Available via license: CC BY 4.0
Content may be subject to copyright.
July, 2020 Int J Agric & Biol Eng Open Access at https://www.ijabe.org Vol. 13 No.4 205
Novel method for identifying wheat leaf disease images based on
differential amplification convolutional neural network
Mengping Dong1, Shaomin Mu1*, Aiju Shi2, Wenqian Mu1, Wenjie Sun1
(1. College of Information Science and Engineering, Shandong Agricultural University, Taian, Shandong 271018, China;
2. College of Chemistry and Materials Science, Shandong Agricultural University, Taian, Shandong 271018, China)
Abstract: In this study, a differential amplification convolutional neural network (DACNN) was proposed and used in the
identification of wheat leaf disease images with ideal accuracy. The branches added between the deep convolutional layers
can amplify small differences between the real output and the expected output, which made the weight updating more sensitive
to the light errors return in the backpropagation pass and significantly improved the fitting capability. Firstly, since there is no
large-scale wheat leaf disease images dataset at present, the wheat leaf disease dataset was constructed which included eight
kinds of wheat leaf images, and five kinds of data augmentation methods were used to expand the dataset. Secondly, DACNN
combined four classifiers: Softmax, support vector machine (SVM), K-nearest neighbor (KNN) and Random Forest to evaluate
the wheat leaf disease dataset. Finally, the DACNN was compared with the models: LeNet-5, AlexNet, ZFNet and Inception
V3. The extensive results demonstrate that DACNN is better than other models. The average recognition accuracy obtained
on the wheat leaf disease dataset is 95.18%.
Keywords: convolutional neural network, differential amplification, wheat leaf diseases, image identification
DOI: 10.25165/j.ijabe.20201304.4826
Citation: Dong M P, Mu S M, Shi A J, Mu W Q, Sun W J. Novel method for identifying wheat leaf disease images based on
differential amplification convolutional neural network. Int J Agric & Biol Eng, 2020; 13(4): 205–210.
1 Introduction
Wheat is one of the most important rations in China. The
development of the wheat industry is related to the country's food
security and social stability directly. Therefore, it is important for
yield and quantity to recognize wheat leaf diseases. However, at
present, the main method of wheat leaf disease identification is
manual identification, which has low efficiency and accuracy.
In recent years, deep learning has developed in image
recognition. In 1998, LeNet-5 was used for postal code
handwriting recognition, which has a 7-layer network structure[1].
In 2012, Convolutional Neural Network (CNN) was used to
achieve the best result in the ImageNet large-scale visual
recognition challenge, which caused to receive widespread
attention[2]. In 2014, Zeiler et al.[3] implemented ZFNet to
visualize network structure through deconvolution technology.
Simonyan et al.[4] proposed the visual geometry group (VGG)
model that increased the depth of the network by adding a
convolution layer of 3×3 convolution kernels, and used a small
convolution kernel to replace a convolution layer with a larger
convolution kernel, reducing the number of parameters. In 2015,
Received date: 2018-12-04 Accepted date: 2020-05-27
Biographies: Mengping Dong, Graduate student, research interests: artificial
intelligence, Email: dongmengping@126.com; Aiju Shi, Graduate student,
researcher, research interests: pest control, Email: saj31402@163.com; Wenqian
Mu, Undergraduate student, research interests: machine learning, Email:
1663385109@qq.com; Wenjie Sun, Graduate student, research interests:
artificial intelligence, Email: 766469613@qq.com.
*Corresponding author: Shaomin Mu, PhD, Professor, research interests:
machine learning, artificial intelligence, big data. College of Information
Science and Engineering, Shandong Agricultural University, No.61 Daizong
Street, Taian, Shandong 271018, China. Tel: +86-15005486826, Email:
msm@sdau.edu.cn.
Szegedy et al.[5] proposed the GoogleNet with more than 20 layers,
which increased the depth of CNN, improved the utilization rate of
the computer, reduced the parameters, and improved the accuracy.
In 2016, through a series of correction methods that can increase
accuracy and reduce computational complexity, Inception V2 and
Inception V3 were proposed in the paper[6]. He et al.[7] used a
residual network to solve the problem of vanishing gradients, so
that the underlying network can be fully trained. As the depth
increases, so does accuracy. The idea of cross-channel connection
was further extended to multi-layer connections by DenseNet to
improve representation[8]. In 2018, Khan[9] introduced a new
channel improvement idea. The motivation for network training
with channel boosted representations is to use rich representations.
This idea effectively improved the performance of CNN by
learning various features. In 2019, Hou et al.[10] proposed a
method for selecting channels based on the relative of activation,
and proposed weighted channel discarding for regularization of
convolutional layers in CNN.
With the development of deep learning, crop disease
identification has been developed, which not only reduces the
workload but also improves the efficiency of pest identification.
Zeng et al.[11] developed a CNN model with high-order residuals
and parameter sharing feedback to apply to crop disease
recognition in an actual environment. The recognition accuracy
and robustness were better than other methods. Zhang et al.[12]
used the model of VGG 16 to classify the apple leaves disease with
higher accuracy. Amanda et al.[13] proposed use transfer learning
to train a CNN, which had higher recognition accuracy in cassava
disease pest recognition. Mohanty et al.[14] trained the CNN with
54306 healthy and morbid leaf images, and used it to identify 14
kinds of crops and 26 kinds of diseases. Lu et al.[15] used deep
CNN to identify rice leaf diseases, which was more accurate than
traditional machine learning models. Zhang et al.[16] used the
206 July, 2020 Int J Agric & Biol Eng Open Access at https://www.ijabe.org Vol. 13 No.4
LeNet model to identify the diseases of cucumber, which was more
accurate than traditional methods. Huang et al.[17] proposed that
GoogleNet was used to identify disease images of spikes, and the
classification effect was obvious. In 2017, the capsule network
was proposed by Sabour et al.[18]. Since CNN cannot learn spatial
relationships, the pooling layer will lose the information, and the
capsule will adjust the output according to the changes. Deng et
al.[19] proposed the capsule network to classify hyperspectral
images, and the classification accuracy rate exceeded CNN. In
2018, Gan et al.[20] established a hyperspectral inversion model for
chlorophyll content prediction of longan leaves using sparse
self-encoding of classic models of deep learning. The accuracy
can be greatly improved by using deep learning methods. Zhu et
al.[21] used the improved faster region-based convolutional network
(Faster-RCNN) to identify plant leaves, and achieved a high
recognition accuracy than Faster RCNN in the complex
background.
With the increase of network depth, large network models tend
to ignore light feedback errors, which lead to lower convergence
rates[7]. Finally, the large deepening model itself tends to ignore
the details of large-scale data. In view of the above problem, this
study proposes the differential amplification convolutional neural
network (DACNN), which can amplify small differences between
the real output and the expected output. And it has achieved good
results in the identification of wheat leaf disease images. The
differential amplifier branches constructed in the deep neural layers
can make the model more sensitive to the light error of each
iteration feedback. It can alleviate the error omission. Since
there is no large-scale wheat leaf disease images dataset at present,
and the wheat leaf disease dataset was constructed.
2 Materials and methods
The DACNN contains 6 convolutional layers, 3 max-pooling
layers and 3 fully connected layers. To improve the ability of
feature extraction, 3×3 kernels are used to replace the larger
kernels and convolution kernels are fully connected in the last
two layers. In order to alleviate the omission of minor errors in
the backpropagation pass, a branch is added before and after the
deep convolution layer of the differential amplifier, so as to
simulate the difference which achieves the function of error
amplification. In Figure 1, the structure of the traditional CNN
is compared with that of the differential amplification branch, and
the advantage of the latter in the error amplification effect is
proved by theoretical analysis.
Figure 1 Structure of differential amplification branch
2.1 Differential amplification branch
Scheme 1 in Figure 1 is the schematic diagram demonstrating
the CNN that does not add a branch in deep neural layers, similar to
the traditional CNN, whose data stream can be represented by
Equation (1).
1()
l i l l
i
T E w x b
(1)
where, w1 and b1 are the weight matrix and the bias of the lth neural
layer, respectively; xl is the mapping input and Tl+1 is mapping
result of the lth neural layer, respectively, and E() is a linear
activation function.
Scheme 2 in Figure 1 is the schematic diagram demonstrating
the CNN that adds a differential amplification branch in DACNN.
Its data stream satisfies Equation (2).
1( , , )
( , , ) ( ), 0,1,2,...,
l l l l l l
l l l l i l l
i
H x F x w b
F x w b E w x b l L
(2)
where, wl and bl are the weight matrix and the bias of the lth neural
layer, respectively; xl and Hl+1 are the mapping input and mapping
results of the lth neural layer, respectively; Fl () is the mapping
output of convolutional layers and E() is the linear activation
function. Compared to Scheme 1, this structure can strip the
unchanged part xl and highlight the minor change of Fl (xl, wl, bl),
thus making the model more sensitive to error of the back-
propagation pass during each iteration.
Suppose the input feature map is 100. It is expected mapping
results and the actual mapping results in the convolutional layer are
105 and 110 respectively, and Δf6=5, as is shown in Equation (3).
65
65
5
( ) 105
( ) 110
100
fx
fx
x
(3)
f6 and f6′ represent the expected mappings and actual mappings of
the convolutional layer, respectively. ‘′’ represents functions and
variables, etc. in actual situations. In Scheme 1, the ΔT6 is 5
which is shown in Equation (4).
6 5 5
6 5 5
( ) 105
( ) 110
i
i
i
i
T E w x b
T E w x b
(4)
The proportion of ΔT6 is shown in Equation (5).
6
6
6
50.0476
105
TT
PT
(5)
In Scheme 2, there is
6 5 5 5 5 5
6 5 5 5 5 5
( , , ) 105
( , , ) 110
H x F x w b
H x F x w b
(6)
July, 2020 Dong M P, et al. Novel method for wheat leaf disease images based on differential amplification convolutional neural network Vol. 13 No.4 207
Naturally,
5 5 5 5 5 5
5 5 5 5 5 5
( , , ) ( ) 5
( , , ) ( ) 10
i
i
i
i
F x w b E w x b
F x w b E w x b
(7)
And Δf5 = 5 are got. The proportion of Δf5 is shown in
Equation (8).
5
5
5
51
5
FF
PF
(8)
Obviously,
5
F
P
in Scheme 2 is much larger than
6
T
P
in
Scheme 1. Therefore, the network structure in Scheme 2 can
enlarge the error in backpropagation pass between the expected
output and the actual output, which is beneficial to the correct
convergence of the model.
Then, in Scheme 2, Equation (10) can be obtained by recursive
Equation (9).
2 1 1 1 1
1 1 1
()
( ) ( )
l l i l l
i
l i l l i l l
ii
x x E w x b
x E w x b E w x b
(9)
1()
L
L l i i i
i l i
x x E w x b
(10)
For the initial input x0 the mapping result of the Lth neural
layer satisfies Equation (11).
1
00()
L
L i i i
ii
x x E w x b
(11)
From Equations (10) and (11), it can be seen that the
differential amplification effect can be accumulated layer by layer,
thus improving the fitting ability of the model to image pixel
distribution and the identification accuracy to a maximum extent.
2.2 Normalized layers
As noted above, owing to the influence of sunlight, water mist,
dust, and other factors, the range of the signal intensity in gathered
images is extremely wide. Signals with wide ranges of values
often play a major role in model learning, and smaller range signals
have less effect, thus affecting the trend of model coverage.
Moreover, the range of the function domain is limited, so the input
data need to be mapped into this domain. To solve the above
problems, the local response normalization (LRN) is used before
and after the differential amplification branch.
By creating a competition mechanism, LRN can make the
activity of local neurons with the larger response, inhibit other
neurons with smaller feedback, which improves the generalization
ability of the model, and prevent the data from overfitting[2], as is
shown in Equation (12).
min( 1, /2)
( ) ( ) (j) 2
, , ,
j=max(0, / 2)
/ ( ( ) )
N i n
ii
p q p q p q
in
y x k x
(12)
where,
()
,
i
pq
y
is the normalized value, and i is the position of the
channel, which represents the value of the update channel, and p
and q represent the position of the pixel. And the
()
,
i
pq
x
is the
input value, α = 0.0001 is the scaling factor, β = 0.75 is the
exponential term, n = 5 is the local size of the normalized range.
2.3 Dropout
In order to improve the generalization ability and inhibit
overfitting, the dropout strategy[22] is introduced in the differential
amplification branch. When the network propagates forward, it
stops a neuron with a certain probability of p, its activation function
value change from probability p to 0. Dropout reduces the
dependence between neurons by forcing a neuron to interact with
randomly selected neurons and prevents some features from having
effect only under other specific features. So that dropout can
improve the generalization ability of the model. The dropout rate
is set to 0.5 in this study, that is to say, when the neurons pass
dropout, half of them will be set to 0. Figure 2 illustrates the
training process of DACNN with dropout.
Figure 2 Learning processing of DACNN with dropout
In Equation (13), Bernoulli function is used to generate
probability B vector, that is, randomly generate a vector of 0 and 1.
And
()l
i
y
is the input of the upper layer. As is shown in
Equation (14),
()l
i
y
is multiplied by the
()l
i
B
to obtain the
processed signal after masking
()l
i
y
. The output
( 1)l
i
y
is then
calculated by the Equations (15) and (16). The whole procedure
is indicated below.
() ( )
l
i
B Bernoulli distribution p
(13)
( ) ( ) ( )l l l
i i i
y B y
(14)
( 1) ( 1) ( ) ( 1)l l l l
i i i i
y w y b
(15)
( 1) ( 1)
()
ll
ii
y f y
(16)
2.4 Exponential linear unit
In this paper, we use the exponential linear unit (ELU) as the
nonlinear activation function, as shown in Equation (17).
0
( 1) 0
x
x if x
ye if x
(17)
ELU is an improved version of the Rectified Linear Unit
(ReLU). Compared to the ReLU function, when the input is
negative, it has a certain output. As shown in Figure 3, the linear
part of the right segment can alleviate the gradient disappearance,
while the soft saturation end makes it more robust to input changes
and noise at the left. The mean value of the output is close to 0,
and the convergence speed of the ELU is fast.
Figure 3 Exponential linear unit
2.5 Experimental setup
The computer model is HP EliteDesk 880 G2 TWR, the
processor is Intel(R) Core(TM) i7-6700K CPU @ 3.40 GHz, and
the RAM is 16 GB. Furthermore, the operating system is Ubuntu
14.04.4 64 bits. Training a deep CNN on the large-scale images
through a large number of iterations largely relies on GPUs with
the high performance. Its basic configuration is listed in Table 1.
The Python is utilized as the programming language to adapt to the
208 July, 2020 Int J Agric & Biol Eng Open Access at https://www.ijabe.org Vol. 13 No.4
core of TensorFlow.
Table 1 Basic Characteristics of GPUs
Configuration parameter
Parameter value
Chip model
NVIDIA GeForce GTX 1080
RAM capacity
8192M
RAM interface
256-bit
Core frequency
1759/1936 MHz
Memory frequency
10206/10400 MHz
Stream processor
2560
Raster processing unit
64
RAMDAC frequency
400 MHz
Maximum resolution
7680×4320
3 Construction of the dataset
As there are no large-scale images of wheat leaf diseases,
therefore, images were collected from several wheat planting bases
in Shandong province. Then, they were expanded by 5 kinds of
data augmentation techniques to construct the wheat leaf disease
dataset. It is expected that these experiments can shorten the
distance between the theoretical research of neural networks and
the practical agricultural application.
3.1 Acquisition of images
The wheat leaf images were collected from the wheat planting
bases of Shandong Province of China. The number of the original
dataset is 8326, containing normal leaf and 7 kinds of diseases,
which are mechanical damage leaf, powdery mildew, bacterial leaf
streak, cochliobolus heterostrophus, stripe rust, leaf rust and
bacterial leaf blight. The images were taken with a Canon EOS
80D (18-200 mm). The image format is JPEG and each image is
a 24-bit color bitmap. The numbers and proportions of the wheat
leaf disease image in the original dataset are shown in Table 2, and
the samples of wheat leaf disease images are shown in Table 3.
Table 2 Number and proportion of the wheat disease image in original dataset
Name
Normal leaf
Mechanical
damage leaf
Powdery
mildew
Bacterial leaf
streak
Cochliobolus
heterostrophus
Stripe rust
Leaf rust
Bacterial leaf
blight
Number
1016
1237
1182
962
1046
939
1061
883
Proportion/%
0.122
0.148
0.141
0.115
0.125
0.112
0.127
0.110
Table 3 Samples of wheat leaf disease images
Normal
leaf
Mechanical
damage leaf
Powdery
mildew
Bacterial
leaf streak
Cochliobolus
heterostrophus
Stripe
rust
Leaf
rust
Bacterial leaf
blight
Sample 1
Sample 2
Sample 3
3.2 Data preprocessing
The CNN self-learning relies on iterative training on a
large-scale dataset. If the amount of data is too small, it is prone
to cause the overfitting, which makes the training error very small
while the testing error very large[23]. In order to increase the size
and diversity of original dataset, 5 ways are adopted to implement
dataset augmentation which are add Gaussian noise, color jittering,
fancy PCA, mirror horizontally and Gaussian blur, as shown in
Table 4, and the images processed by the methods of data
augmentation are shown in Table 5.
Data augmentation can produce 6 corresponding enhanced
images of every category of wheat leaf disease images. Finally,
the number of data augmentation of wheat leaf diseases is 41630,
the number and proportion of each kind of wheat leaf disease
images are shown in Table 6.
Table 4 Method of data augmentation
Name
Detail operations
Gaussian noise
Add 30% Gaussian noise to the original image.
Color jittering
Increase saturation and brightness by 20% and contrast by 30%.
Fancy PCA
Change the intensity of RGB channel, then perform PCA on all RGB pixel values, and obtain a 3×3 covariance matrix; A new covariance
moment is obtained by multiplying the eigenvalue by a random variable with a mean value of 0 and a standard deviation of 0.1 Gaussian
distribution.
Mirror horizontally
Mirror the left and right parts of the image with the vertical central axis of the image as the center.
Gaussian blur
Each pixel takes the average value of the surrounding pixels, when calculating the average value, the fuzzy was affected by the blur radius,
and the blur radius is set to 2.
July, 2020 Dong M P, et al. Novel method for wheat leaf disease images based on differential amplification convolutional neural network Vol. 13 No.4 209
Table 5 Images processed by the data augmentation
Disease type
Original image
Gaussian Noise
Color jittering
Fancy PCA
Mirror horizontally
Gaussian blur
Stripe
rust
Bacterial leaf blight
Cochliobolus
heterostrophus
Leaf
rust
Powdery mildew
Table 6 Number and proportion of the wheat disease images in dataset
Name
Normal leaf
Mechanical
damage leaf
Powdery
mildew
Bacterial leaf
streak
Cochliobolus
heterostrophus
Stripe
rust
Leaf
rust
Bacterial
leaf blight
Number
5080
6185
5910
4810
5230
4695
5305
4415
Proportion/%
0.122
0.148
0.141
0.115
0.125
0.112
0.127
0.110
4 Results and discussion
4.1 DACNN-Softmax, DACNN-SVM, DACNN-KNN and
DACNN-Random Forest
In this experiment, DACNN is combined with softmax,
support vector machine (SVM), K-nearest neighbor (KNN), and
Random Forest to identify the augmented dataset, which aims at
investigating the effect of different classifiers on identification
results by observing their trend of accuracy change. In KNN, k is
set to 100. Radius Basis Function (RBF) is used in SVM. The
penalty parameter C, γ, and slack variable ζ are initialized to 10,
0.02 and 0.001, respectively. The number of decision trees in
Random Forest is 200, and the Gini index is used:
2
( ) 1 c
i
i
Gini D p
, where c represents the number of categories
in the dataset and is set to 8. pi represents the proportion of the ith
category of samples in all samples. The 4 models are iterated
50000 times on the augmented dataset, and save the intermediate
model every 5000 iterations and validation it with the test dataset.
Their change procedure of identification accuracy is shown in
Figure 4.
Figure 4 Accuracy of DACNN-Softmax, DACNN-SVM,
DACNN-KNN and DACNN-Random Forest
It can be seen from Figure 4 that when the models are
convergent, the identification accuracy of DACNN-SVM and
DACNN-Softmax are 95.32% and 96.09%, respectively, which is
obviously superior to the accuracies of DACNN-KNN and
DACNN-Random Forest of 90.37% and 89.96%. Furthermore,
through the experiment process, we can see that the identification
accuracy of DACNN-SVM is higher than that of
DACNN-Softmax when the number of iterations is small. This
is because the number of iterations is small and the data
throughput is small in the early experiment, and SVM is just a
classification algorithm based on statistical learning theory,
which replaces Empirical Risk Minimization (ERM) with
Structure Risk Minimization (SRM). It is suitable for small
sample data classification, so it has higher recognition accuracy
than Softmax in the early stage.
4.2 DACNN, Inception V3, LeNet-5, AlexNet and ZFNet
In order to verify the performance of DACNN, it is compared
with Inception V3, Lenet-5, AlexNet and ZFNet. LeNet-5
consists of 3 convolutional layers, 2 subsampling layers, and 3
fully connected layers, which have been widely used in digital
handwriting recognition; Both AlexNet and ZFNet contain 5
convolutional layers, 3 subsampling layers, and 3 fully connected
layers. However, the former uses two GPU sparse connection
structures, while ZFNet uses only one GPU dense connection
structure. Inception V3 works by performing multiple
convolution and pooling operation on the image and outputs a deep
feature map. In the above experimental environment, the 5
models are iterated 50 000 times on the augmented dataset and save
the intermediate model every 5 000 iterations and validation it with
the test dataset. The training process of the model is shown in
Figure 5.
It can be seen from Figure 5 that when the number of iterations
is close to 25 000, the DACNN begins to converge, the average
identification accuracy of DACNN is about 95.18%, which is
higher than the accuracy of Inception V3 94.31%, AlexNet 91.54%
and ZFNet 92.79%, and is obviously higher than the accuracy of
LeNet-5 89.15%. DACNN owns higher identification accuracy
for the wheat leaf disease images. The error amplification effect
of DACNN can be accumulated layer by layer, which makes the
network more capable of fitting the pixel distribution of the image
and improves the classification accuracy.
210 July, 2020 Int J Agric & Biol Eng Open Access at https://www.ijabe.org Vol. 13 No.4
Figure 5 Training process of DACNN, Inception V3, Lenet-5,
AlexNet and ZFNet
5 Conclusions
In this study, we deal with the recognition of the wheat leaf
disease image by proposing a novel method named DACNN, a
differential amplification convolutional neural network.
Especially, branches before and after the deep convolution layer in
DACNN were added to simulate the differential amplifier and
realize the function of error amplification. Then, there is no
standard dataset of wheat leaf diseases, constructing the wheat leaf
disease dataset. Finally, the experimental results with
Inception-V3, AlexNet, ZFNet and LeNet-5 and combined with
four classifiers, which are Softmax, SVM, KNN and Random
Forest on the wheat leaf diseases dataset show the superiority of
DACNN. For future work, we plan to apply DACNN to other
types of visual tasks, such as object detection.
Acknowledgements
This work is supported by First Class Discipline Funding of
Shandong Agricultural University (XXXY201703).
[References]
[1] Lecun Y, Bottou L, Bengio Y, Haffner P. Gradient-based learning
applied to document recognition. Proc.IEEE, 1998; 86(11): 2278–2324.
[2] Krizhevsky A, Sutskever I, Hinton G E. ImageNet classification with
deep convolutional neural networks. Proceedings of Advances in Neural
Information Processing Systems, 2012; 25: 1097−1105.
[3] Zeiler M D, Fergus R. Visualizing and understanding convolutional
networks. Computer Vision-ECCV, IEEE, 2014; 8689: 818–833.
[4] Simonyan K, Zisserman A. Very deep convolutional networks for
large-scale image recognition. arXiv preprint, 2014; 6: 1–47. arXiv
1409.1556.
[5] Szegedy C, Liu W, Jia Y, Sermanet P, Reed S, Anguelov D, et al. Going
deeper with convolutions. In: 2015 IEEE Conference on Computer
Vision and Pattern Recognition (CVPR), Boston: IEEE, 2015; pp.1–9.
[6] Szegedy C, Vanhoucke V, Ioffe S, Shlens J, Wojna Z. Rethinking the
inception architecture for computer vision. IEEE Conference on
Computer Vision and Pattern Recognition, IEEE, 2016; pp.2818–2826.
[7] He K M, Zhang X Y, Ren S Q, Sun J. Deep residual learning for image
recognition. In: 2016 IEEE Conference on Computer Vision and Pattern
Recognition (CVPR), Las Vegas: IEEE, 2016; pp.770–778.
[8] Huang G, Liu Z, Laurens V D M, Weinberger K Q. Densely connected
convolutional networks. Computer Vision and Pattern Recognition, IEEE,
2017; pp.4700-4708. doi: 10.1109/CVPR.2017.243.
[9] Khan A, Sohail A, Ali A. A new channel boosted convolutional neural
network using transfer learning. arXiv preprint, 2018. arXiv:1804.08528.
[10] Hou S H, Wang Z L. Weighted channel dropout for regularization of
deep convolutional neural network. AAAI Conference on Artificial
Intelligence, 2019; 33: 8425–8432.
[11] Zeng W H, Li M, Li Z, Xiong Y. High-order residual and
parameter-sharing feedback convolutional neural network for crop disease
recognition. Acta Electronica Sinica, 2019; 47(9): 1979–1986.
[12] Zhang K, Guo Y R, Wang X S, Yuan J S, Ding Q L. Multiple feature
reweight Dense Net for image classification. IEEE Access, 2019; 7:
9872–9880.
[13] Amanda R, Kelsee B, Peter M, Babuali A, James L, David P. Deep
learning for image-based cassava disease detection. Frontiers in Plant
Science, 2017; 8: 1852. doi: 10.3389/fpls.2017.01852.
[14] Mohanty S P, Hughes D P, Salathé M. Using deep learning for
image-based plant disease detection. Frontiers in Plant Science, 2016; 7:
1419. doi: 10.3389/fpls.2016.01419.
[15] Lu Y, Yi S J, Zeng N Y, Liu Y R, Zhang Y. Identification of rice
diseases using deep convolutional neural networks. Neurocomputing,
2017; 267(Dec.6): 378–384.
[16] Zhang S W, Xie Z Q, Zhang Q Q. Application research on convolutional
neural network for cucumber leaf disease recognition. Jiangsu Journal of
Agricultural Sciences, 2018; 34(1): 56 – 61.
[17] Huang S P, Sun C, Qi L, Ma X, Wang W J. Rice panicle blast
identification method based on deep convolution neural network.
Transactions of the CSAE, 2017; 33(20): 169 – 176.
[18] Sabour S, Frosst N, Hinton G E. Dynamic routing between capsules.
Neural Information Processing Systems, 2017; pp.3856–3866.
[19] Deng F, Pu S L, Chen X H, Shi Y S, Yuan T, Pu S Y. Hyperspectral
image classification with capsule network using limited training samples.
Sensors, 2018; 18(9): 3153. doi: 10.3390/s18093153.
[20] Gan H M, Yue X J, Hong T S, Ling K J, Wang L H, Cen Z Z. A
hyperspectral inversion model for predicting chlorophyll content of Longan
leaves based on deep learning. Journal of South China Agricultural
University, 2018; 39(3): 102–110. (in Chinese)
[21] Zhu, X L, Zhu M, Ren H. Method of plant leaf recognition based on
improved deep convolutional neural network. Cognitive Systems
Research, 2018; 52(Dec.): 223–233.
[22] Srivastava N, Hinton G, Krizhevsky A Sutskever I, Salakhutdinov R.
Dropout: A simple way to prevent neural networks from overfitting.
Journal of Machine Learning Research, 2014; 15: 1929–1958.
[23] Hu J L, Lu J W, Tan Y P, Zhou J. Deep transfer metric learning. IEEE
Transactions on Image Processig, 2016; 25(12): 5576–5588.
