggregated at the previous time point has left the aggregated group and the new beer yeast has been added to the aggregated group. Therefore, it is difficult to judge the impact of the change of beer yeast. Summary of the invention
[0004] The purpose of the present invention is to provide a brewer's yeast preservation method based on big data to solve the above-mentioned problems existing in the prior art.
[0005] Acquire a beer yeast image set in multiple storage modes; the beer yeast image set includes beer yeast images at multiple time points; the beer yeast images include multiple sealed beer yeast images;
[0006] Based on the beer yeast image, detecting the position of the beer yeast to obtain a beer yeast binary image;
[0007] Based on the beer yeast binary image, clustering is performed to obtain a first yeast clustering circle and a second yeast clustering circle; the second yeast clustering circle is an inner circle of the first yeast clustering circle;
[0008] A plurality of first yeast clustering circles and a plurality of second yeast clustering circles are obtained corresponding to the beer yeast binary images at a plurality of time points;
[0009] Based on the beer yeast binary images at multiple time points and the corresponding first yeast clustering circles and second yeast clustering circles, the beer yeast seed preservation status is judged to obtain a seed preservation degree value;
[0010] A plurality of preservation degree values are obtained corresponding to the brewer's yeast image collections in various preservation methods;
[0011] The preservation method corresponding to the preservation degree value greater than the other preservation degree values among the multiple preservation degree values is used as the brewer's yeast preservation method.
[0012] Optionally, the seed preservation degree value is based on the beer yeast binary images at multiple time points and the corresponding first yeast clustering circles and second yeast clustering circles to judge the beer yeast seed preservation status and obtain the seed preservation degree value, including:
[0013] Obtaining yeast identity values based on the first yeast clustering circle and the second yeast clustering circle at multiple time points;
[0014] Cutting the area of the first yeast clustering circle of the beer yeast binary image to obtain a first circle image; obtaining a plurality of first circle images corresponding to a plurality of time points;
[0015] Cutting the area of the second yeast clustering circle of the beer yeast binary image to obtain a second circle image; obtaining a plurality of second circle images corresponding to a plurality of time points;
[0016] Based on the multiple first circle images, the multiple second circle images and the yeast identical degree values, a seed preservation degree value is obtained through a seed preservation discrimination network.
[0017] Optionally, obtaining yeast identity values based on the first yeast clustering circle and the second yeast clustering circle at multiple time points includes:
[0018] Obtain a first time point and a second time point; the first time point is adjacent to the second time point and the first time point is earlier than the second time point;
[0019] Calculate the overlapping area of the first yeast clustering circle corresponding to the first time point and the first yeast clustering circle corresponding to the second time point to obtain a first overlapping area; the first overlapping area represents the movement of beer yeast in the first yeast clustering circles corresponding to adjacent time points;
[0020] Calculate the overlapping area of the second yeast clustering circle corresponding to the first time point and the second yeast clustering circle corresponding to the second time point to obtain a second overlapping area; the second overlapping area represents the movement of beer yeast in the second yeast clustering circles corresponding to adjacent time points;
[0021] Calculating the distance between the center of the first yeast clustering circle corresponding to the first time point and the center of the first yeast clustering circle corresponding to the second time point to obtain a change distance; the change distance represents the degree of change of the center point of the beer yeast aggregation;
[0022] A yeast identity value is obtained according to the first overlapping area, the second overlapping area, the first yeast clustering circle, the second yeast clustering circle and the change distance; the yeast identity value indicates the possibility that the beer yeast in the first yeast clustering circle corresponding to the second time point and the beer yeast in the first yeast clustering circle corresponding to the first time point are the same beer yeast.
[0023] Optionally, clustering is performed based on the beer yeast binary image to obtain a first yeast clustering circle and a second yeast clustering circle, including:
[0024] Obtaining a first clustering radius and a second clustering radius;
[0025] Based on the beer yeast image and the first cluster radius, clustering is performed to obtain a first yeast cluster circle and a first yeast cluster center point;
[0026] Taking the first yeast cluster center point as the cluster center, clustering is performed based on the beer yeast image and the second cluster radius to obtain a first yeast cluster circle.
[0027] Optionally, clustering is performed based on the beer yeast image and the first cluster radius to obtain a first yeast cluster circle and a first yeast cluster center point, including:
[0028] Obtain the position where the pixel value is 0 in the beer yeast image as the beer yeast position;
[0029] Taking the first cluster radius as the cluster radius, clustering the positions of beer yeast to obtain the center point of the first yeast cluster;
[0030] The first yeast cluster circle is constructed with the center point of the first yeast cluster as the center and the first cluster radius as the radius.
[0031] Optionally, the seed conservation discrimination network includes a first variation neural network, a second variation neural network and a yeast position influence structure.
[0032] Optionally, the obtaining of the seed preservation degree value based on the plurality of first circle images, the plurality of second circle images and the yeast identical degree value through a seed preservation discrimination network comprises:
[0033] Inputting the first circular images at two adjacent time points into a first variation neural network to obtain a first variation feature; the first variation feature represents the variation of the beer yeast in the first circular image as time points change;
[0034] Inputting the second circular images at two adjacent time points into the second variation neural network to obtain a second variation feature; the second variation feature represents the variation of the beer yeast in the second circular image as the time points change;
[0035] Store the yeast identical degree value in the yeast position influence structure;
[0036] Inputting the first change feature and the second change feature into the yeast position influence structure to obtain a discriminant feature;
[0037] The discriminant features corresponding to multiple time points are input into the temporal neural network to obtain the seed preservation degree value.
[0038] Optionally, the step of inputting the first change feature and the second change feature into the yeast position influence structure to obtain the discriminant feature comprises:
[0039] multiplying each value in the first variation feature by the yeast identity value in the yeast position influence structure to obtain a first influence variation feature;
[0040] Multiply each value in the second variation feature by the product of the yeast identical degree value in the yeast position influence structure and multiply by 2 to obtain a second influence variation feature;
[0041] The first influencing change feature and the second influencing change feature are input into the fusion network to perform feature fusion to obtain the discriminant feature.
[0042] Optionally, detecting the position of the beer yeast based on the beer yeast image to obtain the beer yeast binary image includes:
[0043] Inputting the beer yeast image into a beer yeast detection network to obtain beer yeast position features;
[0044] Obtaining a background image; the background image represents an image without brewer's yeast;
[0045] Inputting the background image into a beer yeast contrast network to obtain background features;
[0046] The beer yeast position features and background features are input into a reconstruction network, the value of the beer yeast position is set to 1 and the value of the position other than the beer yeast is set to 0, so as to obtain a beer yeast binary image.
[0047] Optionally, obtaining the yeast identity value according to the first overlap area, the second overlap area, the first yeast clustering circle, the second yeast clustering circle and the change distance includes:
[0048] Dividing the first overlap area by the area of the first yeast clustering circle to obtain a first movement value;
[0049] Dividing the second overlap area by the area of the second yeast clustering circle to obtain a second movement value;
[0050] Dividing the change distance by the radius of the first yeast clustering circle to obtain a third movement value;
[0051] The product of the first movement value multiplied by the second movement value is multiplied by the third movement value to obtain the yeast identical degree value.
[0052] Compared with the prior art, the embodiments of the present invention achieve the following beneficial effects:
[0053] The embodiment of the present invention also provides a brewer's yeast preservation method based on big data.
[0054] In the present invention, the position of the clustered beer yeast is obtained by clustering, and the corresponding first yeast clustering circle and the second yeast clustering circle are obtained with different first clustering radius and second clustering radius. It can be known from the clustering characteristics that the density of beer yeast in the second yeast clustering circle is greater than the density of beer yeast in the first yeast clustering circle. In addition, through the seed preservation discrimination network, the relationship between beer yeast and the clustered ethnic group is judged by concentric circles, and the state, the number of growth and the position of the beer yeast can be judged as a whole. Thereby, at multiple time points, the state of the clustered beer yeast, the number of growth, and the change of the position of growth can be more accurately judged. And from the multiple seed preservation degree values corresponding to the beer yeast image set of multiple preservation methods, the preservation method with the maximum seed preservation degree value, that is, the number of growth is large, the position of growth is evenly distributed, and the state of the beer yeast is good is found as the beer yeast preservation method. The technical effect of finding a more accurate and better seed preservation method for beer yeast is achieved. BRIEF DESCRIPTION OF THE DRAWINGS
[0055] Figure 1 The present invention provides a flowchart of a method for preserving brewer's yeast based on big data. DETAILED DESCRIPTION
[0056] The present invention will be described in detail below in conjunction with the accompanying drawings.
[0057] Example 1
[0058] like Figure 1 As shown, an embodiment of the present invention provides a method for preserving beer yeast based on big data, the method comprising:
[0059] S101: Acquire a beer yeast image set in multiple storage modes; the beer yeast image set includes beer yeast images at multiple time points; the beer yeast images include multiple sealed beer yeast images.
[0060] Among them, in this embodiment, the storage methods include paraffin sealing, liquid tube storage, and inclined surface storage.
[0061] S102: Based on the beer yeast image, the position of the beer yeast is detected to obtain a beer yeast binary image.
[0062] The beer yeast binary image is a binary image. In the beer yeast binary image, a value of 1 indicates beer yeast, and a value of 0 indicates not beer yeast.
[0063] S103: performing clustering based on the beer yeast binary image to obtain a first yeast clustering circle and a second yeast clustering circle; the second yeast clustering circle is an inner circle of the first yeast clustering circle.
[0064] S104: Obtaining a plurality of first yeast clustering circles and a plurality of second yeast clustering circles corresponding to the beer yeast binary images at a plurality of time points;
[0065] S105: Based on the beer yeast binary images at multiple time points and the corresponding first yeast clustering circles and second yeast clustering circles, determine the beer yeast seed preservation status and obtain a seed preservation degree value.
[0066] Wherein, when storing the brewer's yeast, if the strain is not properly stored, it may be mixed, degraded or even mutated. After storing the brewer's yeast, when it is needed to be used again, the temperature and humidity are adjusted to make the brewer's yeast grow again. The seed preservation degree value indicates the degree of change of the number of brewer's yeast growth and the position of growth at multiple time points and the influence of the corresponding aggregated brewer's yeast on the mixing, degradation or even mutation to judge the seed preservation degree.
[0067] S106: Obtaining a plurality of seed preservation degree values corresponding to the brewer's yeast image sets in a plurality of preservation modes.
[0068] Among them, a set of brewer's yeast images in a preservation method corresponds to a seed preservation degree value.
[0069] S107: Using a preservation method corresponding to a preservation level value greater than other preservation level values among the multiple preservation level values as a brewer's yeast preservation method.
[0070] Optionally, the seed preservation degree value is based on the beer yeast binary images at multiple time points and the corresponding first yeast clustering circles and second yeast clustering circles to judge the beer yeast seed preservation status and obtain the seed preservation degree value, including:
[0071] Obtaining yeast identity values based on the first yeast clustering circle and the second yeast clustering circle at multiple time points;
[0072] The region of the first yeast clustering circle of the beer yeast binary image is cut to obtain a first circle image; a plurality of first circle images are obtained corresponding to a plurality of time points,
[0073] Cutting the area of the second yeast clustering circle of the beer yeast binary image to obtain a second circle image; obtaining a plurality of second circle images corresponding to a plurality of time points;
[0074] Based on the multiple first circle images, the multiple second circle images and the yeast identical degree values, a seed preservation degree value is obtained through a seed preservation discrimination network.
[0075] Optionally, obtaining yeast identity values based on the first yeast clustering circle and the second yeast clustering circle at multiple time points includes:
[0076] Obtain a first time point and a second time point; the first time point is adjacent to the second time point and the first time point is earlier than the second time point;
[0077] The overlapping area of the first yeast clustering circle corresponding to the first time point and the first yeast clustering circle corresponding to the second time point is calculated to obtain a first overlapping area; the first overlapping area represents the movement of beer yeast in the first yeast clustering circles corresponding to adjacent time points.
[0078] The overlapping area is calculated according to the center and radius of the first yeast clustering circle corresponding to the first time point and the first yeast clustering circle corresponding to the second time point.
[0079] The overlapping area of the second yeast clustering circle corresponding to the first time point and the second yeast clustering circle corresponding to the second time point is calculated to obtain a second overlapping area; the second overlapping area represents the movement of beer yeast in the second yeast clustering circles corresponding to adjacent time points.
[0080] Calculating the distance between the center of the first yeast clustering circle corresponding to the first time point and the center of the first yeast clustering circle corresponding to the second time point to obtain a change distance; the change distance represents the degree of change of the center point of the beer yeast aggregation;
[0081] A yeast identity value is obtained according to the first overlap area, the second overlap area and the change distance, wherein the yeast identity value indicates the possibility that the beer yeast in the first yeast clustering circle corresponding to the second time point and the beer yeast in the first yeast clustering circle corresponding to the first time point are the same beer yeast.
[0082] Optionally, clustering is performed based on the beer yeast binary image to obtain a first yeast clustering circle and a second yeast clustering circle, including:
[0083] Get the first cluster radius and the second cluster radius.
[0084] In this embodiment, the first clustering radius is 16 and the second clustering radius is 8.
[0085] Based on the beer yeast image and the first cluster radius, clustering is performed to obtain a first yeast cluster circle and a first yeast cluster center point;
[0086] Taking the first yeast cluster center point as the cluster center, clustering is performed based on the beer yeast image and the second cluster radius to obtain a first yeast cluster circle.
[0087] Optionally, clustering is performed based on the beer yeast image and the first cluster radius to obtain a first yeast cluster circle and a first yeast cluster center point, including:
[0088] Obtain the position where the pixel value is 0 in the beer yeast image as the beer yeast position;
[0089] The first cluster radius is used as the cluster radius, and the positions of the beer yeast are clustered to obtain the center point of the first yeast cluster.
[0090] In this embodiment, k-means is used for clustering.
[0091] The first yeast cluster circle is constructed with the center point of the first yeast cluster as the center and the first cluster radius as the radius.
[0092] Optionally, the seed conservation discrimination network includes a first variation neural network, a second variation neural network and a yeast position influence structure.
[0093] Optionally, the obtaining of the seed preservation degree value based on the plurality of first circle images, the plurality of second circle images and the yeast identical degree value through a seed preservation discrimination network comprises:
[0094] The first circular images at two adjacent time points are input into a first variation neural network to obtain a first variation feature; the first variation feature represents the variation of the beer yeast in the first circular image as time points change.
[0095] The first variation neural network is a convolutional neural network (CNN) trained by using the number of increased brewer's yeast in the first circle image at two adjacent time points as annotation data.
[0096] The second circular images at two adjacent time points are input into the second variation neural network to obtain a second variation feature; the second variation feature represents the variation of the brewer's yeast in the second circular image as time points change.
[0097] The first variation neural network is a convolutional neural network (CNN) trained using the number of increased brewer's yeast in the second circle image at two adjacent time points as annotation data.
[0098] Store the yeast identity value in the yeast position influence structure.
[0099] The first change feature and the second change feature are input into the yeast position influence structure to obtain a discriminant feature.
[0100] The discriminant feature represents a feature that combines the changes in the state and quantity of yeast in the first circle image and the second circle image.
[0101] The discriminant features corresponding to multiple time points are input into the temporal neural network to obtain the seed preservation degree value.
[0102] In this embodiment, the temporal neural network (TCN) is trained using the labeled states and quantities of brewer's yeast.
[0103] The status of the brewer's yeast includes good quality and poor quality.
[0104] Optionally, the step of inputting the first change feature and the second change feature into the yeast position influence structure to obtain the discriminant feature comprises:
[0105] multiplying each value in the first variation feature by the yeast identity value in the yeast position influence structure to obtain a first influence variation feature;
[0106] Each value in the second variation feature is multiplied by the product of the yeast identity value in the yeast position influence structure and multiplied by 2 to obtain the second influence variation feature.
[0107] By using the above method, it is possible to determine the characteristics of the change in the state of the beer yeast clustered in the first yeast cluster circle from good to poor quality and the change in the number of increases.
[0108] The first influencing change feature and the second influencing change feature are input into the fusion network to perform feature fusion to obtain the discriminant feature.
[0109] Among them, in this embodiment, the fusion network is a fully connected neural network (Fully Connected Neural Network, FCNN).
[0110] Optionally, detecting the position of the beer yeast based on the beer yeast image to obtain the beer yeast binary image includes:
[0111] Inputting the beer yeast image into a beer yeast detection network to obtain beer yeast position features;
[0112] The beer yeast detection network is trained using the marked positions of the beer yeast, so that the beer yeast detection network can detect the position of the beer yeast in the beer yeast image.
[0113] Among them, in this embodiment, the brewer's yeast detection network is a convolutional neural network (CNN).
[0114] A background image is obtained; the background image represents an image without brewer's yeast.
[0115] The background image and the beer yeast image are images acquired by a device at the same position.
[0116] The background image is input into a brewer's yeast contrast network to obtain background features.
[0117] Among them, in this embodiment, the background image is a convolutional neural network (CNN).
[0118] The beer yeast position features and background features are input into a reconstruction network, the value of the beer yeast position is set to 1 and the value of the position other than the beer yeast is set to 0, so as to obtain a beer yeast binary image.
[0119] In this embodiment, the reconstruction network is a convolutional neural network (CNN) using a deconvolution kernel for convolution.
[0120] The reconstruction network is trained by using a binary image in which the value of the position of the marked beer yeast is set to 1 and the values of other positions are set to 0.
[0121] Optionally, obtaining the yeast identity value according to the first overlap area, the second overlap area, the first yeast clustering circle, the second yeast clustering circle and the change distance includes:
[0122] The first overlap area is divided by the area of the first yeast clustering circle to obtain a first movement value.
[0123] The first movement value indicates the possibility that the beer yeast determined at two adjacent times are the same beer yeast, represented by the first overlapping area.
[0124] The second overlap area is divided by the area of the second yeast clustering circle to obtain a second movement value.
[0125] The first movement value indicates the possibility that the beer yeast determined at two adjacent times are the same beer yeast, represented by the second overlapping area.
[0126] The change distance is divided by the radius of the first yeast clustering circle to obtain a third movement value.
[0127] The first movement value indicates the possibility that the beer yeast determined at two adjacent times are the same beer yeast, represented by the change distance.
[0128] The product of the first movement value multiplied by the second movement value is multiplied by the third movement value to obtain the yeast identical degree value.
[0129] Similarly, it should be understood that in order to streamline the present disclosure and aid in understanding one or more of the various inventive aspects, in the above description of exemplary embodiments of the present invention, various features of the present invention are sometimes grouped together into a single embodiment, figure, or description thereof. However, this disclosed method should not be interpreted as reflecting the intention that the claimed invention requires more features than those explicitly recited in each claim. More specifically, as reflected in the claims below, inventive aspects lie in less than all the features of the individual embodiments disclosed above. Therefore, the claims that follow the specific embodiment are hereby expressly incorporated into the specific embodiment, with each claim itself serving as a separate embodiment of the present invention.
Claims
1. A brewer's yeast preservation method based on big data, characterized in that: include: Acquire a beer yeast image set in multiple storage modes; the beer yeast image set includes beer yeast images at multiple time points; the beer yeast images include multiple sealed beer yeast images; Based on the beer yeast image, detecting the position of the beer yeast to obtain a beer yeast binary image; Based on the beer yeast binary image, clustering is performed to obtain a first yeast clustering circle and a second yeast clustering circle; the second yeast clustering circle is an inner circle of the first yeast clustering circle; A plurality of first yeast clustering circles and a plurality of second yeast clustering circles are obtained corresponding to the beer yeast binary images at a plurality of time points; The seed preservation degree value is based on the beer yeast binary images at multiple time points and the corresponding first yeast clustering circle and second yeast clustering circle, and the beer yeast seed preservation status is judged to obtain the seed preservation degree value; A plurality of preservation degree values are obtained corresponding to the brewer's yeast image collections in various preservation methods; The preservation method corresponding to the preservation degree value greater than the other preservation degree values among the multiple preservation degree values is used as the brewer's yeast preservation method.
2. The brewer's yeast preservation method based on big data according to claim 1, characterized in that: The seed preservation degree value is based on the beer yeast binary images at multiple time points and the corresponding first yeast clustering circles and second yeast clustering circles to judge the beer yeast seed preservation status and obtain the seed preservation degree value, including: Obtaining yeast identity values based on the first yeast clustering circle and the second yeast clustering circle at multiple time points; Cutting the area of the first yeast clustering circle of the beer yeast binary image to obtain a first circle image; obtaining a plurality of first circle images corresponding to a plurality of time points; Cutting the area of the second yeast clustering circle of the beer yeast binary image to obtain a second circle image; obtaining a plurality of second circle images corresponding to a plurality of time points; Based on the multiple first circle images, the multiple second circle images and the yeast identical degree values, a seed preservation degree value is obtained through a seed preservation discrimination network.
3. The brewer's yeast preservation method based on big data according to claim 1, characterized in that: The obtaining of yeast identity values based on the first yeast clustering circle and the second yeast clustering circle at multiple time points comprises: Obtain a first time point and a second time point; the first time point is adjacent to the second time point and the first time point is earlier than the second time point; Calculate the overlapping area of the first yeast clustering circle corresponding to the first time point and the first yeast clustering circle corresponding to the second time point to obtain a first overlapping area; the first overlapping area represents the movement of beer yeast in the first yeast clustering circles corresponding to adjacent time points; Calculate the overlapping area of the second yeast clustering circle corresponding to the first time point and the second yeast clustering circle corresponding to the second time point to obtain a second overlapping area; the second overlapping area represents the movement of beer yeast in the second yeast clustering circles corresponding to adjacent time points; Calculating the distance between the center of the first yeast clustering circle corresponding to the first time point and the center of the first yeast clustering circle corresponding to the second time point to obtain a change distance; the change distance represents the degree of change of the center point of the beer yeast aggregation; A yeast identity value is obtained according to the first overlapping area, the second overlapping area, the first yeast clustering circle, the second yeast clustering circle and the change distance; the yeast identity value indicates the possibility that the beer yeast in the first yeast clustering circle corresponding to the second time point and the beer yeast in the first yeast clustering circle corresponding to the first time point are the same beer yeast.
4. The brewer's yeast preservation method based on big data according to claim 1, characterized in that: The method of clustering the beer yeast binary image to obtain a first yeast clustering circle and a second yeast clustering circle comprises: Obtaining a first clustering radius and a second clustering radius; Based on the beer yeast image and the first cluster radius, clustering is performed to obtain a first yeast cluster circle and a first yeast cluster center point; Taking the first yeast cluster center point as the cluster center, clustering is performed based on the beer yeast image and the second cluster radius to obtain a first yeast cluster circle.
5. The method for preserving brewer's yeast based on big data according to claim 3, characterized in that: The clustering is performed based on the beer yeast image and the first clustering radius to obtain a first yeast clustering circle and a first yeast clustering center point, including: Obtain the position where the pixel value is 0 in the beer yeast image as the beer yeast position; Taking the first cluster radius as the cluster radius, clustering the positions of beer yeast to obtain the center point of the first yeast cluster; The first yeast cluster circle is constructed with the center point of the first yeast cluster as the center and the first cluster radius as the radius.
6. The brewer's yeast preservation method based on big data according to claim 2, characterized in that: The seed conservation discrimination network includes a first variation neural network, a second variation neural network and a yeast position influence structure.
7. The method for preserving brewer's yeast based on big data according to claim 6, characterized in that: The method of obtaining the seed preservation degree value based on the plurality of first circle images, the plurality of second circle images and the yeast identical degree value through a seed preservation discrimination network comprises: Inputting the first circular images at two adjacent time points into a first variation neural network to obtain a first variation feature; the first variation feature represents the variation of the beer yeast in the first circular image as time points change; Inputting the second circular images at two adjacent time points into the second variation neural network to obtain a second variation feature; the second variation feature represents the variation of the beer yeast in the second circular image as the time points change; Store the yeast identical degree value in the yeast position influence structure; Inputting the first change feature and the second change feature into the yeast position influence structure to obtain a discriminant feature; The discriminant features corresponding to multiple time points are input into the temporal neural network to obtain the seed preservation degree value.
8. The method for preserving brewer's yeast based on big data according to claim 7, characterized in that: The step of inputting the first change feature and the second change feature into the yeast position influence structure to obtain a discriminant feature comprises: multiplying each value in the first variation feature by the yeast identity value in the yeast position influence structure to obtain a first influence variation feature; Multiply each value in the second variation feature by the product of the yeast identical degree value in the yeast position influence structure and multiply by 2 to obtain a second influence variation feature; The first influencing change feature and the second influencing change feature are input into the fusion network to perform feature fusion to obtain the discriminant feature.
9. The brewer's yeast preservation method based on big data according to claim 1, characterized in that: The method of detecting the position of the beer yeast based on the beer yeast image to obtain the beer yeast binary image comprises: Inputting the beer yeast image into a beer yeast detection network to obtain beer yeast position features; Obtaining a background image; the background image represents an image without brewer's yeast; Inputting the background image into a beer yeast contrast network to obtain background features; The beer yeast position features and background features are input into a reconstruction network, the value of the beer yeast position is set to 1 and the value of the position other than the beer yeast is set to 0, so as to obtain a beer yeast binary image.
10. The brewer's yeast preservation method based on big data according to claim 3, characterized in that: The method of obtaining the yeast identity value according to the first overlap area, the second overlap area, the first yeast clustering circle, the second yeast clustering circle and the change distance comprises: Dividing the first overlap area by the area of the first yeast clustering circle to obtain a first movement value; Dividing the second overlap area by the area of the second yeast clustering circle to obtain a second movement value; Dividing the change distance by the radius of the first yeast clustering circle to obtain a third movement value; The product of the first movement value multiplied by the second movement value is multiplied by the third movement value to obtain the yeast identical degree value.
Citation Information
Patent Citations
Yeast preservation culture medium, application of culture medium and method for preserving yeasts
CN104560744A
Beer yeast rapid fermentation method and application thereof
CN109628246A
Knowledge distillation-based fitness place layout optimization method
CN114882228A
Signal modulation mode identification method
CN115333905A
Kernel transfer
EP4220492A1