CN112419288B - Unmanned vegetable greenhouse planting method based on computer vision - Google Patents
Unmanned vegetable greenhouse planting method based on computer vision Download PDFInfo
- Publication number
- CN112419288B CN112419288B CN202011368215.0A CN202011368215A CN112419288B CN 112419288 B CN112419288 B CN 112419288B CN 202011368215 A CN202011368215 A CN 202011368215A CN 112419288 B CN112419288 B CN 112419288B
- Authority
- CN
- China
- Prior art keywords
- potato
- image
- disease
- layer
- picture
- 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
- 235000013311 vegetables Nutrition 0.000 title claims abstract description 51
- 238000000034 method Methods 0.000 title claims abstract description 29
- 201000010099 disease Diseases 0.000 claims abstract description 62
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 claims abstract description 62
- 238000012545 processing Methods 0.000 claims abstract description 26
- 230000005540 biological transmission Effects 0.000 claims abstract description 10
- 230000008054 signal transmission Effects 0.000 claims abstract description 5
- 235000002595 Solanum tuberosum Nutrition 0.000 claims description 56
- 244000061456 Solanum tuberosum Species 0.000 claims description 56
- 238000013527 convolutional neural network Methods 0.000 claims description 20
- 238000011176 pooling Methods 0.000 claims description 12
- 238000012549 training Methods 0.000 claims description 11
- 238000012360 testing method Methods 0.000 claims description 9
- 238000013135 deep learning Methods 0.000 claims description 7
- 238000011156 evaluation Methods 0.000 claims description 7
- 206010035148 Plague Diseases 0.000 claims description 6
- 241000607479 Yersinia pestis Species 0.000 claims description 6
- 238000004383 yellowing Methods 0.000 claims description 6
- 238000000605 extraction Methods 0.000 claims description 5
- 235000012015 potatoes Nutrition 0.000 claims description 5
- 238000001914 filtration Methods 0.000 claims description 4
- 238000002372 labelling Methods 0.000 claims description 3
- 238000011160 research Methods 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 2
- 230000008569 process Effects 0.000 claims description 2
- 239000000284 extract Substances 0.000 claims 3
- 238000012935 Averaging Methods 0.000 claims 1
- 238000013499 data model Methods 0.000 claims 1
- 238000001514 detection method Methods 0.000 abstract description 15
- 238000004458 analytical method Methods 0.000 abstract 1
- 238000010191 image analysis Methods 0.000 abstract 1
- 230000006870 function Effects 0.000 description 9
- 238000005516 engineering process Methods 0.000 description 5
- 241000196324 Embryophyta Species 0.000 description 3
- 241000282414 Homo sapiens Species 0.000 description 2
- 230000004913 activation Effects 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 230000007170 pathology Effects 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- 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
- G06N3/045—Combinations of networks
-
- 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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20024—Filtering details
- G06T2207/20032—Median filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A40/00—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production
- Y02A40/10—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production in agriculture
- Y02A40/25—Greenhouse technology, e.g. cooling systems therefor
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
Aiming at the problem that the existing vegetable greenhouse is difficult to judge diseases, the invention provides an unmanned vegetable greenhouse planting method based on computer vision, which comprises the following steps: the image information acquisition and transmission device, the upper computer processing end and the early warning display device; the image signal acquisition and transmission device comprises a picture positioning acquisition device and a wireless signal transmission device; the early warning display device comprises a disease prompt device and a disease display device. The image signal acquisition device acquires the vegetable image on the vegetable site and carries out wireless transmission, the upper computer processing unit receives the image signal, then carries out analysis and processing, and obtains a final result, and the final result is displayed on the early warning display device through the control of the upper computer. The unmanned vegetable greenhouse planting method adopts computer vision to carry out image analysis on vegetables and establish a vegetable disease detection model, and compared with the traditional manual interpretation, the model has higher efficiency; the accuracy is higher, and the recognition speed is faster.
Description
Technical Field
The invention relates to the technical field of agricultural vegetable planting, in particular to a greenhouse method for intelligently judging diseases based on computer vision.
Background
Any crop yield is related to its disease, and untimely disease treatment can lead to severe yield slip, which is more important for vegetables. At present, vegetable planting is mostly carried out by adopting greenhouse management technology, improving production technology, strictly controlling vegetable temperature, humidity and illumination within proper ranges, thereby obtaining better quality products and a thicker report. Along with the progress of social science, the vegetable greenhouse is slowly developed from original manual management to unmanned automatic management, and environmental conditions in the greenhouse are automatically and effectively controlled and monitored, so that the labor cost is reduced, the working efficiency is improved, and the yield of vegetables is further increased.
Although there are many patent technologies related to vegetable greenhouses, there are few technologies related to disease detection and treatment. Diseased vegetables may exhibit varying degrees of characteristics in terms of color, texture, etc. The machine is used for replacing manual work to judge diseases, so that the cost of the vegetable greenhouse can be further reduced, the working efficiency is further improved, and the unmanned automatic management in the true sense is realized. Therefore, a vegetable greenhouse planting method for automatically monitoring diseases is urgently needed in the market.
With the development of artificial intelligence, computer vision can be used for extracting vegetable characteristics and judging diseases, and the accuracy of judgment is further improved due to the introduction of deep learning. The convolutional neural network is used as a high-performance image recognition technology and provides a new idea for accurately detecting and recognizing vegetable diseases.
Disclosure of Invention
The invention aims to provide a vegetable greenhouse planting method capable of distinguishing vegetable diseases, and provides a target device and a detection method for acquiring and detecting the vegetable diseases, so that the problem that the existing vegetable greenhouse lacks an automatic disease detection system of a system is solved, and the labor is further relieved.
The invention relates to an unmanned vegetable greenhouse planting method based on computer vision, which comprises the following steps: the image information acquisition and transmission device, the upper computer processing end and the early warning device;
the image signal acquisition and transmission device comprises: and the picture positioning and collecting device and the wireless signal transmitting device.
The picture positioning and collecting device comprises a camera, a rotating shaft motor, a position sensor and a gear track; the rotating shaft motor drives the gear track to move so as to drive the camera on the track to move, and when the position sensor area where the vegetables are located is reached, the image positioning and acquisition is stopped.
The wireless signal transmitting device transmits the acquired picture information to the upper computer processing end of the main control through the wireless signal transmission bus, the next picture positioning acquisition is carried out after the processing is successful, the picture information transmitted by the wireless signal transmitting device and received by the upper computer processing end is analyzed in a vegetable disease detection model established by utilizing a convolutional neural network and a deep learning mode to obtain a corresponding prediction result, when the upper computer processing end detects that vegetables are diseased and gives out disease types, the disease display device can automatically display the disease types, and the disease prompt device can be started to inform a shed owner of adopting corresponding processing measures in time.
The disease display device consists of classifying small lamps under each plant of vegetables, and the corresponding small lamps can display after judging the disease types, so that measures are taken by a shed owner conveniently.
The disease vegetable detection method of the invention is introduced by taking potato as an example, and an internal processing mechanism of an upper computer processing end establishes a potato disease detection model through a convolutional neural network and deep learning, and comprises the following specific steps:
(1) Collecting a large amount of potato leaf pictures of five common diseases including 'potato plague', 'ring spot', 'potato yellowing leaf curl', 'potato wilt', 'potato anthracnose' and a healthy and normal state potato leaf picture;
(2) Classifying and labeling the six states of the collected potato images, and establishing a training set and a testing set required by a model training sequence by a data set according to a 9:1 relation.
(3) The potato image is preprocessed, which comprises the following steps:
3a) Image cutting, namely removing redundant parts by cutting and reserving a research theme;
3b) Adjusting the resolution of the image, and adjusting the original image to 800 x 600 pixel resolution under the condition of ensuring that the proportion of the image is unchanged and the DPI is unchanged;
3c) The invention adopts pixel average method to convert color RGB color image into gray image, namely, R, G, B pixel points at corresponding positions are averaged to obtain the pixel point at the position, the formula is:
3d) The invention adopts a simple median filtering method to carry out nonlinear processing on the image. Let g, g (i, j) be the pixel point with coordinates (i, j) in the image g, S be the field space of the pixel point for obtaining K pixels by using a given template, and arrange the pixels from small to large according to the gray value and take the median as the new gray value, the formula is:
g’(i,j)＝Med{g(x,y),(x,y)∈S}
(4) The invention preferably uses a convolutional neural network to extract and identify the characteristics of the processed picture, and the basic structure of the convolutional neural network comprises a convolutional layer, a pooling layer and a full-connection layer.
Furthermore, the invention establishes a depth convolution neural network structure with the depth of 10, which comprises 3 convolution layers, 3 maximum pooling layers, 2 full connection layers, a softmax layer and an output layer;
Furthermore, the invention adopts 3 different convolution kernels in 3 layers of convolution layers, wherein one layer is used for extracting edge features, two layers are used for extracting contour features, and three layers are used for extracting texture features.
Furthermore, the invention adds a modified linear unit activation Function, namely a ReLU Function, after each convolution layer, and the formula is as follows:
Furthermore, the pooling layer of the invention adopts the maximum pooling algorithm capable of overlapping 2 x 2 windows.
The characteristic extraction and disease identification of potato leaves by adopting the convolutional neural network depth are output by adopting softmax, and the specific steps are as follows:
4a) The potatoes in six states are singly divided into one type, namely healthy potato leaves are in a type 1, potato plague leaves are in a type 2, ring rot are in a type 3, potato yellowing leaf curl are in a type 4, potato wilt is in a type 5, and potato anthracnose is in a type 6;
4b) When the category 1 is a positive sample, the rest categories are negative samples, namely the marking data is (1,0,0,0,0,0);
when the category 2 is a positive sample, the remaining categories are negative samples, i.e. the marking data is (0,1,0,0,0,0);
when the category 3 is a positive sample, the remaining categories are negative samples, i.e. the marking data is (0,0,1,0,0,0);
when the category 4 is a positive sample, the remaining categories are negative samples, i.e. the marking data is (0,0,0,1,0,0);
when the category 5 is a positive sample, the remaining categories are negative samples, i.e. the marking data is (0,0,0,0,1,0);
when the category 6 is a positive sample, the remaining categories are negative samples, i.e. the marking data is (0,0,0,0,0,1);
4c) The 10-layer deep convolutional neural network structure constructed by the invention is used for training the potato picture data set marked with the class to obtain the probability of the corresponding class, and the data is continuously trained by using the counter propagation principle until the final gradient is reduced to the minimum value, as shown in fig. 3, so that the probability of the predicted sample is close to 0 on the rest classes and is close to 1 on the class.
(5) And (3) taking the test set into the trained model to judge the accuracy rate of the model, namely, the test set is used for evaluating whether the trained model can be put into use.
Further, two single digital evaluation indexes are combined, namely a new evaluation mode is used:
precision ratio P: true marking of the correct proportion in all potato leaves marked with disease;
recall ratio R: for all plots, the disease duty cycle was identified;
Compared with the traditional machine learning and manual interpretation, the invention has the beneficial effects that:
(1) The method is more intelligent and automatic, any disease can not be put through, and diseases which are not noticed or missed by human beings can be detected;
(2) The device is more convenient, releases manpower and improves the working efficiency;
(3) The detection speed is faster, and the precision is more accurate.
Drawings
FIG. 1 is a schematic structural diagram of an unmanned vegetable greenhouse method based on computer vision;
FIG. 2 is a flow chart of a vegetable disease detection model;
FIG. 3 is a diagram of a back propagation prediction method;
FIG. 4 is a sample output classification chart
Detailed Description
The following describes the technical aspects of the present invention with reference to examples, but the present invention is not limited to the following examples.
Examples
The invention provides an unmanned vegetable greenhouse planting method based on computer vision, which is shown in fig. 1 and comprises the following steps: the image information acquisition and transmission device, the upper computer processing end and the early warning display device; the image information acquisition and transmission device comprises a picture positioning acquisition device and a wireless signal transmission device; the early warning display device comprises a disease prompt device and a disease display device;
The image information acquisition and transmission device is used for acquiring and transmitting images of vegetables in the greenhouse; the upper computer processing end processes and detects the picture information sent by the image information acquisition and sending device to give a corresponding interpretation result;
The picture positioning and collecting device collects picture information of each plant of vegetables from a greenhouse vegetable planting area, the collected picture signals are transmitted to the upper computer processing end through the wireless signal transmitting device, the upper computer processing end which receives the picture signals can analyze the received pictures through a vegetable disease detection model established by convolutional neural network deep learning, and a corresponding interpretation result is obtained. Judging whether the early warning equipment is started or not according to the detection result given by the processing end of the upper computer, if the corresponding pathology is detected, the disease prompting device prompts the shed owner to have the corresponding disease, and the disease display device displays the corresponding disease lamp on the corresponding vegetable patient strain.
As shown in fig. 2, the unmanned vegetable greenhouse planting method based on computer vision provided by the invention is a vegetable disease detection model established by a convolutional neural network deep learning mode, and takes potatoes as an example, and specifically comprises the following steps:
(1) Collecting a large amount of potato leaf pictures of five common diseases including 'potato plague', 'ring spot', 'potato yellowing leaf curl', 'potato wilt', 'potato anthracnose' and a healthy and normal state potato leaf picture;
(2) Classifying and labeling the six states of the collected potato images, and establishing a training set and a testing set required by a model training sequence by a data set according to a 9:1 relation.
(3) Pre-processing the potato image, comprising:
3a) Image cutting, namely removing redundant parts by cutting and reserving a research theme;
3b) Adjusting the resolution of the image, and adjusting the original image to 800 x 600 pixel resolution under the condition of ensuring that the proportion of the image is unchanged and the DPI is unchanged;
3c) The invention adopts pixel average method to convert color RGB color image into gray image, namely, R, G, B pixel points at corresponding positions are averaged to obtain the pixel point at the position, the formula is:
3d) The invention adopts a simple median filtering method to carry out nonlinear processing on the image. Let g, g (i, j) be the pixel point with coordinates (i, j) in the image g, S be the field space of the pixel point for obtaining K pixels by using a given template, and arrange the pixels from small to large according to the gray value and take the median as the new gray value, the formula is:
g’(i,j)＝Med{g(x,y),(x,y)∈S}
(4) The invention preferably uses a convolutional neural network to extract and identify the characteristics of the processed picture, and the basic structure of the convolutional neural network comprises a convolutional layer, a pooling layer and a full-connection layer.
In the invention, the model establishes a depth convolution neural network structure with the depth of 10, and comprises 3 convolution layers, 3 maximum pooling layers, 2 full connection layers, a softmax layer and an output layer; according to the invention, 3 different convolution kernels are adopted in three convolution layers, wherein one layer is for extracting edge characteristics, the other layer is for extracting main contour characteristics, and the three layers are for extracting texture characteristics; the pooling layers all employ a maximum pooling algorithm that can overlap a2 x 2 window, as shown in fig. 3.
In the invention, each convolution layer is added with a modified linear unit activation Function, namely a ReLU Function, and the formula is as follows:
furthermore, the invention adopts convolutional neural network depth to output the characteristic extraction and disease identification of potato leaves by using softmax, and comprises the following specific steps:
4a) The potatoes in six states are singly divided into one type, namely healthy potato leaves are in a type 1, potato plague leaves are in a type 2, ring rot are in a type 3, potato yellowing leaf curl are in a type 4, potato wilt is in a type 5, and potato anthracnose is in a type 6;
4b) When the category 1 is a positive sample, the rest categories are negative samples, namely the marking data is (1,0,0,0,0,0);
when the category 2 is a positive sample, the remaining categories are negative samples, i.e. the marking data is (0,1,0,0,0,0);
when the category 3 is a positive sample, the remaining categories are negative samples, i.e. the marking data is (0,0,1,0,0,0);
when the category 4 is a positive sample, the remaining categories are negative samples, i.e. the marking data is (0,0,0,1,0,0);
when the category 5 is a positive sample, the remaining categories are negative samples, i.e. the marking data is (0,0,0,0,1,0);
when the category 6 is a positive sample, the remaining categories are negative samples, i.e. the marking data is (0,0,0,0,0,1);
4c) The 10-layer deep convolutional neural network structure constructed by the invention is used for training the potato picture data set marked with the class to obtain the probability of the corresponding class, and the data is continuously trained by using the counter propagation principle until the average value of the total loss function of all samples is minimum, namely the cost function is minimum, so that the probability of the predicted sample is close to 0 on the rest classes and is close to 1 on the class. As shown in fig. 4.
The loss function in the embodiment of the invention is l= - (y ilogai+(1-yi)log(1-ai)
The cost function is:
Where y i is the class label data of the i-th sample, and a i is the output probability of the i-th sample.
Further, taking class 1 as an example, when the ith input is a healthy leaf, that is, class 1 is a positive sample, the other classes are negative samples, the labeled class data y i is (1,0,0,0,0,0), the data of the output layer is substituted into the loss function, the weight threshold is obtained by using the backward propagation gradient descent, and the data is substituted for performing the second training, and the cycle is repeated until the final curve of l is flattened.
(5) And (3) taking the test set into the trained model to judge the accuracy rate of the model, namely, the test set is used for evaluating whether the trained model can be put into use.
In the invention, a new evaluation standard is provided by combining the prior evaluation standard, and the content is as follows:
precision ratio P: true marking of the correct proportion in all potato leaves marked with disease;
Recall ratio R: for all disease graphs, identifying the duty ratio of the disease;
The new evaluation mode is as follows:
and the larger N, the better the evaluation result.
The early warning prompting device provided by the embodiment of the invention can judge whether to start or not according to the detection result given by the processing end of the upper computer. If the corresponding pathology is detected, the disease prompt device prompts the shed owner to have the corresponding disease, and the disease display device displays the corresponding disease lamp on the corresponding vegetable disease plant.
According to the vegetable disease detection model in the unmanned vegetable greenhouse planting method based on computer vision, the convolutional neural network deep learning mode is adopted for construction, the detection effect accuracy and speed of the model are superior to those of machine learning, the efficiency is far higher than that of human beings, the first time identification can be achieved, and the labor cost can be reduced.
The present invention may be better implemented as described above, and the above examples are merely illustrative of preferred embodiments of the present invention and not intended to limit the scope of the present invention, and various changes and modifications made by those skilled in the art without making any inventive effort should fall within the scope of protection defined by the present invention.
Claims (5)
1. An unmanned vegetable greenhouse planting method based on computer vision is characterized by comprising the following steps: the image information acquisition and transmission device, the upper computer processing end and the early warning display device; the image information acquisition and transmission device comprises a picture positioning acquisition device and a wireless signal transmission device; the early warning display device comprises a disease prompt device and a disease display device; the picture positioning and collecting device comprises a camera, a rotating shaft motor, a position sensor and a gear track, wherein the rotating shaft motor drives the gear track to move so as to drive the camera on the track to move, and when the position sensor area where vegetables are located is reached, the picture positioning and collecting device stops; the wireless signal transmitting device transmits the acquired picture information to the upper computer processing end of the main control through the wireless signal transmission bus, the upper computer processing end analyzes and processes the acquired picture information through the established data model after receiving the image signal to obtain a final result, and when the upper computer processing end detects that vegetables are diseased and gives out disease types, the disease display device automatically displays the disease types, the disease prompt device is started and notifies a shed owner to take corresponding processing measures;
The potato disease prediction model is established by a convolutional neural network deep learning mode, and the method comprises the following steps:
(1) Collecting potato leaf pictures of five common diseases of potatoes and a healthy and normal potato leaf picture;
(2) Classifying and labeling the six states of the collected potato images, and establishing a training set and a testing set required by a model training sequence by a data set according to a 9:1 relation;
(3) Pretreating the potato image;
(4) Feature extraction and disease identification, wherein a convolutional neural network is used for carrying out feature extraction and identification on the processed picture, and the basic structure of the convolutional neural network comprises a convolutional layer, a pooling layer and a full-connection layer;
(5) The test set is brought into the trained model to judge the accuracy rate, namely, the test set is used for evaluating whether the trained model can be put into use;
the five common potato diseases in the step (1) are "potato plague", "ring spot", "potato yellowing leaf curl", "potato wilt", "potato anthracnose";
The image preprocessing method in the step (3) is carried out according to the following steps:
3a) Image cutting, namely removing redundant parts by cutting and reserving a research theme;
3b) Adjusting the resolution of the image, and adjusting the original image to 800 x 600 pixel resolution under the condition of ensuring that the proportion of the image is unchanged and the DPI is unchanged;
3c) Image gray scale processing, namely converting a color RGB color image into a gray scale image by adopting a pixel average method, namely averaging R, G and B pixel points at corresponding positions to obtain the pixel point at the position;
3d) Image filtering, namely performing nonlinear processing on an image by adopting a simple median filtering method, setting an original image as g, setting g (i, j) as a pixel point with coordinates of (i, j) in the image g, acquiring field spaces of K pixels by using a given template for the pixel point, arranging the pixels from small to large according to gray values, and taking a median value as a new gray value;
the convolutional neural network used in the step (4) performs feature extraction and identification on the processed picture, and comprises the following steps:
4a) The potatoes in six states are singly classified into one type, namely ' healthy potato leaves ' are in a type 1', potato plague leaves ' are in a type 2 ', ring rot is in a type 3', potato yellowing leaf curl is in a type 4', potato wilt is in a type 5, and potato anthracnose is in a type 6;
4b) When the category 1 is a positive sample, the rest categories are negative samples, namely the marking data is (1,0,0,0,0,0);
when the category 2 is a positive sample, the remaining categories are negative samples, i.e. the marking data is (0,1,0,0,0,0);
when the category 3 is a positive sample, the remaining categories are negative samples, i.e. the marking data is (0,0,1,0,0,0);
when the category 4 is a positive sample, the remaining categories are negative samples, i.e. the marking data is (0,0,0,1,0,0);
when the category 5 is a positive sample, the remaining categories are negative samples, i.e. the marking data is (0,0,0,0,1,0);
when the category 6 is a positive sample, the remaining categories are negative samples, i.e. the marking data is (0,0,0,0,0,1);
4c) Training the potato picture data set marked with the class by using the constructed 10-layer deep convolutional neural network structure to obtain the probability of the corresponding class, and continuously training the data by using the counter propagation principle until the final gradient is reduced to the minimum value, so that the probability of the predicted sample is close to 0 on the rest classes and is close to 1 on the class.
2. The unmanned vegetable greenhouse planting method based on computer vision according to claim 1, wherein the 10-layer deep convolutional neural network structure provided in the step 4 c) comprises 3 convolutional layers, 3 maximum pooling layers, 2 full-connection layers, 1 softmax layer and 1 output layer.
3. The unmanned vegetable greenhouse planting method based on computer vision according to claim 2, wherein the 3-layer convolution layer adopts 3 different convolution kernels, one layer extracts edge features, two layers extracts main contour features, and three layers extracts texture features.
4. The unmanned vegetable greenhouse planting method based on computer vision according to claim 2, wherein the pooling layers all adopt a maximum pooling algorithm capable of overlapping 2 x 2 windows.
5. The unmanned vegetable greenhouse planting method based on computer vision according to claim 1, wherein the evaluation standard content is:
precision ratio P: true marking of the correct proportion in all potato leaves marked with disease; recall ratio R: for all figures, the disease duty cycle is identified.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011368215.0A CN112419288B (en) | 2020-11-28 | 2020-11-28 | Unmanned vegetable greenhouse planting method based on computer vision |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011368215.0A CN112419288B (en) | 2020-11-28 | 2020-11-28 | Unmanned vegetable greenhouse planting method based on computer vision |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112419288A CN112419288A (en) | 2021-02-26 |
| CN112419288B true CN112419288B (en) | 2024-05-03 |
Family
ID=74828884
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011368215.0A Active CN112419288B (en) | 2020-11-28 | 2020-11-28 | Unmanned vegetable greenhouse planting method based on computer vision |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112419288B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115053735A (en) * | 2022-07-13 | 2022-09-16 | 黄山四月乡村农艺场有限公司 | Constant temperature greenhouse temperature control system |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101162384A (en) * | 2006-10-12 | 2008-04-16 | 魏珉 | Artificial intelligence plant growth surroundings regulate and control expert decision-making system |
| CN205334168U (en) * | 2016-04-21 | 2016-06-22 | 贾如春 | Based on big data plant diseases and insect pests monitoring and early warning system |
| EP3716207A1 (en) * | 2019-03-26 | 2020-09-30 | Can Ince | Method and apparatus for diagnostic analysis of the function and morphology of microcirculation alterations |
| WO2020199538A1 (en) * | 2019-04-04 | 2020-10-08 | 中设设计集团股份有限公司 | Bridge key component disease early-warning system and method based on image monitoring data |
-
2020
- 2020-11-28 CN CN202011368215.0A patent/CN112419288B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101162384A (en) * | 2006-10-12 | 2008-04-16 | 魏珉 | Artificial intelligence plant growth surroundings regulate and control expert decision-making system |
| CN205334168U (en) * | 2016-04-21 | 2016-06-22 | 贾如春 | Based on big data plant diseases and insect pests monitoring and early warning system |
| EP3716207A1 (en) * | 2019-03-26 | 2020-09-30 | Can Ince | Method and apparatus for diagnostic analysis of the function and morphology of microcirculation alterations |
| WO2020199538A1 (en) * | 2019-04-04 | 2020-10-08 | 中设设计集团股份有限公司 | Bridge key component disease early-warning system and method based on image monitoring data |
Non-Patent Citations (3)
| Title |
|---|
| 信息技术在农作物病虫害监测预警中的应用综述;李素;郭兆春;王聪;陈天恩;袁志高;;江苏农业科学(22);全文 * |
| 深度学习在图像识别中的应用研究综述;郑远攀;李广阳;李晔;;计算机工程与应用(12);全文 * |
| 蔬菜病害识别诊断与预警物联网技术研究与应用;张领先;李鑫星;;蔬菜(08);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112419288A (en) | 2021-02-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111563887B (en) | Intelligent analysis method and device for oral cavity image | |
| CN110717903A (en) | Method for detecting crop diseases by using computer vision technology | |
| CN105894536A (en) | Method and system for analyzing livestock behaviors on the basis of video tracking | |
| KR102002632B1 (en) | Fruit monitoring system and method at the same | |
| CN106845366B (en) | Sugarcane coverage automatic detection method based on image | |
| CN113947734A (en) | Pig physical sign information acquisition method and system | |
| CN112151167A (en) | Intelligent screening method for six-age dental caries of children based on deep learning | |
| CN110826552A (en) | Grape nondestructive automatic detection device and method based on deep learning | |
| CN112419288B (en) | Unmanned vegetable greenhouse planting method based on computer vision | |
| CN111161214A (en) | Binocular vision-based pig weight measurement and drinking behavior identification system and method | |
| CN114898405B (en) | Portable broiler chicken anomaly monitoring system based on edge calculation | |
| CN115812634A (en) | Pre-incubation hatching egg fertilization information nondestructive sorting method and device | |
| CN114842957A (en) | Senile dementia auxiliary diagnosis system and method based on emotion recognition | |
| CN115953690B (en) | Lodging crop identification method for unmanned harvester travel calibration | |
| Poonguzhali et al. | Crop condition assessment using machine learning | |
| CN114359539B (en) | Intelligent identification method for high-spectrum image of parasite in sashimi | |
| CN116686738A (en) | Automatic weighing and identifying system for dairy cows | |
| CN111144379B (en) | Automatic identification method for visual dynamic response of mice based on image technology | |
| CN116977648A (en) | Identification method and system for vegetable soybean phenotype information based on target detection | |
| Yenugudhati | Identification of plant health using machine learning and image techniques | |
| CN114022831A (en) | Binocular vision-based livestock body condition monitoring method and system | |
| Yang et al. | Research and design of a machine vision-based silk cocoon quality inspection system | |
| Valarmathi et al. | Fruit Disease Prediction with Fertilizer Recommendation for Citrus Family using Deep Learning | |
| CN117409403B (en) | Rice spike maturity estimation method based on deep learning | |
| Viduka et al. | The potential of RGB camera for machine learning in non-destructive detection of nutrient deficiencies in apples |
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 |