CN116342919A - Rice disease identification method based on attention and gating mechanism - Google Patents
Rice disease identification method based on attention and gating mechanism Download PDFInfo
- Publication number
- CN116342919A CN116342919A CN202211197823.9A CN202211197823A CN116342919A CN 116342919 A CN116342919 A CN 116342919A CN 202211197823 A CN202211197823 A CN 202211197823A CN 116342919 A CN116342919 A CN 116342919A
- Authority
- CN
- China
- Prior art keywords
- rice
- disease
- attention
- identification
- gating mechanism
- 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.)
- Pending
Links
- 235000007164 Oryza sativa Nutrition 0.000 title claims abstract description 34
- 201000010099 disease Diseases 0.000 title claims abstract description 34
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 title claims abstract description 34
- 235000009566 rice Nutrition 0.000 title claims abstract description 34
- 230000007246 mechanism Effects 0.000 title claims abstract description 25
- 238000000034 method Methods 0.000 title claims abstract description 19
- 240000007594 Oryza sativa Species 0.000 title claims description 3
- 241000209094 Oryza Species 0.000 claims abstract description 33
- 238000012549 training Methods 0.000 claims abstract description 11
- 210000002569 neuron Anatomy 0.000 claims description 13
- 230000001580 bacterial effect Effects 0.000 claims description 4
- 238000012360 testing method Methods 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 3
- 238000012795 verification Methods 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 2
- 241000607479 Yersinia pestis Species 0.000 abstract description 11
- 238000005516 engineering process Methods 0.000 abstract description 5
- 238000013136 deep learning model Methods 0.000 abstract description 2
- 238000013527 convolutional neural network Methods 0.000 abstract 2
- 238000013135 deep learning Methods 0.000 abstract 1
- 230000006870 function Effects 0.000 description 9
- 230000004913 activation Effects 0.000 description 5
- 241000196324 Embryophyta Species 0.000 description 4
- 241000238631 Hexapoda Species 0.000 description 4
- 235000013339 cereals Nutrition 0.000 description 3
- 239000004973 liquid crystal related substance Substances 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 230000002860 competitive effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000005284 excitation Effects 0.000 description 1
- 235000013305 food Nutrition 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
Images
Classifications
-
- 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
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- 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
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
- G06V20/10—Terrestrial scenes
- G06V20/188—Vegetation
-
- 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
Abstract
The invention discloses a rice disease identification method based on attention and a gating mechanism, and relates to the technical field of disease and pest identification. Firstly, acquiring a rice disease image data set, and expanding a training sample data set by using a picture enhancement technology; and secondly, constructing a rice disease identification basic model based on a deep learning technology. The model uses a convolutional neural network technology and refers to the idea of a residual network to form a convolutional neural network identification model based on residual connection; third, a simple method is used to implement the three-dimensional attention module, and the channel and the context global information are modeled in combination with a gating mechanism. Finally, intelligent identification of rice diseases is realized through a classifier algorithm. The method is based on a three-dimensional attention mechanism combined with a gating channel unit and based on a residual error connection deep learning model to identify different types of rice diseases.
Description
Technical Field
The invention relates to the technical field of pest image recognition, in particular to a rice disease recognition method based on attention and a gating mechanism.
Background
Rice is one of main grain crops in China, the rice planting area in China is the second place in the world, and the planting area accounts for about 25% of the cultivated area in China. The population taking rice as main food in the whole country accounts for about 60% of the total population in the whole country, so that the rice yield is guaranteed to be related to the national folk life guarantee in the country.
According to the forecast of the national agricultural technology popularization service center, the occurrence area of the serious plant diseases and insect pests of 2021-year grains in China is 1.4 hundred million hm2 times in the country, and the serious plant diseases and insect pests can be in a resending situation. Among them, the estimated occurrence area of rice pest and disease damage of 2021 is 8154 ten thousand hm2 times, which will produce huge loss for the grain yield and economic development in our country. Therefore, rapid identification of rice pests and provision of control measures would be advantageous in reducing the impact of the pests on rice yield.
Identification of traditional crop diseases and insect pests mainly depends on a manual identification mode. Typically relying primarily on manual identification by agricultural professionals in the local plant protection sector. However, the manual identification method mainly has the problems of low identification efficiency, large workload and the like, and secondly, the method has no popularization by manually identifying the plant diseases and insect pests. Moreover, vast basic-level farmers do not have expert knowledge in pest and disease identification.
Therefore, how to realize intelligent recognition by means of computer technology can greatly improve the working efficiency and the application range.
Disclosure of Invention
The invention aims to provide a rice disease identification method based on an attention and gating mechanism, which overcomes the defects in the prior art.
In order to achieve the above purpose, the technical scheme adopted by the invention is as follows:
the rice disease identification method based on the attention and the gating mechanism comprises the following steps:
s1, collecting a rice disease data set;
s2, dividing the data set into a training set, a verification set and a test set;
s3, preprocessing the training sample, including data enhancement and normalization;
s4, referring to the neuroscience theory, the importance of each neuron is planned by setting an energy loss function, so that the problem that the traditional attention module can only realize attention excitation in a channel or space dimension is solved. The method can realize the attention activation operation in the three-dimensional space of the channel and the space at the same time;
s5, modeling global information by using two regularization items, and capturing competition and cooperation relations between channels by setting a regularization module and a gating mechanism; in this way, the cooperative relationship between the low-level features can be acquired, while the competitive relationship between the high-level features can be acquired. Therefore, lower level features may result in more general feature attributes, while higher level features may result in higher level semantic features related to tasks.
S6, identifying the types of the rice diseases through a classifier algorithm.
Further, the step S1 specifically includes: disease data sets are collected through a field shooting mode, and the following various common rice diseases affecting the rice yield are collected: the rice leaf spot, bacterial leaf blight, rice sheath blight, bacterial leaf streak, false smut and rice blast are important to study, and each disease sample is not less than 500.
Compared with the prior art, the invention has the advantages that: the method is based on a three-dimensional attention mechanism combined with a gating channel unit and based on a residual error connection deep learning model to identify different types of rice diseases.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that 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 the rice disease identification method based on the attention and the gating mechanism.
Detailed Description
The preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings so that the advantages and features of the present invention can be more easily understood by those skilled in the art, thereby making clear and defining the scope of the present invention.
Referring to FIG. 1, the invention provides a rice disease identification method based on an attention and gating mechanism, comprising the following steps:
step one:
and acquiring a crop disease image database in the farmland field. The crop disease and pest data set used by the invention comes from field scene shooting, and the total number of samples is 9982, and the total number of the samples is 6.
Step two:
the data set is divided into a training set, a validation set and a test set. The segmentation ratio is 8:1:1, the number of training samples after segmentation is 7986, the number of verification samples is 998, and the number of test samples is 998.
Step three:
the training set is data enhanced to expand the number of training samples. The data enhancement mode firstly cuts samples with different sizes into the same size, the initial cut size is 1.25 times of the sample size in the training process, and then the initial cut size is adjusted to the input picture size in the training process from the cut pictures in a central cutting mode, and the initial cut size is generally the sample size of (224 ). Then random rotation (-15 deg., 15 deg.) is carried out, and finally normalization treatment is carried out. The normalized mean and variance are ([ 0.485, 0.456, 0.406 ]) and ([ 0.229, 0.224, 0.225 ]) for each of the three R, G, B channels. The mean and variance described above are derived from computing their mean and variance for all samples in the ImageNet dataset.
Step four: three-dimensional attention mechanism implementation based on energy minimization
In order to implement the three-dimensional attention mechanism, the importance of each neuron needs to be calculated, i.e., the importance of each pixel needs to be calculated. According to neuroscience theory, an informative neuron will typically suppress surrounding neurons, i.e. the activated neurons should be given a higher weight, while the neurons surrounding it are given a lower weight. To achieve this function, it is necessary to distinguish between activated neurons and non-activated neurons, i.e. consider them as a classification task. Thus, the following loss function is set:
for simplicity, it is set to a binary label, i.e., the positive sample is set to +1 and the negative sample is set to-1. Meanwhile, in order to improve generalization capability, a regularization term is added, and the final energy loss function is as follows:
the energy function has a simple closed-form solution:
wherein:
the minimum energy finally obtained is as follows:
the above formula means: the lower the energy, the greater the distinction of the neuron t from the surroundings, indicating a higher importance. Therefore, the importance of each neuron t isAnd then, outputting the activation weight value of each neuron by the importance input activation function of each neuron, and finally multiplying the activation weight value by the input picture to realize the attention mechanism.
(2) Channel regularization
To establish the relationship between channels, two regularization methods are also used for construction.
(3) Gating mechanism
In order to obtain competing and cooperative relationships between channel features, a set of trainable parameters needs to be set up and the relationships between features learned.
Wherein, the liquid crystal display device comprises a liquid crystal display device,weights representing gating mechanisms, +.>Representing the bias of the gating mechanism, implementing the gating mechanism between channels. Then input it into an activation function, here chosen +.>The function implements a gating mechanism. Meanwhile, the information of the original picture is considered, and the activated weight and the original picture element are added and then multiplied by the picture. Finally, the following gating mechanism function is constructed:
implementation of gating mechanisms to implement channels and global contexts.
Although the embodiments of the present invention have been described with reference to the accompanying drawings, the patentees may make various modifications or alterations within the scope of the appended claims, and are intended to be within the scope of the invention as described in the claims.
Claims (2)
1. The rice disease identification method based on the attention and the gating mechanism is characterized by comprising the following steps:
s1, collecting a rice disease data set;
s2, dividing the data set into a training set, a verification set and a test set;
s3, preprocessing the training sample, including data enhancement and normalization;
s4, referring to the neuroscience theory, the importance of each neuron is planned by setting an energy loss function;
s5, modeling global information by using two regularization items, and capturing competition and cooperation relations between channels by setting a regularization module and a gating mechanism;
s6, identifying the types of the rice diseases through a classifier algorithm.
2. The method for identifying rice diseases based on the attention and gate control mechanism according to claim 1, wherein the step S1 specifically comprises: disease data sets are collected through a field shooting mode, and the following various common rice diseases affecting the rice yield are collected: the rice leaf spot, bacterial leaf blight, rice sheath blight, bacterial leaf streak, false smut and rice blast are important to study, and each disease sample is not less than 500.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211197823.9A CN116342919A (en) | 2022-09-29 | 2022-09-29 | Rice disease identification method based on attention and gating mechanism |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211197823.9A CN116342919A (en) | 2022-09-29 | 2022-09-29 | Rice disease identification method based on attention and gating mechanism |
Publications (1)
Publication Number | Publication Date |
---|---|
CN116342919A true CN116342919A (en) | 2023-06-27 |
Family
ID=86879568
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202211197823.9A Pending CN116342919A (en) | 2022-09-29 | 2022-09-29 | Rice disease identification method based on attention and gating mechanism |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116342919A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116863340A (en) * | 2023-08-16 | 2023-10-10 | 安徽荃银超大种业有限公司 | Rice leaf disease identification method based on deep learning |
-
2022
- 2022-09-29 CN CN202211197823.9A patent/CN116342919A/en active Pending
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116863340A (en) * | 2023-08-16 | 2023-10-10 | 安徽荃银超大种业有限公司 | Rice leaf disease identification method based on deep learning |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Jasim et al. | Plant leaf diseases detection and classification using image processing and deep learning techniques | |
Militante et al. | Plant leaf detection and disease recognition using deep learning | |
Ai et al. | Research on recognition model of crop diseases and insect pests based on deep learning in harsh environments | |
CN111563431A (en) | Plant leaf disease and insect pest identification method based on improved convolutional neural network | |
Alimboyong et al. | Classification of plant seedling images using deep learning | |
CN108549685A (en) | Behavior analysis method, device, system and readable storage medium storing program for executing | |
Der Yang et al. | Real-time crop classification using edge computing and deep learning | |
Zeng et al. | Identification of maize leaf diseases by using the SKPSNet-50 convolutional neural network model | |
CN114596468A (en) | Method and device for identifying plant diseases and insect pests and training model, electronic equipment and storage medium | |
CN116342919A (en) | Rice disease identification method based on attention and gating mechanism | |
US20220076068A1 (en) | Method for constructing a convolution neural network based on farmland images, electronic device using the same | |
Li et al. | Maize leaf disease identification based on WG-MARNet | |
Ma et al. | Algorithm of strawberry disease recognition based on deep convolutional neural network | |
Mahbub et al. | Detect bangladeshi mango leaf diseases using lightweight convolutional neural network | |
Singh et al. | Performance evaluation of plant leaf disease detection using deep learning models | |
Singh et al. | Performance Analysis of CNN Models with Data Augmentation in Rice Diseases | |
Bonik et al. | A Convolutional Neural Network Based Potato Leaf Diseases Detection Using Sequential Model | |
Seelwal et al. | Automatic Detection of Rice Diseases using Deep Convolutional Neural Networks with SGD and ADAM | |
CN114049935A (en) | HER2 image classification system based on multi-convolution neural network | |
CN116188855A (en) | Multi-scale plant disease identification method, device, storage medium and apparatus | |
Deshpande et al. | Detection of Plant Leaf Disease by Generative Adversarial and Deep Convolutional Neural Network | |
Agarwal et al. | Plant leaf disease classification using deep learning: A survey | |
CN113344009A (en) | Light and small network self-adaptive tomato disease feature extraction method | |
Zhang et al. | Automatic pest identification system in the greenhouse based on deep learning and machine vision | |
Ren et al. | Recognition of common pests in agriculture and forestry based on convolutional neural networks |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
TA01 | Transfer of patent application right |
Effective date of registration: 20231208 Address after: 230000 No.40, Nongke South Road, Hefei City, Anhui Province Applicant after: AGRICULTURAL ECONOMY AND INFORMATION RESEARCH OF ANHUI ACADEMY OF AGRICULTURAL SCIENCES Address before: 230000 40 Nongke South Road, Luyang District, Hefei City, Anhui Province Applicant before: Anhui Prov. Academy of Agricultural Sciences |
|
TA01 | Transfer of patent application right | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |