CN117037152B - Machine vision-based botrytis cinerea control effect analysis method and device - Google Patents
Machine vision-based botrytis cinerea control effect analysis method and device Download PDFInfo
- Publication number
- CN117037152B CN117037152B CN202311301316.XA CN202311301316A CN117037152B CN 117037152 B CN117037152 B CN 117037152B CN 202311301316 A CN202311301316 A CN 202311301316A CN 117037152 B CN117037152 B CN 117037152B
- Authority
- CN
- China
- Prior art keywords
- botrytis cinerea
- control effect
- colony
- machine vision
- effect analysis
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 241000123650 Botrytis cinerea Species 0.000 title claims abstract description 83
- 230000000694 effects Effects 0.000 title claims abstract description 53
- 238000004458 analytical method Methods 0.000 title claims abstract description 33
- 238000005286 illumination Methods 0.000 claims abstract description 19
- 239000003814 drug Substances 0.000 claims abstract description 17
- 230000002265 prevention Effects 0.000 claims abstract description 17
- 238000012545 processing Methods 0.000 claims abstract description 16
- 230000011218 segmentation Effects 0.000 claims abstract description 16
- 238000011161 development Methods 0.000 claims abstract description 9
- 230000018109 developmental process Effects 0.000 claims abstract description 9
- 238000000034 method Methods 0.000 claims abstract description 4
- 230000001580 bacterial effect Effects 0.000 claims description 27
- 238000012549 training Methods 0.000 claims description 22
- 230000007246 mechanism Effects 0.000 claims description 18
- 230000000844 anti-bacterial effect Effects 0.000 claims description 9
- 235000015097 nutrients Nutrition 0.000 claims description 9
- 230000004927 fusion Effects 0.000 claims description 7
- 230000004913 activation Effects 0.000 claims description 6
- 230000001502 supplementing effect Effects 0.000 claims description 6
- 235000016709 nutrition Nutrition 0.000 claims description 4
- 238000012795 verification Methods 0.000 claims description 4
- 238000002372 labelling Methods 0.000 claims description 3
- 238000012360 testing method Methods 0.000 claims description 3
- 239000004615 ingredient Substances 0.000 claims description 2
- 229940079593 drug Drugs 0.000 abstract 1
- 235000012041 food component Nutrition 0.000 abstract 1
- 241000196324 Embryophyta Species 0.000 description 11
- 239000003899 bactericide agent Substances 0.000 description 7
- 238000011156 evaluation Methods 0.000 description 6
- 229930014626 natural product Natural products 0.000 description 6
- 241001465180 Botrytis Species 0.000 description 4
- 208000031888 Mycoses Diseases 0.000 description 4
- 235000013399 edible fruits Nutrition 0.000 description 4
- 244000052616 bacterial pathogen Species 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 3
- 239000005760 Difenoconazole Substances 0.000 description 2
- 239000005802 Mancozeb Substances 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- BQYJATMQXGBDHF-UHFFFAOYSA-N difenoconazole Chemical compound O1C(C)COC1(C=1C(=CC(OC=2C=CC(Cl)=CC=2)=CC=1)Cl)CN1N=CN=C1 BQYJATMQXGBDHF-UHFFFAOYSA-N 0.000 description 2
- 201000010099 disease Diseases 0.000 description 2
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 230000002070 germicidal effect Effects 0.000 description 2
- 230000002401 inhibitory effect Effects 0.000 description 2
- 230000035764 nutrition Effects 0.000 description 2
- 230000003449 preventive effect Effects 0.000 description 2
- 239000003223 protective agent Substances 0.000 description 2
- 238000012827 research and development Methods 0.000 description 2
- 238000012216 screening Methods 0.000 description 2
- 235000013311 vegetables Nutrition 0.000 description 2
- WURBVZBTWMNKQT-UHFFFAOYSA-N 1-(4-chlorophenoxy)-3,3-dimethyl-1-(1,2,4-triazol-1-yl)butan-2-one Chemical compound C1=NC=NN1C(C(=O)C(C)(C)C)OC1=CC=C(Cl)C=C1 WURBVZBTWMNKQT-UHFFFAOYSA-N 0.000 description 1
- PXMNMQRDXWABCY-UHFFFAOYSA-N 1-(4-chlorophenyl)-4,4-dimethyl-3-(1H-1,2,4-triazol-1-ylmethyl)pentan-3-ol Chemical compound C1=NC=NN1CC(O)(C(C)(C)C)CCC1=CC=C(Cl)C=C1 PXMNMQRDXWABCY-UHFFFAOYSA-N 0.000 description 1
- HZJKXKUJVSEEFU-UHFFFAOYSA-N 2-(4-chlorophenyl)-2-(1H-1,2,4-triazol-1-ylmethyl)hexanenitrile Chemical compound C=1C=C(Cl)C=CC=1C(CCCC)(C#N)CN1C=NC=N1 HZJKXKUJVSEEFU-UHFFFAOYSA-N 0.000 description 1
- 241000219104 Cucurbitaceae Species 0.000 description 1
- 206010059866 Drug resistance Diseases 0.000 description 1
- 239000005780 Fluazinam Substances 0.000 description 1
- 239000005867 Iprodione Substances 0.000 description 1
- 239000005811 Myclobutanil Substances 0.000 description 1
- 239000005869 Pyraclostrobin Substances 0.000 description 1
- 239000005828 Pyrimethanil Substances 0.000 description 1
- 235000004789 Rosa xanthina Nutrition 0.000 description 1
- 241000220222 Rosaceae Species 0.000 description 1
- 241000208292 Solanaceae Species 0.000 description 1
- 239000005839 Tebuconazole Substances 0.000 description 1
- 239000005858 Triflumizole Substances 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 239000004480 active ingredient Substances 0.000 description 1
- 239000011149 active material Substances 0.000 description 1
- 239000013543 active substance Substances 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 239000013043 chemical agent Substances 0.000 description 1
- 229930002875 chlorophyll Natural products 0.000 description 1
- 235000019804 chlorophyll Nutrition 0.000 description 1
- ATNHDLDRLWWWCB-AENOIHSZSA-M chlorophyll a Chemical compound C1([C@@H](C(=O)OC)C(=O)C2=C3C)=C2N2C3=CC(C(CC)=C3C)=[N+]4C3=CC3=C(C=C)C(C)=C5N3[Mg-2]42[N+]2=C1[C@@H](CCC(=O)OC\C=C(/C)CCC[C@H](C)CCC[C@H](C)CCCC(C)C)[C@H](C)C2=C5 ATNHDLDRLWWWCB-AENOIHSZSA-M 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 230000003111 delayed effect Effects 0.000 description 1
- 230000006806 disease prevention Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000003912 environmental pollution Methods 0.000 description 1
- 230000005284 excitation Effects 0.000 description 1
- UZCGKGPEKUCDTF-UHFFFAOYSA-N fluazinam Chemical compound [O-][N+](=O)C1=CC(C(F)(F)F)=C(Cl)C([N+]([O-])=O)=C1NC1=NC=C(C(F)(F)F)C=C1Cl UZCGKGPEKUCDTF-UHFFFAOYSA-N 0.000 description 1
- 230000004907 flux Effects 0.000 description 1
- 244000053095 fungal pathogen Species 0.000 description 1
- 230000005802 health problem Effects 0.000 description 1
- 238000010191 image analysis Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 230000009545 invasion Effects 0.000 description 1
- ONUFESLQCSAYKA-UHFFFAOYSA-N iprodione Chemical compound O=C1N(C(=O)NC(C)C)CC(=O)N1C1=CC(Cl)=CC(Cl)=C1 ONUFESLQCSAYKA-UHFFFAOYSA-N 0.000 description 1
- 230000000670 limiting effect Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 231100000053 low toxicity Toxicity 0.000 description 1
- 244000005700 microbiome Species 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000000056 organ Anatomy 0.000 description 1
- 230000007918 pathogenicity Effects 0.000 description 1
- 239000000447 pesticide residue Substances 0.000 description 1
- QXJKBPAVAHBARF-BETUJISGSA-N procymidone Chemical compound O=C([C@]1(C)C[C@@]1(C1=O)C)N1C1=CC(Cl)=CC(Cl)=C1 QXJKBPAVAHBARF-BETUJISGSA-N 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- HZRSNVGNWUDEFX-UHFFFAOYSA-N pyraclostrobin Chemical compound COC(=O)N(OC)C1=CC=CC=C1COC1=NN(C=2C=CC(Cl)=CC=2)C=C1 HZRSNVGNWUDEFX-UHFFFAOYSA-N 0.000 description 1
- ZLIBICFPKPWGIZ-UHFFFAOYSA-N pyrimethanil Chemical compound CC1=CC(C)=NC(NC=2C=CC=CC=2)=N1 ZLIBICFPKPWGIZ-UHFFFAOYSA-N 0.000 description 1
- 230000002829 reductive effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 208000024891 symptom Diseases 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 230000009885 systemic effect Effects 0.000 description 1
- HSMVPDGQOIQYSR-KGENOOAVSA-N triflumizole Chemical compound C1=CN=CN1C(/COCCC)=N/C1=CC=C(Cl)C=C1C(F)(F)F HSMVPDGQOIQYSR-KGENOOAVSA-N 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/69—Microscopic objects, e.g. biological cells or cellular parts
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/255—Detecting or recognising potential candidate objects based on visual cues, e.g. shapes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/70—Labelling scene content, e.g. deriving syntactic or semantic representations
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Medical Informatics (AREA)
- Databases & Information Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Measuring Or Testing Involving Enzymes Or Micro-Organisms (AREA)
Abstract
The invention relates to a machine vision-based botrytis cinerea control effect analysis method and a machine vision-based botrytis cinerea control effect analysis device, wherein a camera is used for shooting and collecting botrytis cinerea culture images in a culture dish, and condition factors in an observation device are recorded; processing the botrytis cinerea culture image through a semantic segmentation model, and taking the colony size data as a label of the colony segmentation image; the method comprises the steps of converting an RGB color image by taking illumination intensity, temperature and nutritional components as RGB values of the RGB color image and taking medicament concentration as intensity values of the RGB color image, inputting the obtained RGB color image and labeled colony segmentation images as a strain growth prevention network prediction model, and predicting the growth and development conditions of colonies under various condition factors by using the strain growth prevention network prediction model. The invention can accurately analyze the growth condition of botrytis cinerea during dosing, and can be used for assisting the development of plant gray mold control medicines.
Description
Technical Field
The invention belongs to the technical field of image analysis and microorganism growth analysis, and particularly relates to a botrytis cinerea control effect analysis method and device based on machine vision.
Background
Botrytis cinerea is a pathogenic fungus of crops with wide distribution, can infect various crops including vegetables, fruit trees, flowers and the like, and can cause gray mold of the plants after the botrytis cinerea infects the plants. Gray mold is one of main fungal diseases which damage crops, 470 plants such as Solanaceae, cucurbitaceae, rosaceae and the like can be damaged, gray villous hypha and spores can grow on the surfaces of the plants when the plants are affected by the gray mold, and then the fruits, leaves and flowers of the plants are softened and rotten. Meanwhile, botrytis cinerea can infect a plurality of organs such as stems, leaves, flowers, fruits and the like of plants and simultaneously generate symptoms, so that the control of gray mold becomes important and difficult. Gray mold is a main disease of important cash crops such as vegetables, fruits, flowers and the like in agriculture, is easy to burst under the conditions of high temperature and high humidity, has very high propagation speed and strong pathogenicity, and can greatly influence the yield of the cash crops. With the adjustment of agricultural planting structures, the popularization and application of planting technologies of artificial environment facilities such as greenhouses, greenhouses and the like, the central planting environment with high temperature and high humidity is warmed, so that fungal diseases are more difficult to control.
At present, in the prevention and treatment measures of gray mold, chemical prevention and treatment are still mainly carried out, and difenoconazole, myclobutanil, tebuconazole, triflumizole and pyraclostrobin are all commonly used for preventing and treating fungal diseases of crops, such as mancozeb, fluazinam, triadimefon, pyrimethanil, procymidone, iprodione and the like. The prevention and control effect of the preventive and protective agent represented by mancozeb is limited, and the single use of the preventive and protective agent cannot meet the prevention and control requirements of users on diseases. The systemic bactericide represented by difenoconazole can inhibit the growth and invasion of pathogenic bacteria, and can also influence the physiological indexes of crops to a certain extent, and is generally characterized in that chlorophyll synthesis is inhibited, the growth of crops is slow, and the resistance of plants to pathogenic bacteria can be reduced after long-term use. Meanwhile, the long-time application of chemical agents further causes the problems of enhanced drug resistance of pathogenic bacteria, pesticide residues, environmental pollution and the like. The prevention and control of fungal diseases can enter into vicious circle which is mutually restricted by environmental influence problem, crop health problem and disease prevention and control problem.
As a natural source bactericidal active substance, the natural product bactericidal agent (plant source bactericidal agent) has the advantages of low toxicity, good diversity of attack targets and easy degradation. However, at present, the types of plant-source bactericides are still limited to a few existing varieties, the popularization and the application are delayed, and meanwhile, application short plates such as slow onset of action, short duration and the like still exist. In consideration of many factors such as research and development cost, personnel investment and the like, natural active material screening is still limited to manual evaluation, screening flux is low, accuracy is greatly influenced by manual factors, and product research and development efficiency of natural product bactericides is hindered to a certain extent.
Disclosure of Invention
In order to accelerate the research progress of natural product bactericidal active ingredients, the invention provides a botrytis cinerea control effect analysis method and device based on machine vision. According to the invention, the effect of inhibiting the growth of botrytis cinerea, which is achieved by natural product bactericides under different condition factors, is studied, the growth of botrytis cinerea under each condition factor is observed and collected, and experimental data are accurately analyzed through a strain growth control effect network prediction model.
The invention discloses a machine vision-based botrytis cinerea control effect analysis method, which comprises the following steps:
step one: placing a culture dish of Botrytis cinerea on an observation device, shooting and collecting a Botrytis cinerea culture image in the culture dish through a camera, and recording condition factors in the observation device, wherein the condition factors comprise illumination intensity, temperature, nutrient components of the culture dish and drop-added medicament concentration;
step two: labeling the botrytis cinerea culture image, and expanding the labeled botrytis cinerea culture image sample into a training data set, wherein the training data set comprises a training set, a verification set and a test set;
step three: processing a botrytis cinerea culture image of a training data set through a semantic segmentation model, segmenting out the position of a bacterial colony in a culture dish and the shape of a bacterial colony, calculating the size of the culture dish in the botrytis cinerea culture image, calculating bacterial colony size data (comprising bacterial colony radius and bacterial colony area) in the culture dish according to the size of the culture dish and the proportional relation between the bacterial colony and the culture dish, and taking the bacterial colony size data as a label of a bacterial colony segmentation image;
step four: taking illumination intensity, temperature and nutrient components as RGB values of an RGB color image, taking medicament concentration as intensity values of the RGB color image, converting the RGB color image, taking the obtained RGB color image and a labeled colony segmentation image as a strain growth prevention network prediction model for input, and predicting the growth and development conditions of colonies under various condition factors by using the strain growth prevention network prediction model; the strain growth control effect network prediction model is obtained by adding a channel attention mechanism and combining a multi-head self-attention module (SE_MHSA module) and a multi-scale feature extractor to improve the RepVGG network training, wherein the channel attention mechanism and the multi-head self-attention module are combined;
step five: and (3) establishing a botrytis cinerea growth curve according to the prediction result of the step four, and judging the antibacterial efficiency of each condition factor on botrytis cinerea through the colony radius and the colony area of each time point.
Further preferably, the strain growth control effect network prediction model is composed of a RepVGG network, a channel attention mechanism, a multi-head self-attention module and a multi-scale feature extractor, wherein the output of the RepVGG network enters the channel attention mechanism, the multi-head self-attention module and then the multi-scale feature extractor.
Further preferably, the RepVGG network sequentially comprises three-stage convolution groups, each stage of convolution group is composed of 3×3 two-dimensional convolution, 1×1 two-dimensional convolution and a Relu activation function, in each stage of convolution group, the 3×3 two-dimensional convolution and the 1×1 two-dimensional convolution respectively carry out convolution processing on input features, then fusion is carried out, the fusion features are processed through the Relu activation function and then serve as output, the output enters the next convolution group, and the output of the last convolution group enters a channel attention mechanism and combines a multi-head self-attention module.
Further preferably, the multi-scale feature extractor is comprised of a plurality of convolution levels.
Further preferably, in the fourth step, a strain growth control effect network prediction model is trained by using a training data set, and condition factors are input into the trained strain growth control effect network prediction model to obtain predicted colony size data.
Further preferably, according to the predicted colony size data, the illumination intensity, the temperature, the nutrient components and the medicament concentration are combined, the influence condition of a strain growth control effect network prediction model on the colony area under various condition factors is researched through a saproline (shape) value algorithm, and an ash botrytis cinerea growth control effect analysis chart is output.
Further preferably, the specific formula of the saprolitic (shape) value algorithm is:
；
wherein,representing Condition factors->Influence factor on the growth status of Botrytis cinerea, < >>Representation->A set of individual condition factors,/->=/>Condition factor set->Is->But does not include the conditional factors +.>，/>Representing Condition factor set +.>The number of conditional factors involved,/->Representing the number of all conditional factors>Representing Condition factor set +.>Growth size of lower colony, ++>Representing Condition factor set +.>Binding to Condition factors->Colony growth size thereafter.
The invention provides a machine vision-based botrytis cinerea control effect analysis device, which comprises a base, an observation device, a camera module and a data processing module, wherein the data processing module performs data processing according to the machine vision-based botrytis cinerea control effect analysis method.
Further preferably, the base is provided with pulleys, and the observation device and the camera module are respectively arranged on the base through telescopic supporting rods.
Further preferably, the observation device comprises a botrytis cinerea observation box, an observation table, a light supplementing lamp, a light intensity sensor and a temperature sensor are arranged in the botrytis cinerea observation box, the observation table is used for placing a culture dish, the light supplementing lamp is used for adjusting illumination intensity and a light irradiation angle, the light intensity sensor is used for measuring the illumination intensity in the botrytis cinerea observation box in real time, and the temperature sensor is used for measuring the temperature in the observation box in real time.
Further preferably, the camera module comprises an industrial camera and a camera support, and the camera of the industrial camera is used for accurately shooting Botrytis cinerea culture dishes at different heights and multiple angles.
According to the invention, the growth of the botrytis cinerea under various conditions is automatically observed through machine vision, condition factors are collected, RGB color image conversion is carried out according to illumination intensity, temperature, nutritional ingredients and medicament concentration, then colony segmentation images are used as the input of a strain growth control effect network prediction model, experimental data are accurately analyzed, and a saproline (shape) value algorithm is provided for researching the influence condition of the saproline on the colony area under various condition factors, and a botrytis cinerea growth control effect analysis chart is output. The invention establishes a natural product active bactericidal substance evaluation system with high efficiency, accuracy and low cost, and simultaneously provides guidance for the establishment of a medicament control system by introducing environmental parameters, thereby realizing rapid new medicament development and application guidance for plant gray mold.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings that are needed in the embodiments of the present invention will be briefly described below, and it is obvious that the drawings described below are only some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a machine vision-based botrytis cinerea control effect analysis method.
FIG. 2 is a diagram of a model for predicting the growth and prevention effect network and analysis of the prevention effect.
FIG. 3 is a graph showing the analysis of the growth control effect of Botrytis cinerea obtained by the present invention.
FIG. 4 is a schematic diagram of the machine vision-based botrytis cinerea control effect analysis device.
Detailed Description
The present invention will be described in further detail with reference to the following examples in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
As shown in FIG. 1, the machine vision-based botrytis cinerea control effect analysis method comprises the following steps:
step one: placing a culture dish of botrytis cinerea on an observation device, shooting and collecting a botrytis cinerea culture image in the culture dish through a camera, and recording condition factors in the observation device; the condition factors comprise illumination intensity, temperature, nutrient components of the culture dish and the concentration of the dropwise added medicament;
step two: labeling the botrytis cinerea culture image, and expanding the labeled botrytis cinerea culture image sample into a training data set, wherein the training data set comprises a training set, a verification set and a test set; wherein, the botrytis cinerea culture image, illumination intensity, temperature, nutrition components and medicament concentration are transmitted to a computer in real time in a wireless local area network or wired connection mode;
step three: processing a botrytis cinerea culture image of a training data set through a semantic segmentation model, segmenting out the position of a bacterial colony in a culture dish and the shape of a bacterial colony, calculating the size of the culture dish in the botrytis cinerea culture image, calculating bacterial colony size data (comprising bacterial colony radius and bacterial colony area) in the culture dish according to the size of the culture dish and the proportional relation between the bacterial colony and the culture dish, and taking the bacterial colony size data as a label of a bacterial colony segmentation image; the semantic segmentation model of the embodiment is a Unet semantic segmentation model built in a computer through a Pytorch framework;
step four: taking illumination intensity, temperature and nutrient components as RGB values of an RGB color image, taking medicament concentration as intensity values of the RGB color image, converting the RGB color image, taking the obtained RGB color image and a labeled colony segmentation image as a strain growth prevention network prediction model for input, and predicting the growth and development conditions of colonies under various condition factors by using the strain growth prevention network prediction model; the strain growth control effect network prediction model is obtained by adding a channel attention mechanism and combining a multi-head self-attention module (SE_MHSA module) to improve a RepVGG network training, and the SE_MHSA module is formed by combining a SE module (Squeeze and Excitation) and a multi-head self-attention module (MultiHead SelfAttention); the SE module is mainly used for weighting the characteristic channels and is used for enhancing the attention degree of the network model to different characteristic channels, and the multi-head self-attention module (MultiHead SelfAttention) is mainly used for capturing long-distance dependency relations among input characteristics. By combining the two modules, the expression capacity of the model can be enhanced at different levels, so that the strain growth control effect network prediction model has higher analysis accuracy; the strain growth control effect network prediction model is built by a computer through a Pytorch frame;
step five: and (3) establishing a botrytis cinerea growth curve according to the prediction result of the step four, and judging the antibacterial efficiency of each condition factor on botrytis cinerea through the colony radius and the colony area of each time point. According to the time point or the culture time of the experiment, the colony area is correlated with the observation record time, and a Botrytis cinerea growth curve is established. The botrytis cinerea growth curve is a curve graph of the change of the colony area along with time, and the growth and development conditions of the colony in each time period can be more intuitively and specifically observed through the botrytis cinerea growth curve. Meanwhile, a growth threshold can be set, and if the colony area is continuously increased within a period of time and exceeds the set threshold, the colony area is judged to be normal in growth and development; if the colony area grows slowly or stagnates and does not reach the set threshold, it is determined that the growth is inhibited, that is, the natural product type germicide (plant-derived germicide) is effective.
According to the embodiment, the bacterial colony size data of the botrytis cinerea in the culture dish can be identified and calculated according to the camera imaging principle:
；
wherein,indicated are radius of the dish, +.>Denoted focal length of camera, +.>The distance of the camera to the dish is shown, and the radius of the dish in the Botrytis cinerea culture image is calculated by the known or measurable parameters>：
；
Then, recognizing and dividing the outline of the culture dish and the outline of the botrytis cinerea colony in the botrytis cinerea culture image by a semantic division model, and calculating the radius ratio of the culture dish and the colony according to the obtained division imageThus dividing the colony size of Botrytis cinerea in the image +.>The method comprises the following steps:
；
the actual radius of the colony can be obtained：
；
Then from the actual radius gauge of the bacterial colonyCalculating colony area：
。
As shown in FIG. 2, the strain growth prevention network prediction model is composed of a RepVGG network, a channel attention mechanism, a multi-head self-attention module (SE_MHSA module) and a multi-scale feature extractor, wherein the input of the RepVGG network is a labeled colony segmentation image and an RGB color image formed by conversion of illumination intensity, temperature, nutrition components and medicament concentration, and the output of the RepVGG network enters the channel attention mechanism, the multi-head self-attention module (SE_MHSA module) and then enters the multi-scale feature extractor. The RepVGG network sequentially comprises three-level convolution groups, each level convolution group consists of 3X 3 two-dimensional convolution, 1X 1 two-dimensional convolution and a Relu activation function, in each level convolution group, the 3X 3 two-dimensional convolution and the 1X 1 two-dimensional convolution respectively carry out convolution processing on input features, then fusion is carried out, the fusion features are processed through the Relu activation function and then serve as output, the output enters the next convolution group, and the output of the last convolution group enters a channel attention mechanism and combines a multi-head self-attention module (SE_MHSA module).
The multi-scale feature extractor is comprised of a plurality of convolution levels. Each level is capable of extracting features of a particular scale from the input image. While the convolution kernel size and stride are different for each level, thereby making the receptive field size different for each level. Larger receptive fields facilitate capturing global or large scale features, and smaller receptive fields facilitate local or small scale features. Meanwhile, the multi-scale feature extractor further comprises a feature fusion layer, feature graphs with different scales are fused together, and multi-scale information can be reserved in higher feature representation.
After training by a strain growth control effect network prediction model, inputting condition factors into the trained strain growth control effect network prediction model to obtain predicted colony size data, wherein the aim is to improve the interpretation of a neural network by using a saproline (shape) value algorithm in the field of game theory, and inputting four parameters for experiments: the influence of the light intensity, the temperature, the nutrient components and the medicament concentration on the colony area under various condition factors is researched, and an ash botrytis growth prevention effect analysis chart is output.
The specific formula of the saprolitic (shape) value algorithm is:
；
wherein,representing Condition factors->Influence factor on the growth status of Botrytis cinerea, < >>Representation->A set of individual condition factors,/->=/>Condition factor set->Is->But does not include the conditional factors +.>，/>Representing Condition factor set +.>IncludedThe number of conditional factors of->Representing the number of all conditional factors>Representing Condition factor set +.>Growth size of lower colony, ++>Representing Condition factor set +.>Binding to Condition factors->Colony growth size thereafter.
As shown in the figure 3, the horizontal axis of the figure shows a saprolitic value (SHAP value), and a SHAP value greater than 0 indicates that the factor promotes the predicted value of the colony area, has the effect of promoting the growth of bacterial cakes, a SHAP value smaller than 0 indicates that the factor reduces the predicted value of the colony area and plays the effect of inhibiting the growth of bacterial cakes, and a larger absolute value of the SHAP value indicates that the condition factor has a larger influence on the growth of bacterial cakes. The vertical axis indicates the influence of each condition factor on colony growth, and the higher the characteristic value of a certain condition factor, the larger the influence of the condition factor on colony growth and development. In the experiment of the method, the nutrient components of the culture dish have the greatest influence on the colony area, and the illumination intensity, the medicament concentration and the temperature are respectively adopted.
The training data set of this embodiment includes training set 525 row data and verification set 175 row data, and two main evaluation indexes such as training Loss (Loss) and Mean Absolute Error (MAE) are selected, where the mean absolute error can measure the mean absolute value of the prediction error of the network model, so that the smaller the value, the better the value. And (3) uniformly running 1000 rounds of the improved network of different RepVGGs to obtain evaluation index data. The two evaluation index comparison data for different RepVGG modified networks after run using the training dataset are shown in table 1.
Table 1 evaluation index contrast data for different RepVGG improved networks
In the table, ECA stands for effective channel attention mechanism; CBAM stands for convolutional block attention module (channel attention mechanism and spatial attention mechanism are used in mixture); the SE MHSA module represents a channel attention mechanism in combination with a multi-headed self-attention module.
As shown in FIG. 4, the machine vision-based Botrytis cinerea control effect analysis device provided by the embodiment comprises a base 1, an observation device 2, a camera module 3 and a data processing module 4, wherein the base 1 is provided with pulleys, so that the whole device is convenient to move. The observation device 2 and the camera module 3 are respectively arranged on the base 1 through telescopic supporting rods, so that the height can be conveniently adjusted. The observation device 2 comprises an ash botrytis observation box, an observation table, a light supplementing lamp, a light intensity sensor and a temperature sensor are arranged in the ash botrytis observation box, the observation table is used for placing a culture dish, the light supplementing lamp can adjust illumination intensity and an angle of light irradiation, the light intensity sensor can measure the illumination intensity in the ash botrytis observation box in real time, and the temperature sensor is used for measuring the temperature in the observation box in real time. The camera module 3 comprises an industrial camera and a camera bracket, and a camera of the industrial camera shoots the Botrytis cinerea culture dish accurately at different heights and multiple angles. The data processing module 4 is completed by a computer, and the data processing module 4 performs data processing according to the botrytis cinerea control effect analysis method based on machine vision.
The above-described invention is merely representative of embodiments of the present invention and should not be construed as limiting the scope of the invention, nor any limitation in any way as to the structure of the embodiments of the present invention. It should be noted that it will be apparent to those skilled in the art that various changes and modifications can be made herein without departing from the spirit and scope of the invention as defined by the appended claims.
Claims (10)
1. The machine vision-based botrytis cinerea control effect analysis method is characterized by comprising the following steps of:
step one: placing a culture dish of Botrytis cinerea on an observation device, shooting and collecting a Botrytis cinerea culture image in the culture dish through a camera, and recording condition factors in the observation device, wherein the condition factors comprise illumination intensity, temperature, nutrient components of the culture dish and drop-added medicament concentration;
step two: labeling the botrytis cinerea culture image, and expanding the labeled botrytis cinerea culture image sample into a training data set, wherein the training data set comprises a training set, a verification set and a test set;
step three: processing the botrytis cinerea culture image of the training data set through a semantic segmentation model, segmenting out the position of the bacterial colony in the culture dish and the shape of the bacterial colony, calculating the size of the culture dish in the botrytis cinerea culture image, calculating the bacterial colony size data in the culture dish according to the size of the culture dish and the proportional relation between the bacterial colony and the culture dish, and taking the bacterial colony size data as a label of the bacterial colony segmentation image;
step four: taking illumination intensity, temperature and nutrient components as RGB values of an RGB color image, taking medicament concentration as intensity values of the RGB color image, converting the RGB color image, taking the obtained RGB color image and a labeled colony segmentation image as a strain growth prevention network prediction model for input, and predicting the growth and development conditions of colonies under various condition factors by using the strain growth prevention network prediction model; the strain growth control effect network prediction model is obtained by adding a channel attention mechanism and combining a multi-head self-attention module and a multi-scale feature extractor to improve the RepVGG network training, wherein the channel attention mechanism and the multi-head self-attention module are combined by a SE module and a multi-head self-attention module;
step five: and (3) establishing a botrytis cinerea growth curve according to the prediction result of the step four, and judging the antibacterial efficiency of each condition factor on botrytis cinerea through the colony radius and the colony area of each time point.
2. The machine vision-based botrytis cinerea control effect analysis method according to claim 1, wherein the strain growth control effect network prediction model is composed of a RepVGG network, a channel attention mechanism, a multi-head self-attention module and a multi-scale feature extractor, wherein the output of the RepVGG network enters the channel attention mechanism, combines with the multi-head self-attention module and then enters the multi-scale feature extractor.
3. The machine vision-based botrytis cinerea control effect analysis method according to claim 2, wherein the RepVGG network sequentially comprises three-level convolution groups, each level of convolution group comprises 3×3 two-dimensional convolution, 1×1 two-dimensional convolution and a Relu activation function, in each level of convolution group, the 3×3 two-dimensional convolution and the 1×1 two-dimensional convolution respectively carry out convolution processing on input features, then fusion is carried out, the fusion features are processed by the Relu activation function and then serve as output, the output enters the next convolution group, and the output of the last convolution group enters a channel attention mechanism to be combined with a multi-head self-attention module.
4. The machine vision based botrytis cinerea control effect analysis method of claim 2, wherein the multi-scale feature extractor consists of a plurality of convolution levels.
5. The machine vision-based botrytis cinerea control effect analysis method according to claim 1, wherein in the fourth step, a strain growth control effect network prediction model is trained by using a training data set, and condition factors are input into the trained strain growth control effect network prediction model to obtain predicted colony size data.
6. The machine vision-based botrytis cinerea control effect analysis method based on claim 5, wherein according to predicted colony size data, the influence condition of a strain growth control effect network prediction model on colony areas under various condition factors is studied through a saproline value algorithm in combination with illumination intensity, temperature, nutritional ingredients and medicament concentration, and a botrytis cinerea growth control effect analysis chart is output.
7. The machine vision based botrytis cinerea control effect analysis method of claim 6, wherein the specific formula of the saprolitic value algorithm is:
；
wherein,representing Condition factors->Influence factor on the growth status of Botrytis cinerea, < >>Representation->A set of individual condition factors,/->=/>Condition factor set->Is->But does not include the conditional factors +.>，/>Representing Condition factor set +.>The number of conditional factors involved,/->Representing the number of all conditional factors>Representing Condition factor set +.>Growth size of lower colony, ++>Representing Condition factor set +.>Binding to Condition factors->Colony growth size thereafter.
8. The botrytis cinerea control effect analysis device based on machine vision is characterized by comprising a base, an observation device, a camera module and a data processing module, wherein the data processing module processes data according to the botrytis cinerea control effect analysis method based on machine vision as set forth in any one of claims 1-7.
9. The machine vision-based botrytis cinerea control effect analysis device according to claim 8, wherein the base is provided with pulleys, the observation device and the camera module are respectively arranged on the base through telescopic supporting rods, the camera module comprises an industrial camera and a camera support, and a camera of the industrial camera is used for accurately shooting botrytis cinerea culture dishes at different heights and multiple angles.
10. The machine vision-based botrytis cinerea control effect analysis device according to claim 8, wherein the observation device comprises a botrytis cinerea observation box, an observation table, a light supplementing lamp, a light intensity sensor and a temperature sensor are arranged in the botrytis cinerea observation box, the observation table is used for placing a culture dish, the light supplementing lamp is used for adjusting illumination intensity and an angle of light irradiation, the light intensity sensor is used for measuring the illumination intensity in the botrytis cinerea observation box in real time, and the temperature sensor is used for measuring the temperature in the observation box in real time.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311301316.XA CN117037152B (en) | 2023-10-10 | 2023-10-10 | Machine vision-based botrytis cinerea control effect analysis method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311301316.XA CN117037152B (en) | 2023-10-10 | 2023-10-10 | Machine vision-based botrytis cinerea control effect analysis method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117037152A CN117037152A (en) | 2023-11-10 |
| CN117037152B true CN117037152B (en) | 2024-01-05 |
Family
ID=88626729
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311301316.XA Active CN117037152B (en) | 2023-10-10 | 2023-10-10 | Machine vision-based botrytis cinerea control effect analysis method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117037152B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102628794A (en) * | 2012-04-19 | 2012-08-08 | 江苏大学 | Method for quickly measuring total quantity of livestock meat bacteria based on hyperspectral imaging technology |
| CN111368643A (en) * | 2020-02-12 | 2020-07-03 | 杭州电子科技大学 | Escherichia coli dynamic growth monitoring method |
| CN111415709A (en) * | 2020-03-16 | 2020-07-14 | 北京君立康生物科技有限公司 | Method and device for predicting colony growth image, electronic device and storage medium |
| CN112116599A (en) * | 2020-08-12 | 2020-12-22 | 南京理工大学 | Sputum smear tubercle bacillus semantic segmentation method and system based on weak supervised learning |
| KR20210035948A (en) * | 2019-09-24 | 2021-04-02 | (주)다울 | Growth analysis system based on machine learning |
| CN115761225A (en) * | 2022-11-03 | 2023-03-07 | 北京科技大学 | Image annotation method based on neural network interpretability |
-
2023
- 2023-10-10 CN CN202311301316.XA patent/CN117037152B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102628794A (en) * | 2012-04-19 | 2012-08-08 | 江苏大学 | Method for quickly measuring total quantity of livestock meat bacteria based on hyperspectral imaging technology |
| KR20210035948A (en) * | 2019-09-24 | 2021-04-02 | (주)다울 | Growth analysis system based on machine learning |
| CN111368643A (en) * | 2020-02-12 | 2020-07-03 | 杭州电子科技大学 | Escherichia coli dynamic growth monitoring method |
| CN111415709A (en) * | 2020-03-16 | 2020-07-14 | 北京君立康生物科技有限公司 | Method and device for predicting colony growth image, electronic device and storage medium |
| CN112116599A (en) * | 2020-08-12 | 2020-12-22 | 南京理工大学 | Sputum smear tubercle bacillus semantic segmentation method and system based on weak supervised learning |
| CN115761225A (en) * | 2022-11-03 | 2023-03-07 | 北京科技大学 | Image annotation method based on neural network interpretability |
Non-Patent Citations (3)
| Title |
|---|
| Computer vision and machine learning applied in the mushroom industry: a critical review;Hua Yin等;Computers and Electronics in Agriculture;第1-15页 * |
| 基于光谱和图像信息融合的玉米霉变程度在线检测;沈飞等;食品科学;第40卷(第16期);第274-280页 * |
| 基于迁移学习的单目菌落深度提取算法;邓相舟;张荣福;;光学仪器(02);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117037152A (en) | 2023-11-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Ferrante et al. | Agronomic management for enhancing plant tolerance to abiotic stresses: High and low values of temperature, light intensity, and relative humidity | |
| Trivellini et al. | Pitaya, an attractive alternative crop for Mediterranean region | |
| Opara et al. | Quantification of on-farm pomegranate fruit postharvest losses and waste, and implications on sustainability indicators: South African case study | |
| Islam | Plant disease detection using CNN model and image processing | |
| Gulavnai et al. | Deep learning for image based mango leaf disease detection | |
| Haque et al. | Planting arrangement and effects of planting density on tropical fruit crops—A Review | |
| CN107619853A (en) | A kind of wheat scab resistance authentication method | |
| Peavey et al. | Effect of shading on red colour and fruit quality in blush pears “ANP-0118” and “ANP-0131” | |
| Szot et al. | Effect of the Climatic Conditions in Central Europe on the Growth and Yield of Cornelian Cherry Cultivars | |
| Zhu et al. | Machine learning methods for efficient and automated in situ monitoring of peach flowering phenology | |
| Jovanović-Cvetković et al. | Influence of interaction effects of the different pollenizers on the Blatina variety (Vitis vinifera L.) grape cluster and seed characteristics | |
| CN117037152B (en) | Machine vision-based botrytis cinerea control effect analysis method and device | |
| Lin et al. | Data-driven modeling for crop growth in plant factories | |
| Wu et al. | Research progress on efficient pollination technology of crops | |
| Bieniasz et al. | The effect of titanium organic complex on pollination process and fruit development of apple cv. topaz | |
| CN116636423B (en) | Efficient cultivation method of poria cocos strain | |
| Xu et al. | Horticultural practices in early spring to mitigate the adverse effect of low temperature on fruit set in ‘Lapins’ sweet cherry | |
| Alshallash et al. | GA3 and hand thinning improves physical, chemical characteristics, yield and decrease bunch compactness of Sultanina grapevines (Vitis vinifera L.) | |
| Lee et al. | Temperature changes affected spring phenology and fruit quality of apples grown in high-latitude region of South Korea | |
| Cabrera et al. | Papaya (Carica papaya L.) phenology under different agronomic conditions in the subtropics | |
| Williams et al. | Evaluating the effects of early pruning, leaf removal, and shoot thinning on ‘MidSouth’grapes over two consecutive vintages in South Mississippi | |
| Lustosa da Silva et al. | Breeding short-day strawberry genotypes for cultivation in tropical and subtropical regions | |
| Tombesi et al. | Grapevine response to stress generated by excessive temperatures during the budburst | |
| Zhao et al. | Screening and evaluation of excellent blackberry cultivars and strains based on nutritional quality, antioxidant properties, and genetic diversity | |
| Nicolosi et al. | Phenotypic plasticity in bud fruitfulness expressed in two distinct wine grape cultivars grown under three different pedoclimatic conditions |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |