CN112801942B - Citrus yellow dragon disease image identification method based on attention mechanism - Google Patents
Citrus yellow dragon disease image identification method based on attention mechanism Download PDFInfo
- Publication number
- CN112801942B CN112801942B CN202011632531.4A CN202011632531A CN112801942B CN 112801942 B CN112801942 B CN 112801942B CN 202011632531 A CN202011632531 A CN 202011632531A CN 112801942 B CN112801942 B CN 112801942B
- Authority
- CN
- China
- Prior art keywords
- feature
- disease
- attention
- citrus yellow
- attention 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.)
- Active
Links
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
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- 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
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
-
- 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
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30128—Food products
Abstract
The invention relates to the field of image recognition, and particularly discloses a citrus yellow dragon disease image recognition method based on an attention mechanism, which comprises the following steps of: collecting disease leaf and disease fruit data of citrus yellow dragon disease as positive samples and other data not of citrus yellow dragon disease as negative samples, constructing an image classification network based on an attention mechanism, designing a loss function, inputting a training set into the image classification network based on the attention mechanism, performing supervision training by adopting the loss function, and inputting a verification set into a trained model for verification in the training process. And loading the trained model parameters to an image classification network based on an attention mechanism, and sequentially inputting the citrus yellow crohn's disease images of the test set to the network for reasoning to obtain a citrus yellow crohn's disease image classification result. The method for detecting the citrus yellow dragon disease has the advantages that key features of the citrus yellow dragon disease are selected independently through the attention mechanism module, and the method for detecting the citrus yellow dragon disease is low in cost, high in efficiency and high in identification rate based on image identification.
Description
Technical Field
The invention relates to the field of image recognition, in particular to a citrus yellow dragon disease image recognition method based on an attention mechanism.
Background
Citrus yellow dragon disease is the most destructive disease in the current global citrus production, and is caused by bacillus phloem of citrus yellow dragon disease, citrus seedlings carrying citrus yellow dragon disease pathogenic bacteria are taken as a transmission way, and citrus psyllids and the like are taken as transmission media. The citrus growing areas in nearly 50 countries and regions of the world have been infected, and once created the important citrus producing areas of florida, holy-Paul, brazil, guangdong, etc., in the United states, no method for radically treating citrus yellow dragon disease has been found yet. The agricultural departments all over the world attach great importance to the prevention and treatment of citrus yellow dragon diseases, 9 and 15 days in 2020, which are listed in a list of crop diseases and insect pests by agricultural rural areas.
Early diagnosis and early treatment, citrus yellow dragon disease detection is important for prevention and control work, and at present, there are mainly several detection methods: the PCR detection is carried out, whether the pathogen DNA sequence in the citrus body is consistent with the yellow dragon disease DNA sequence is detected by instrument and equipment, the identification by a professional institution is needed, and the detection efficiency is low; the hyperspectral remote sensing of the unmanned aerial vehicle is used for large-area detection, so that the cost is high; based on the characteristics of low cost and high efficiency of a computer vision scheme, the method is widely used for identifying diseases and insect pests in recent years, but the disease characteristics of citrus yellow dragon disease belong to fine granularity characteristics, and the identification rate of a general image identification algorithm is low.
Disclosure of Invention
Aiming at the defects existing in the background technology, the invention provides the citrus yellow dragon disease image recognition method based on the attention mechanism, thereby overcoming the defects that the disease characteristics of the citrus yellow dragon disease belong to fine granularity characteristics and the recognition rate of a general image recognition algorithm is lower.
In order to achieve the purpose, the invention provides a citrus yellow long disease identification method based on an attention mechanism, which comprises the following steps:
collecting citrus yellow dragon disease image data as a positive sample and collecting non-yellow dragon disease image data as a negative sample, and dividing the citrus yellow dragon disease image data into a training set, a verification set and a test set according to disease categories;
establishing an image classification network, wherein the image classification network comprises an input layer, a feature extraction module, an attention mechanism module, a feature fusion module and an output layer; the input layer is used for receiving the positive sample and the negative sample; the feature extraction module is used for automatically extracting feature images of citrus yellow dragon diseases through a neural learning network; the attention mechanism module is used for generating an attention graph from the feature graph; the feature fusion module is used for carrying out feature fusion on the feature map and the attention map to obtain a feature matrix; the output layer is used for converting the feature matrix into a classified prediction vector for output;
designing a center loss function and a cross entropy loss function, supervising the process of attempting to learn the citrus yellow dragon disease characteristics by adopting the center loss function, and supervising the classification process of classifying the prediction vectors by adopting the cross entropy loss function;
inputting the training set into the image classification network, performing supervision training by adopting the center loss function and the cross entropy loss function, and inputting the verification set into the trained image classification network for verification in the training process;
and loading the trained network parameters to the initial image classification network, and sequentially inputting the disease leaf and disease fruit image data of the test set to the image classification network for testing to obtain the citrus yellow dragon disease image classification result.
Preferably, in the above technical solution, the input layer is used for scaling the citrus yellow dragon disease image to 448x448x3 and normalizing to [ -1,1].
Preferably, in the above technical solution, the feature extraction module uses an acceptance v3 network structure to perform feature extraction on citrus yellow long disease image data of the input layer, and outputs a feature map F with a size of 26x26x768, where 26x26 represents a width and a height of the feature map F, and 768 is a channel number of the feature map F.
Preferably, in the above technical solution, the attention mechanism module is composed of a 1x1 convolution layer, a Batch Normalization layer and a ReLU layer, and inputs a feature map F with a size of 26x26x768, and outputs an attention map a with a size of 26x26x32, where 26x26 represents the width and height of the attention map a, 32 represents the number of attention map a channels, denoted as M, and each attention map a channel may correspond to a feature region of citrus yellow dragon disease.
Preferably, in the above technical solution, the inputs of the feature fusion module are a feature map F and an attention map a, and each channel a of the attention map a is respectively shown in formula (1) k Multiplying the matrix corresponding elements with the characteristic images F of all channels to obtain the characteristic images F of the attention seeking channels k Size 26x26x768, and for feature map F k Performing global average pooling operation to obtain a 1x1x 768-sized attention characteristic diagram f as shown in formula (2) k Finally, 32 attention characteristic graphs f k Splicing, as shown in a formula (3), to obtain a feature matrix P of 32x768, namely a final output result of the feature fusion module;
F k ＝A k ⊙F(k＝1，2，...，M) (1)
f k ＝g(F k ) (2)
preferably, in the above technical solution, the input of the output layer is a feature matrix P, the feature matrix P is normalized by L2 norm, and then is subjected to a flat operation, converted into 24576-dimensional feature vectors, and input into a full-connection layer with a channel number of 6, and finally a 6-dimensional classification prediction vector is obtained as output.
Preferably, in the above technical solution, the classifying process for supervising the classified prediction vector by using a cross entropy Loss function CE Loss specifically includes: firstly, inputting a 6-dimensional prediction vector into a CE Loss function to calculate cross entropy Loss, wherein the CE Loss function is shown as a formula (4), and x is as follows cls Predictive value, x, representing true class cls j A predicted value representing a j-th category;
using a centre loss functionThe process of central Loss supervision striving to learn the characteristics of citrus yellow dragon disease specifically includes: f is as shown in formula (5) k For the attention map of the kth 1x1x768 size, c denotes the feature center matrix, initialized to an all zero matrix of the 6x32x768 size,for the kth attention profile at the feature center of the true class cls, the feature center size is 1x1x768, for each attention channel k, f is calculated k And->The Euclidean distance of (2) is summed to obtain a Center Loss;
the characteristic center is calculated by the formula (6)Updating is carried out, and the updating rate is controlled to be 0.05.
CE Loss＝-x cls +log(∑ j exp(x j ))(j＝1，2，...，N) (4)
Compared with the prior art, the invention provides the citrus yellow dragon disease image identification method based on the attention mechanism, compared with the existing citrus yellow dragon disease image identification method, the method has the advantages that key characteristics of the citrus yellow dragon disease are selected independently through introducing the attention mechanism module according to the mottled yellowing of the citrus yellow dragon disease leaves, coloring near the pedicles and uncolored visual characteristics of the top parts of the fruits, the citrus yellow dragon disease detection scheme with low cost, high efficiency and high identification rate is realized based on image identification, and the classification effect of the citrus yellow dragon disease model is improved.
Drawings
Fig. 1 is a flowchart of image classification according to an embodiment of the present invention.
Fig. 2 is a network configuration diagram of an embodiment of the present invention.
Detailed Description
The following detailed description of embodiments of the invention is, therefore, to be taken in conjunction with the accompanying drawings, and it is to be understood that the scope of the invention is not limited to the specific embodiments.
The main design thought of the citrus yellow long disease image identification method based on the attention mechanism in the embodiment is as follows:
(1) And (3) data set preparation: collecting the data of the disease leaves and fruits of the citrus yellow dragon disease as positive samples, and the data of other non-yellow dragon diseases as negative samples, classifying according to disease types, and randomly dividing the data into a training set, a verification set and a test set according to the ratio of 0.8:0.1:0.1.
(2) An image classification network based on an attention mechanism is constructed: the device consists of an input layer, a feature extraction module, an attention mechanism module, a feature fusion module and an output layer.
(3) Designing a loss function: and a cross entropy Loss CE Loss and Center Loss joint Loss function is adopted, and the CE Loss is responsible for supervising the classification process.
(4) Model training: inputting the training set into the image classification network based on the attention mechanism in the step (2), performing supervision training by adopting the loss function in the step (3), and inputting the verification set into a trained model for verification in the training process.
(5) Model reasoning: loading the trained model parameters in the step (4) to the image classification network based on the attention mechanism in the step (2), and sequentially inputting the citrus yellow dragon disease images of the test set in the step (1) to the network for reasoning to obtain the citrus yellow dragon disease image classification result.
More specifically, referring to fig. 1, the method for identifying citrus yellow long disease image based on the attention mechanism comprises the following steps:
step S1, collecting disease leaves and disease fruits data of citrus yellow dragon diseases as positive samples, and other data of non-yellow dragon diseases as negative samples, classifying according to disease types, and randomly dividing the disease types into a training set, a verification set and a test set according to the proportion of 0.8:0.1:0.1;
step S2, constructing an image classification network based on an attention mechanism, as shown in FIG. 2: the device consists of an input layer, a feature extraction module, an attention mechanism module, a feature fusion module and an output layer:
input layer: and (3) scaling the citrus yellow dragon disease image in the step S1 to 448x448x3, normalizing to [ -1,1], and accelerating the convergence speed of model training.
The feature extraction module is used for automatically extracting visual feature images of citrus yellow dragon diseases through a neural network. More specifically, the feature extraction module adopts an acceptance v3 network structure to perform feature extraction on image data of an input layer, and outputs a feature map with a size of 26x26x768, denoted as F, wherein 26x26 represents the width and height of the feature map F, and 768 is the channel number of the feature map F.
The attention mechanism module is used for generating attention force diagrams from the feature graphs F output by the feature extraction module, and the attention force diagrams are denoted as A. The attention mechanism module consists of a 1x1 convolution layer, a Batch Normalization layer and a ReLU layer, inputs a feature map F of 26x26x768 size, outputs an attention map a of 26x26x32, wherein 26x26 represents the width and height of the attention map a, 32 represents the number of channels of the attention map a, denoted as M, and manually set parameters. Each attention map a channel may correspond to a characteristic region of citrus yellow dragon disease, for example, a colored region near the citrus pedicel, and key features of citrus yellow dragon disease may be selected autonomously by model training.
The feature fusion module is used for carrying out feature fusion on the feature graph F output by the feature extraction module and the attention graph A output by the attention mechanism module. The input of the feature fusion module is a feature graph F output by the feature extraction module and an attention graph A output by the attention mechanism module, and each channel A of the attention graph A is respectively shown in the formula (1) k Multiplying the matrix corresponding elements with the feature images F of all channels to obtain the feature images F of the channels of the attention map A k Size 26x26x768, and for F k Performing global average pooling operation to obtain attention characteristic diagram with 1x1x768 size as shown in formula (2)f k Finally, 32 f k And (3) splicing, wherein a feature matrix P of 32x768 is obtained as shown in a formula (3), and the final output result of the feature fusion module is obtained.
F k ＝A k ⊙F(k＝1，2，...，M) (1)
f k ＝g(F k ) (2)
The function of the output layer module is to map the feature matrix P of the feature fusion module to the classification prediction vector. Specifically, the input of the output layer is the feature matrix P output by the feature fusion module, L2 norm normalization is carried out on the P, then the flat operation is carried out, the P is converted into 24576-dimensional feature vectors, the 24576-dimensional feature vectors are input into a full-connection layer with the channel number of 6, and finally 6-dimensional classification prediction vectors are obtained as output.
Step S3, designing a loss function: and adopting a joint Loss function of a cross entropy Loss function CE Loss and a center Loss function CenterLoss, wherein the CE Loss function is responsible for supervising the classification process. Firstly, inputting a 6-dimensional predictive vector of an output layer in the step S2 into a CE Loss function to calculate cross entropy Loss, wherein the CE Loss is shown as a formula (4), and x is shown as the formula (4) cls Predictive value, x, representing true class cls j Representing the predicted value of the j-th class.
The centrloss function is responsible for supervising the learning process of the attention mechanism, as shown in equation (5), f k For the kth 1x1x768 size attention profile in step S2, c represents the feature center matrix, initialized to a 6x32x768 size all zero matrix,for the kth attention profile at the feature center of the true class cls, the feature center size is 1x1x768, for each attention channel k, f is calculated k And->Is of the European typeAnd obtaining a Center Loss function by summing the distances. The feature center can be +.>Updating is carried out, and the updating rate is controlled to be 0.05.
CE Loss＝-x cls +log(∑ j exp(x j ))(j＝1，2，...，N) (4)
And S4, inputting the training set into the image classification network based on the attention mechanism in the step S2, performing supervision training by adopting the loss function in the step S3, and inputting the verification set into a trained model for verification in the training process.
And S5, loading the model parameters trained in the step S4 to an image classification network based on an attention mechanism in the step S2, and sequentially inputting the citrus yellow dragon disease images of the test set in the step S1 to the network for reasoning to obtain a citrus yellow dragon disease image classification result.
In summary, compared with the existing citrus yellow dragon disease image identification method, the citrus yellow dragon disease image identification method based on the attention mechanism is characterized in that key features of the citrus yellow dragon disease are selected independently through the attention mechanism module according to the mottled yellowing of the citrus yellow dragon disease leaves, coloring near the pedicel and uncolored visual features of the top part of the citrus yellow dragon disease, the citrus yellow dragon disease detection scheme with low cost, high efficiency and high identification rate is realized based on image identification, and the classification effect of the citrus yellow dragon disease model is improved.
The foregoing descriptions of specific exemplary embodiments of the present invention are presented for purposes of illustration and description. It is not intended to limit the invention to the precise form disclosed, and obviously many modifications and variations are possible in light of the above teaching. The exemplary embodiments were chosen and described in order to explain the specific principles of the invention and its practical application to thereby enable one skilled in the art to make and utilize the invention in various exemplary embodiments and with various modifications as are suited to the particular use contemplated. It is intended that the scope of the invention be defined by the claims and their equivalents.
Claims (7)
1. The citrus yellow dragon disease identification method based on the attention mechanism is characterized by comprising the following steps of:
collecting citrus yellow dragon disease image data as a positive sample and collecting non-yellow dragon disease image data as a negative sample, and dividing the citrus yellow dragon disease image data into a training set, a verification set and a test set according to disease categories;
establishing an image classification network, wherein the image classification network comprises an input layer, a feature extraction module, an attention mechanism module, a feature fusion module and an output layer; the input layer is used for receiving the positive sample and the negative sample; the feature extraction module is used for automatically extracting feature images of citrus yellow dragon diseases through a neural learning network; the attention mechanism module is used for generating an attention graph from the feature graph; the feature fusion module is used for carrying out feature fusion on the feature map and the attention map to obtain a feature matrix; the output layer is used for converting the feature matrix into a classified prediction vector for output;
designing a center loss function and a cross entropy loss function, supervising the process of attempting to learn the citrus yellow dragon disease characteristics by adopting the center loss function, and supervising the classification process of classifying the prediction vectors by adopting the cross entropy loss function;
inputting the training set into the image classification network, performing supervision training by adopting the center loss function and the cross entropy loss function, and inputting the verification set into the trained image classification network for verification in the training process;
and loading the trained network parameters to the initial image classification network, and sequentially inputting the disease leaf and disease fruit image data of the test set to the image classification network for testing to obtain the citrus yellow dragon disease image classification result.
2. An attention-based citrus yellow crohn's disease identification method according to claim 1, wherein the input layer is for scaling the citrus yellow crohn's disease image to 448x448x3 and normalizing to [ -1,1].
3. The method for identifying citrus yellow long disease based on the attention mechanism according to claim 2, wherein the feature extraction module adopts an acceptance v3 network structure to perform feature extraction on the citrus yellow long disease image data of the input layer, and outputs a feature map F with a size of 26x26x768, wherein 26x26 represents the width and height of the feature map F, and 768 is the channel number of the feature map F.
4. A method for identifying citrus yellow croaker disease based on an attention mechanism as in claim 3, wherein the attention mechanism module consists of a 1x1 convolution layer, a Batch Normalization layer and a ReLU layer, inputs a feature map F of 26x26x768 size, outputs an attention map a of 26x26x32, wherein 26x26 represents the width and height of the attention map a, 32 represents the number of attention map a channels, denoted as M, each of which may correspond to a feature region of citrus yellow croaker disease.
5. The method for identifying citrus yellow long based on attention mechanism of claim 4, wherein the inputs of the feature fusion module are a feature map F and an attention map a, each channel a of the attention map a being respectively as shown in formula (1) k Multiplying the matrix corresponding elements with the characteristic images F of all channels to obtain the characteristic images F of the attention seeking channels k Size 26x26x768, and for feature map F k Performing global average pooling operation to obtain a 1x1x 768-sized attention characteristic diagram f as shown in formula (2) k Finally, 32 attention characteristic graphs f k And (3) splicing, wherein a feature matrix P of 32x768 is obtained as shown in a formula (3), and the final output result of the feature fusion module is obtained.
F k ＝A k ⊙F(k＝1，2，...，M) (1)
f k ＝g(F k ) (2)
6. The method for identifying citrus yellow long disease based on attention mechanism according to claim 5, wherein the input of the output layer is a feature matrix P, the feature matrix P is normalized by L2 norm, then is subjected to a flat operation, is converted into 24576-dimensional feature vectors, and is input into a fully-connected layer with a channel number of category number 6, and finally a 6-dimensional classification prediction vector is obtained as output.
7. The method for identifying citrus yellow long based on an attention mechanism of claim 6, wherein the classifying process for supervising the classified prediction vector by using a cross entropy Loss function CE Loss specifically comprises: firstly, inputting a 6-dimensional prediction vector into a CE Loss function to calculate cross entropy Loss, wherein the CE Loss function is shown as a formula (4), and x is as follows cls Predictive value, x, representing true class cls j A predicted value representing a j-th category;
the process of taking central Loss function Center Loss supervision attention to try to learn citrus yellow dragon disease characteristics specifically comprises the following steps: f is as shown in formula (5) k For the attention map of the kth 1x1x768 size, c denotes the feature center matrix, initialized to an all zero matrix of the 6x32x768 size,for the kth attention profile at the feature center of the true class cls, the feature center size is 1x1x768, for each attention channel k, f is calculated k And->Is summed to obtain the centerr Loss；
CE Loss＝-x cls +log(∑ j exp(x j ))(j＝1，2，...，N) (4)
The characteristic center is calculated by the formula (6)Updating is carried out, and the updating rate is controlled to be 0.05.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011632531.4A CN112801942B (en) | 2020-12-31 | 2020-12-31 | Citrus yellow dragon disease image identification method based on attention mechanism |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011632531.4A CN112801942B (en) | 2020-12-31 | 2020-12-31 | Citrus yellow dragon disease image identification method based on attention mechanism |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112801942A CN112801942A (en) | 2021-05-14 |
| CN112801942B true CN112801942B (en) | 2023-10-13 |
Family
ID=75808454
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011632531.4A Active CN112801942B (en) | 2020-12-31 | 2020-12-31 | Citrus yellow dragon disease image identification method based on attention mechanism |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112801942B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113627258B (en) * | 2021-07-12 | 2023-09-26 | 河南理工大学 | Apple leaf pathology detection method |
| CN114677606B (en) * | 2022-04-28 | 2023-04-18 | 电子科技大学 | Citrus fine-grained disease identification method based on attention mechanism and double-branch network |
| CN115731189B (en) * | 2022-11-22 | 2023-06-23 | 河北师范大学 | Enteroscopy image lesion recognition method and device, electronic equipment and storage medium |
| CN115690920B (en) * | 2023-01-03 | 2023-04-14 | 智慧眼科技股份有限公司 | Credible living body detection method for medical identity authentication and related equipment |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110110751A (en) * | 2019-03-31 | 2019-08-09 | 华南理工大学 | A kind of Chinese herbal medicine recognition methods of the pyramid network based on attention mechanism |
| CN111382676A (en) * | 2020-02-25 | 2020-07-07 | 南京大学 | Sand image classification method based on attention mechanism |
| CN111489334A (en) * | 2020-04-02 | 2020-08-04 | 暖屋信息科技(苏州)有限公司 | Defect workpiece image identification method based on convolution attention neural network |
| CN111696101A (en) * | 2020-06-18 | 2020-09-22 | 中国农业大学 | Light-weight solanaceae disease identification method based on SE-Inception |
| AU2020102885A4 (en) * | 2020-10-20 | 2020-12-17 | Xijing University | Disease recognition method of winter jujube based on deep convolutional neural network and disease image |
| CN112101424A (en) * | 2020-08-24 | 2020-12-18 | 深圳大学 | Generation method, identification device and equipment of retinopathy identification model |
-
2020
- 2020-12-31 CN CN202011632531.4A patent/CN112801942B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110110751A (en) * | 2019-03-31 | 2019-08-09 | 华南理工大学 | A kind of Chinese herbal medicine recognition methods of the pyramid network based on attention mechanism |
| CN111382676A (en) * | 2020-02-25 | 2020-07-07 | 南京大学 | Sand image classification method based on attention mechanism |
| CN111489334A (en) * | 2020-04-02 | 2020-08-04 | 暖屋信息科技(苏州)有限公司 | Defect workpiece image identification method based on convolution attention neural network |
| CN111696101A (en) * | 2020-06-18 | 2020-09-22 | 中国农业大学 | Light-weight solanaceae disease identification method based on SE-Inception |
| CN112101424A (en) * | 2020-08-24 | 2020-12-18 | 深圳大学 | Generation method, identification device and equipment of retinopathy identification model |
| AU2020102885A4 (en) * | 2020-10-20 | 2020-12-17 | Xijing University | Disease recognition method of winter jujube based on deep convolutional neural network and disease image |
Non-Patent Citations (1)
| Title |
|---|
| 苏鸿 等.基于区域卷积神经网络模型的广西柑橘病虫害识别方法研究.《西南农业学报》.2020,第33卷(第04期),805-810. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112801942A (en) | 2021-05-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112801942B (en) | Citrus yellow dragon disease image identification method based on attention mechanism | |
| CN109829430B (en) | Cross-modal pedestrian re-identification method and system based on heterogeneous hierarchical attention mechanism | |
| Shaheen et al. | Impact of automatic feature extraction in deep learning architecture | |
| CN106682694A (en) | Sensitive image identification method based on depth learning | |
| CN106326874A (en) | Method and device for recognizing iris in human eye images | |
| CN107871136A (en) | The image-recognizing method of convolutional neural networks based on openness random pool | |
| CN107563389A (en) | A kind of corps diseases recognition methods based on deep learning | |
| CN106997475A (en) | A kind of insect image-recognizing method based on parallel-convolution neutral net | |
| CN108171318A (en) | One kind is based on the convolutional neural networks integrated approach of simulated annealing-Gaussian function | |
| CN104091181A (en) | Injurious insect image automatic recognition method and system based on deep restricted Boltzmann machine | |
| CN114676769A (en) | Visual transform-based small sample insect image identification method | |
| Ma et al. | Algorithm of strawberry disease recognition based on deep convolutional neural network | |
| CN116994295B (en) | Wild animal category identification method based on gray sample self-adaptive selection gate | |
| Alharbi et al. | Wheat disease classification using continual learning | |
| CN113516097B (en) | Plant leaf disease identification method based on improved EfficentNet-V2 | |
| Nigam et al. | Wheat rust disease identification using deep learning | |
| CN109886097A (en) | A kind of method for detecting fatigue driving based on artificial fish school optimization H-ELM | |
| CN115631462A (en) | AM-YOLOX-based strawberry disease and pest detection method and system | |
| CN108846327A (en) | A kind of intelligent distinguishing system and method for mole and melanoma | |
| Costales et al. | Development of a Prototype Application for Rice Disease Detection Using Convolutional Neural Networks | |
| CN114580552A (en) | Small sample tea image classification method based on migration model | |
| Poorni et al. | Detection of rice leaf diseases using convolutional neural network | |
| Gaikwad | Identification of fungi infected leaf diseases using deep learning techniques | |
| Anwar et al. | Olive Leaves Disease Detection and Classification using Deep Learning | |
| CN110458219A (en) | A kind of Φ-OTDR vibration signal recognizer based on STFT-CNN-RVFL |
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 |