Hindawi
Journal of Food Quality
Volume 2022, Article ID 9211700, 12 pages
https://doi.org/10.1155/2022/9211700
Research Article
Design and Evaluation of a Hybrid Technique for Detecting
Sunflower Leaf Disease Using Deep Learning Approach
Arun Malik ,1 Gayatri Vaidya ,2 Vishal Jagota ,3 Sathyapriya Eswaran ,4
Akash Sirohi ,1 Isha Batra ,1 Manik Rakhra ,1 and Evans Asenso 5
1Department of Computer Science and Engineering, Lovely Professional University, Phagwara, Punjab 144411, India
2Department of Studies in Food Technology, Davangere University, Davangere, Karnataka, India
3Department of Mechanical Engineering, Madanapalle Institute of Technology & Science, Madanapalle, Andhra Pradesh, India
4Department of Agricultural Extension, Amrita School of Agricultural Sciences, Amrita Vishwa Vidyapeetham University,
Coimbatore 642109, India
5Department of Agricultural Engineering, School of Engineering Sciences, University of Ghana, Accra, Ghana
Correspondence should be addressed to Evans Asenso; easenso@ug.edu.gh
Received 8 January 2022; Revised 22 February 2022; Accepted 10 March 2022; Published 8 April 2022
Academic Editor: Muhammad Faisal Manzoor
Copyright © 2022 Arun Malik et al. &is is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Agriculture and plants, which are a component of a nation’s internal economy, play an important role in boosting the economy of
that country. It becomes critical to preserve plants from infection at an early stage in order to be able to treat them. Previously,
recognition and classification were carried out by hand, but this was a time-consuming operation. Nowadays, deep learning
algorithms are frequently employed for recognition and classification tasks. As a result, this manuscript investigates the diseases of
sunflower leaves, specifically Alternaria leaf blight, Phoma blight, downy mildew, and Verticillium wilt, and proposes a hybrid
model for the recognition and classification of sunflower diseases using deep learning techniques. VGG-16 andMobileNet are two
transfer learning models that are used for classification purposes, and the stacking ensemble learning approach is used to merge
them or create a hybrid model from the two models. &is work makes use of a data set that was built by the author with the
assistance of Google Images and comprises 329 images of sunflowers divided into five categories. On the basis of accuracy, a
comparison is made between several existing deep learning models and the proposed model using the same data set as the
original comparison.
1. Introduction cure bladder disorders, and it can also act as an antioxidant.
Sunflower leaves have several uses in animal husbandry and
Sunflower originated in 2100 BCE in Mexico, and it is also many industries. Sunflower leaves are affected by many
known as Helianthus. Sunflower seeds and their leaves have diseases, but in this thesis, Alternaria leaf blight, Phoma
several uses and benefits. Sunflower can be used for food blight, downy mildew, and Verticillium wilt are considered:
because it has many nutrients in its seeds and leaves.
Sunflower roots have the soaking ability by which it is able to Alternaria leaf blight: Alternaria helianthi is the fungal
soak the radioactive substance too. It is also a good source of plant pathogen responsible for Alternaria leaf blight, as
vitamins. Due to its therapeutic properties, sunflower is also shown in Figure 1. South Africa is the major area of
used in the treatment of various diseases such as malaria, Alternaria diseases. It is a potential disease occurring in
arthritis by reducing swelling, gastroenteritis, chest pain, and the producing areas of sunflower.
respiratory tract disorders. Sunflower leaves have properties Symptoms: concentric rings of 0.2–0.5mm diameter
that can cure insect bites, snake bites, spider bites, head- with dark brown to black lesion appear on the leaves
aches, etc. Its leaves have diuretic properties by which it can and stems. When the spores on the leaves or stems
2 Journal of Food Quality
Figure 1: Alternaria leaf blight. Figure 2: Downy mildew.
come in the contact of moisture and start penetrating,
then the infection process starts.
Treatment: fungicides are sprayed directly on infected
plants, coupled with improved sanitation and crop
rotation.
Downy mildew: this disease is caused by a plant
pathogen called Plasmopara halstedii. Plasmopara
halstedii oospores produce thin walls, which are re-
sistant structures, sexually produced that are funda-
mental for its continuation, as shown in Figure 2.
Entering a territory, the annihilation of the microbe is
troublesome because of the arrangement of oospores,
which can stay in soil for a long time. Figure 3: Phoma blight.
Symptoms: initial symptoms are visible on the upper
surface like large, angular or blocky, yellow areas.
&ey rapidly expand and become brown-like lesions,
mature. &e under surface of infected leaves appears
water-soaked.
Phoma blight: this disease is caused by the plant
pathogenic fungus called Phomamacdonaldii, as shown
in Figure 3, and it is also a common disease caused by
soil-borne fungi.
Symptoms: this is perceived as a huge dull sore on the
stem, begins from the leaf and reaches the petiole. &e
enormous patches on the tail become generally per-
ceptible after the petal drops.
Treatment: 4-year crop rotation is a good treatment Figure 4: Verticillium wilt.
for curing Phoma blight, which reduces the disease by
controlling stem weevil.
Verticillium wilt: Verticillium dahliae, a soil-borne computer vision methods to classify or recognize the dis-
fungus, is the main cause of the disease. It is also called eases in the earlier stage.
“Vert,” and it starts from the lower leaves and moves to In the past, diseases were classified manually without the
the whole leaves, as shown in Figure 4. use of electronic devices, which required more time, wasmore expensive, and had a higher chance of error because
Sunflower and its leaves have various uses and benefits. the entire process was carried out by humans. However,
&ey are beneficial for human beings, animals, and the computer vision (also known as machine learning) has made
environment; so if it gets infected, then it is a big loss to the it possible to reduce the processing time of classification and
environment. If we can identify or recognize the diseases in also provide better accuracy than the manual method. It is
the initial stage, then we can save the sunflower plant and its also capable of classifying multiple images at the same time.
leaves, thus saving the environment and source of many In the field of machine learning, there are three main
nutrients. &ere are some ancient methods and some learning approaches: supervised learning, unsupervised
Journal of Food Quality 3
learning, and reinforcement learning. In machine learning, 2. Related Work
there are many different approaches, such as linear re-
gression, decision tree, logistic regression, SVM, random Zhong and Zhao [11] used the DenseNet-121 deep learning
forest, Naive Bayes, KNN, and k-means. Although machine technique to classify the 6 apple diseases with three methods
learning algorithms give decent accuracy, they fall short of for which they used the data set of 2462 images, and they
the target, i.e., we were unable to identify images with high concluded that their proposed method gave the accuracy of
accuracy when employing machine learning algorithms. 93.51%, 93.31%, and 93.71%, respectively.
When it comes to text value prediction and categorization, Ji et al. [12] proposed a CNN based on an integrated
machine learning algorithms are the most effective tools. method to classify the grape leaf diseases for which they used
Deep learning began to have an influence on picture cate- a data set from the PlantVillage database, and they con-
gorization at around this time. cluded that their proposed method gave 99.17% and 98.57%
Early identification and prediction of plant diseases is accuracy for validation and test, respectively.
one of the most crucial needs for developing agriculture, Uguz and Uysal [13] developed a CNN model to classify
which is important to our country’s economy. It supports the olive leaf diseases, i.e., Aculus olearius and olive peacock
the economy and feeds a large population. And, by earlier spot disease.&ey used the data set of 3400 olive images from
detection, we can conserve the plants and avoid losses. Turkey and used the transfer learning techniques, i.e., VGG-
Deep learning techniques are frequently used to classify or 16 and VGG-19, and they concluded that their proposed
forecast outcomes [1, 2]. Deep learning techniques are model gave an accuracy of 95%.
used as active methods for the classification of plant Nanehkaran et al. [14] proposed a CNN method for the
diseases [3], and the main classifier used is convocational detection of leaf diseases, and they divided their method into
neural network (CNN). &e CNN is one of the most two parts—image segmentation and image classification;
recommended models for the classification or recognition they also proposed a segmentation algorithm based on in-
with the help of images whether we have a large or small tensity and LAB. &ey concluded that their proposed
data set [4]. As deep learning dynamically analyses method gave a detection accuracy of 75.59%.
structured characteristics, there is no need to manually Jiang et al. [15] proposed the hybrid method of CNN and
design the feature extraction function and classifier. Deep SVM to classify rice diseases. &ey used CNN for the feature
learning methods surpass machine learning as image extraction of disease images and then used SVM for the
classifiers with CNN being the best. &e CNN is widely classification of the diseases with a 10-fold cross-validation
recognized as the finest and most effective image classifier method. &ey concluded that their proposed method gave a
on both small and large data sets, and it serves as the test accuracy of 96.8%.
foundation for all deep learning models [5, 6]. It is a basic Ghosal et al. [16] developed a CNNmodel with the help of
deep learning model used for the classification of images VGG-16 for the classification of rice leaf diseases, and to train
according to some patterns or features. One of the best the model, they created their own data set of rice images and
properties is that CNN trained in a supervised manner concluded that their method gave an accuracy of 92.46%.
with the help of existing data [7]. &e CNN is a basic Jasim et al. [3] proposed a method for the detection and
architecture over which many models are formed such as classification of tomatoes, pepper, and potato leaf diseases.
AlexNet, GoogLeNet, and LeNet. &e CNN architecture is &ey collected 20,636 images from the PlantVillage database
presented in Figure 5. &e CNN attains the name of best to make their data set. &ey classified 12 leaf disease classes
image recognition system and is the most recommended and 3 healthy leaf classes with the help of the CNN method
system for the purpose of recognition and classification concluded that their model gave a training accuracy of
[8]. It performs well in the plant disease detection task. It 98.29%, and 98.029% for testing.
is the finest technique for object identification. Any neural Md. Rasel Howlader et al. in [17] have studied guava leaf
architecture should be able to be paired with any feature diseases, using deep CNN to see infection, and their designs
exactor, depending on the requirements. Data pre- characterise for important diseases, for example, algal leaf
processing is necessary for models to operate correctly. area, whitefly, and rust.&ey collected their own data set and
Many infections (viral or fungal) may be difficult to even developed their own model. &ey conclude that their
identify due to overlapping signs [9, 10]. model gave an accuracy of 98.74%.
Other strategies are based on ensemble learning, which is &e mango leaf disease called anthracnose was managed
used to develop several models and then merge the models by Singh et al. [18]. In their study, they tried to suggest a
with the assistance of ensemble methods to enhance the concept in addition to a monetarily savvy program for which
results. Most of the time, the ensemble technique outper- they make use of a multilayer convolutional neural com-
forms the single model in terms of performance. Some of the munity (MCNN). &ey prepared their model with a con-
ensemble approaches are bagging (also known as boosting), tinuing image on the Faculty of J&K. &ey concluded their
majority voting (also known as weighted average), and unit provided an accuracy of 97.13% for sickness called
stacking (also known as stacking ensemble). In this research, anthracnose.
two approaches of ensemble learning technique—namely, Geetharamani and Arun [19] proposed a disease ID type
stacking ensemble and weighted average technique—are of a novel grow leaf which is determined by Deep CNN.&ey
applied, and the outcomes of the both are compared with prepared the model of theirs with a receptive dataset of
one another. thirty-nine photos as well as groundwork photographs. &is
4 Journal of Food Quality
Input convolution sampling convolution sampling
Layer 1 1 2 2
Input Data Input Data
Feature Extraction And Classification
Figure 5: CNN architecture.
study includes six parameters on photos as turn, gamma A neural method to cluster the ailments of vegetation
remedy, clamor infusion, PCA, and scaling, and then photo using an image was proposed by Sijiang Huang et al. in [24].
flipping is utilized by them.&ey declared due to the exercise A unique deep neural network structure in this study is used
exhibitions on the purposed method are expanded. &ey that can accurately categorise plant kinds and illness using a
contrast the model of theirs and also the previous body as single picture of a plant leaf. &e proposed model is com-
well as the main reason which their unit provides much posed of two sub-models: a leaf segmentation model that
better results with 96.46% exactness. utilises a U-Net to effectively separate the leaves in the
Jiang et al. [20] took care of 5 types of infection in original image from the background, and a plant disease
apple leaves by making use of the deep learning approach classification model that utilises two-head network to
based on improved convolutional neural structures. Same classify plant diseases using features extracted from various
authors prepared the model of theirs with frequent popular pre-trained models. Experiments reveal that final
photographs of apple leaf within the wake of using a few model obtains a classification accuracy of 0.9807 for plants
images controlling methods. &ey used the GoogLeNet and a disease identification accuracy of 0.8745.
system in addition to upgraded CNN and the utilisation of Singh [25] focused on the use of sunflower in oil pro-
INAR (SSD with Inception component as well as Rainbow duction and agriculture areas. He mentioned that it is dif-
connection) when designing a Panasonic phone. &e ficult to detect the damage or disease of sunflower for a full
proposed model yields 78.80 percent exactness with an farm. &erefore, he proposed an algorithm for segmenting
excellent place velocity of 23.13 fps, which was substan- and classifying the images of sunflower leaf. &e practical
tially better than prior versions. swarm optimization method is used for the classification of
Akshay and Vani [21] made sure that the convolutional disease.
neural structure would become the greatest technique to Sujithra and Ukrit [26] identified that it is not easy to
perceive the maladies. &ey proposed a unit to evaluate the figure out the disease on the leaf as most of the leaves are
infection of tomato foliage and created their very own data found damaged. &ey analyzed various deep leaning and
set from the PlantVillage database. &ey stated that their image processing methods to classify the disease. Further-
model provided 99.25% precision. &ey compared their more, it was said that neutral network algorithms like
model with different neural networks such as ResNet-50, support vector help identifying and classifying leaf diseases.
VGGNet, and also LeNet. Montecchia et al. [27] recognized that the soil-borne
Hasan et al. [22] tried to find the right technique to figure disease that affect sunflower is SVW (sunflower Verticillium
out the jute diseases (chlorosis along with yellowish mosaic). wilt and leaf mottle). &ey carried out their work in infected
Jute was selected for the very first time.&ey used 600 images fields of the Argentina region. &is paper highlights the
as a data set and also inferred their proposed method various disease descriptors depending on disease incidence
provided 96% exactness without using any image-creating and severity.&is paper presents a way for resisting the SVW
method. of sunflower.
Mohit Agarwal et al. in [23] tried to find out the Huu Quan Cap et al. in [28] ascertain that identification
diseases within apple foliage through the use of neural of growing diseases was accomplished in a couple of ways:
telephone system strategies. &ey released a data set from they used only a small photo, i.e., a few of the tips of the
the town's grow the town initiative. &ey utilised a photograph as information, and on the off chance that they
multilayer neural network to create their suggested style used a continuous photo, they saved additional time by
and also compare it to two CNN models, InceptionV3 and utilising a leaf limitation method with deep understanding.
VGG16, to determine which model is more specific than Additionally, their technical precision was 78 percent within
the others and also provides 99% accuracy, as well as 2.0fps.
additionally determine which their unit captures seven Zhang et al. [29] dealt with the ID and also evaluation
seconds for screening. of maize leaf problems. It basically focused on the various
Journal of Food Quality 5
Table 1: Summary of various crops, their diseases, and the referred models and their accuracy.
Author Crop Disease Model used Accuracy
Yong Zhong and Ming
Zhao Apple All disease DenseNet-121 93.71%
Miaomio Ji et al. Grape Black rot, esca, and isariopsis leaf spot CNN 99.17% (validation) and98.57% (testing)
Sinan Uguz and Nese Olive Aculus olearius and olive peacock spotUysal diseases VGG-16 and VGG-19 95%
99.85% (public data set)
Junde Chen et al. Plants Common MobileNet-V2 and 99.11% (collected
data set)
Y.A. Nanehkaran et al. Plants Common CNN+LAB 75.59%
Shreya Ghosal et al. Rice Leaf blast, leaf blight, and brown spot VGG-16 92.46%
Marwan Adnan Jasim Tomato,
et al. pepper, and Common CNN
98.29% (training) and
potato 98.029% (testing)
Junde Chen et al. Plants Common VGGNet + ImageNet 91.83% and 92%
Md. Rasel Howlader
et al. Guava Whitefly, algal leaf spot, and rust D-CNN 98.74%
Uday Pratap Singh
et al. Mango Anthracnose MCNN 97.13%
Geetharamani G. and
Arun Pandian Plants Common D-CNN 96.46%
Md. Zahid Hasan et al. Jute Chlorosis and yellow mosaic CNN 96%
98.07% (classification)
Sijiang Huang et al. Plants Common U-Net and ResNet and 87.45%
(recognition)
S. Santhana Hari et al. Plants Common CNN 86%
Sammy V. Militante
et al. Plants Common CNN 96.5%
Mercelin Francis and Apple and
C. Deisy tomato Common CNN 87%
Jayme Garcia Arnal
Barbedo Plants Common CNN 87%
Md. Helal Sheikh et al. Maize and Gray leaf spot corn, common rust corn, andpeach bacterial spot peach CNN 99.28%
Mehmet Metin
Ozguven and Kemal Sugar beet Low, severe, and low and severe Faster R-CNN 95.48%
Adem
Malik Hashmat Shadab Sugarcane Cercospora leaf spot, helminthosporium leafet al. spot, rust, red rot, and yellow leaf disease YOLO+FR-CNN 93.20%
Radhamadhab dalai
and Kishore Kumar Plants Bacterial canker, gray mold, blossom end rot, R-CNN 75.43%, 67.85%, 72.13%,
Senapati and whitefly and 49.87%
A.S.M. Farhan Al
Haque et al. Guava Anthracnose, fruit rot, and fruit canker CNN 95.61%
Sukhvir Kaur et al. Plants Common Machine learning —
Siddharth Singh
Chouhan et al. Plants Common BRBFNN 89%
Huu Quan Cap et al. Plants Common CNN 78%
Robert G. de Luna et al. Tomato Phoma rot, leaf miner, and target spot FR-CNN 91.67%
Rutu Gandhi et al. Plants Common CNN+GANs —
Konstantinos
P. Feretinos Plants Common VGG 99.53%
K.R. Aravind et al. Maize Cercospora leaf spot, common rust, and leafblight SVM 83.7%
Edna Chebet Too et al. Plants Common DenseNet 99.75%
Yellow leaf curl, bacterial spot, early blight,
Halil Durmus et al. Tomato late blight, leaf mold, Septoria leaf spot, spider AlexNet and 95.65% and 94.3%
mites, target spot, and mosaic virus SqueezeNet
Usama Mokhtar et al. Tomato Powdery mildew and early bright SVM 99.5%
6 Journal of Food Quality
Image
Dataset
Data Pre-
processing
Training
Set
Test Set VGG-16 MobileNet Test Set
Pred icti
on
iction Pre
d
Stacking Ensemble
Learning
Output Prediction
Figure 6: Proposed hybrid technique’s work flow.
diseases on maize leaf that can occur and also focused on CNN. Although CNNs are powerful and required image
how to mitigate those diseases. With the assistance of the processing, deep CNNs are need to be used to diagnose plant
CIFAR10, they found nine distinct diseases and also to diseases [31, 32].
increase precision of the model on various epochs the
model was hyper-tuned. After hyper-tuning, the model
researchers obtained 98.9 percent accuracy, while with 3. Proposed Hybrid Model for Detecting
Cifar10, accuracy achieved was 98.8 percent. &ey dem- Sunflower Leaf Diseases
onstrate how it is possible to increase precision by in- &is work proposes a hybrid deep learning model for
creasing the number of pooling activities, and they also classifying sunflower illnesses using images, which is based
supplied the long-term value by consolidating brand-new on a deep learning hybrid model. Two models—VGG-16
calculations to increase precision or even discriminate and MobileNet—are combined, and one of the ensemble
brand-new maize diseases. learning techniques, stacking, is used to learn the new model
de Luna et al. [30] had taken a go from the 3 ailments combination. Due to the fact that ensemble models out-
of tomato leaves: Phoma rot, target spot, and leaf miner. perform single models in terms of accuracy, the ensemble
&ey used Diamante Max’s kind of tomato for examining. learning approach is applied in this situation. &e accuracy
&ey produced a motor-controlled image-capturing of some existing deep learning techniques was calculated in
package that catches images from every side of a leaf. &ey order to select these two models. It was discovered that
used CNN of deep learning; they compared their model VGG-16 and MobileNet provide better results than the
with previous designs such as Fast R-CNN giving 80% others, with 81% and 86% accuracy, respectively. Further-
accuracy and transfer learning providing 95.57% exact- more, pretrained networks are best suited for small size data
ness and inferred that their model provided 91.67% sets, and because our data set is small in size, these two
accuracy. pretrained networks were selected for this work. &e ap-
Table 1 presents a summary of various crop diseases and proaches of ensemble learning stacking and weighted av-
their detection models. &us, it is understood that the de- erage are used in conjunction with each other.&e work flow
tection of crop diseases is crucial for agricultural production, of the proposed technique is shown in Figure 6.
quality management, and decision-making. Several projects Geometric transformations such as rotation, shifts, scale,
in this field have focused on deep learning, namely, deep zoom, and flip were accomplished via the use of image
Journal of Food Quality 7
start:
(1) import numpy, sklearn, tensorflow, keras as np1, sk, tf, kr
(2) def convert_img_array(img):
If img not None:
Return array of img
else
Return np1.array[]
(3) for I in imgdir://. . .imgdir contains list of images
imgList[].append(convert_img_array(img))
labelList[].append(i.label)//. . .i.label gives label or class of each images
(4) modify labelList with label_binarizer library
(5) image_list� np1.array(imgList, dtype� np.float16)/225.0
(6) train_test_split(image_list, labelList, 0.2,3)
(7) do Augmentation with Keras.ImageDataGenerator()
(8) Implement Models:
from keras.applications.vgg16 import VGG16
from keras.applications.mobilenet import MobileNet
(9) models[].append(VGG16() & MobileNet)
(10) for i in range models:
Models[i].fit_transform();//to train the models
(11) Now combine prediction values;
For i in range models:
Yhat�models[i].prediction(); //stacked data set
(12) Now create stacked model
stackedx� call process 11
Model� LogisticRegression();
13Now to train the stacked model:
Model.fit(stackedx, test_x);
(14) Now store prediction value of stacked model:
Yhat�Model.predict(stackedx)
(15) Now calculate accuracy:
Sklearn.accuracy_score(yhat, test_y)
end;
ALGORITHM 1: Technique for detecting sunflower leaf disease.
Dataset
77 75 7867
32
Alternaria Downy Phoma Verticillium Healthy
Spot Mildew Blight Wilt Leaves
No_of_images
Figure 7: Data set.
augmentation in this paper. &e ImageDataGenerator from 4. Results and Discussion
Keras was used for the data augmentation process. In ad-
dition to improving the images, this strategy helps to prevent &is section compares the proposed hybrid technique with
overfitting by using a batch size number that selects images the existing techniques on the basis of recognition or
for training purposes in a random manner. &e algorithm classification of sunflower leaf diseases. Data sets play an
for the suggested hybrid approach is detailed in the next important role in classification or anything because if we
section. do not have any data, then there is no meaning for
8 Journal of Food Quality
0.825
0.800
0.775
0.750
0.725
0.700
0.675
0.650
0.625
0 5 10 15 20 25
epoch
train
test
Figure 8: AlexNet.
0.6 0.9
0.8
0.5
0.7
0.4 0.6
0.5
0.3
0.4
0.2 0.3
0 5 10 15 20 25 0 5 10 15 20 25
epoch epoch
train train
test test
Figure 9: CNN. Figure 11: Inception_V3.
0.28
0.9
0.8 0.26
0.7
0.24
0.6
0.5 0.22
0.4 0.20
0.3
0.18
0 5 10 15 20 25 0 5 10 15 20 25
epoch epoch
train train
test test
Figure 10: DenseNet-121. Figure 12: LeNet5.
accuracy accuracy
accuracy
accuracy accuracy
Journal of Food Quality 9
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0 5 10 15 20 25
epoch
train
test
Figure 13: MobileNet.
1.0 1.0
0.9 0.9
0.8
0.8
0.7
0.6 0.7
0.5 0.6
0.4 0.5
0.3
0.4
0.2
0 5 10 15 20 25 0 5 10 15 20 25
epoch epoch
train train
test test
Figure 14: ResNet-50. Figure 15: ResNet50V2.
calculating something or predicting something; so to Initially, some existing deep learning techniques such as
achieve high accuracy in any work, we should have a AlexNet, DenseNet-121, ResNet-101, ResNet-50, ResNet-
proper and well-organized data set. In this work, an or- 50v2, Inception_v3, LeNet5, VGG-16, MobileNet, and
ganized data set or the images containing diseases of CNNs are implemented with 7 layers. For the imple-
sunflower is used.&e data set used in this paper was taken mentation of these techniques, the same process as of the
from Google Images. In total, the data set contains 329 proposed model till step 7 is used, and after that with the
images including 67 images of Alternaria leaf spot, 78 help of Keras library, the object of individual technique is
images of healthy sunflower leaf, 77 images of downy initialized and then finally it is trained with the used data set.
mildew, 75 images of Phoma blight, and 32 images of &e accuracies of various techniques were evaluated, and
Verticillium wilt. Later, the data set was split in the ratio of Figures 8–17 show the graph between accuracy and epoch of
80% and 20%, in which 80% images for training and 20% individual model. &is basically depicts that with more
images for testing. In this data set, all the images are of epoch of the model, the accuracy is increased and the model
high quality, which helps in increasing the accuracy and is making real predictions.
somehow increasing the processing speed. Figure 7 shows After calculating the accuracy of thesemodels, i.e., AlexNet,
bifurcation of the data set used. CNN, DenseNet-121, Inception_V3, ResNet-50, ResNet50V2,
accuracy
accuracy
accuracy
10 Journal of Food Quality
1.0
0.8
0.6
0.4
0.2
0.0
0 5 10 15 20 25
epoch
train
test
Figure 16: ResNet-101.
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0 5 10 15 20 25
epoch
train
test
Figure 17: VGG-16.
Comparison between proposed and Other Models
80
60
40
20
0
Figure 18: Comparison between the proposed and other models.
Accuracy
Inception
DenseNet-121
AlexNet accuracy accuracy
Vgg16
MobileNet
ResNet-50
Models
ResNet-50V2
ResNet-101
LeNet-5
CNN
Proposed
Journal of Food Quality 11
ResNet-101, and VGG-16, it can be concluded that proposed lightweight network,” Multimedia Tools and Applications,
technique gives a better accuracy than the others. vol. 79, pp. 1–19, 2020.
[5] G. Rastogi and R. Sushil, “Cloud computing implementation:
key issues and solutions,” in Proceedings of the 2015 2nd
5. Conclusion and Future Trends International Conference on Computing for Sustainable Global
Sunflower has several advantages, and its leaves and seeds Development (INDIACom), pp. 320–324, IEEE, New Delhi,
are used in a variety of disciplines, making sunflower plants India, March 2015.[6] T. K. Lohani, M. T. Ayana, A. K. Mohammed, M. Shabaz,
essential to our well-being as well as to the environment.&e G. Dhiman, and V. Jagota, “A comprehensive approach of
importance of detecting or recognizing infections in sun- hydrological issues related to ground water using GIS in the
flower plants at an early stage cannot be overstated. How- Hindu holy city of Gaya, India,”World Journal of Engineering,
ever, this may be challenging to do manually. It is made p. 6, 2021.
simple by using deep learning methods or models. In this [7] K. P. Ferentinos, “Deep learning models for plant disease
publication, a hybrid approach for the classification or detection and diagnosis,” Computers and Electronics in Ag-
identification of sunflower leaf diseases is suggested, which riculture, vol. 145, pp. 311–318, 2018.
makes use of deep learning techniques to achieve this [8] J. Ma, K. Du, F. Zheng, L. Zhang, Z. Gong, and Z. Sun, “A
classification or recognition. &is paper discusses four recognition method for cucumber diseases using leaf symp-
sunflower diseases, namely, Alternaria leaf blight, downy tom images based on deep convolutional neural network,”
mildew, Phoma blight, and Verticillium wilt. Alternaria leaf Computers and Electronics in Agriculture, vol. 154, pp. 18–24,
blight is a kind of leaf blight that affects the leaves of 2018.
sunflower plants.&e last step is to do a comparison between [9] S. Vyas and D. Bhargava, “Big data analytics and cognitive
the current approaches and the proposed technique. With computing in smart health systems,” in Smart Health Sys-temsSpringer, Singapore, 2021.
the same data set, the proposed approach exceeds the [10] S. N. H. Bukhari, A. Jain, E. Haq et al., “Machine learning-
competition, with an accuracy of 89.2%. &e creation of a based ensemble model for zika virus T-cell epitope predic-
well-organized collection of sunflower illnesses, either by tion,” Journal of Healthcare Engineering, vol. 2021, Article ID
manually gathering the images of sunflowers or by utilizing a 9591670, 10 pages, 2021.
high-configuration system with many epochs, would allow [11] Y. Zhong and M. Zhao, “Research on deep learning in apple
for improved accuracy in the future. &e graphical com- leaf disease recognition,” Computers and Electronics in Ag-
parison between the proposed technique and other tech- riculture, vol. 168, Article ID 105146, 2020.
niques is presented in Figure 18. [12] M. Ji, L. Zhang, and Q. Wu, “Automatic grape leaf diseases
identification via UnitedModel based on multiple convolu-
tional neural networks,” Information Processing in Agricul-
Data Availability ture, vol. 7, no. 3, pp. 418–426, 2020.
&e data used to support the findings of this study are [13] S. Uğuz and N. Uysal, “Classification of olive leaf diseasesusing deep convolutional neural networks,” Neural Com-
available from the corresponding author upon request. puting & Applications, pp. 1–17, 2020.
[14] Y. A. Nanehkaran, “Recognition of plant leaf diseases based
Conflicts of Interest on computer vision,” Journal of Ambient Intelligence and
Humanized Computing, pp. 1–18, 2020.
&e authors declare that there are no conflicts of interest [15] F. Jiang, Y. Lu, Y. Chen, D. Cai, and G. Li, “Image recognition
regarding the publication of this paper. of four rice leaf diseases based on deep learning and support
vector machine,” Computers and Electronics in Agriculture,
vol. 179, p. 105824, 2020.
References [16] S. Ghosal and K. Sarkar, “Rice leaf diseases classification using
[1] S. O. Mezan, S. M. A. Absi, A. H. Jabbar, M. S. Roslan, and CNN with transfer learning,” in Proceedings of the IEEE
M. A. Agam, “Synthesis and characterization of enhanced Calcutta Conference (CALCON), IEEE, Kolkata, India, Feb-
silica nanoparticle (SiO2) prepared from rice husk ash ruary 2020.
immobilized of 3-(chloropropyl) triethoxysilanea,” Materials [17] Howlader andM. Rasel, “Automatic recognition of Guava leaf
Today: Proceedings, vol. 42, pp. 2464–2468, 2021. diseases using deep convolution neural network,” in Pro-
[2] S. F. Suhel, V. K. Shukla, S. Vyas, and V. P. Mishra, “Con- ceedings of the 2019 International Conference on Electrical,
versation to automation in banking through chatbot using Computer and Communication Engineering (ECCE), IEEE,
artificial machine intelligence language,” in Proceedings of the February 2019.
2020 8th International Conference on Reliability, Infocom [18] U. P. Singh, S. S. Chouhan, S. Jain, and S. Jain, “Multilayer
Technologies and Optimization (Trends and Future Directions) convolution neural network for the classification of mango
(ICRITO), pp. 611–618, IEEE, Noida, India, June 2020. leaves infected by anthracnose disease,” IEEE Access, vol. 7,
[3] M. A. Jasim, M. Jamal, and AL-Tuwaijari, “Plant leaf diseases pp. 43721–43729, 2019.
detection and classification using image processing and deep [19] G. Geetharamani and P. Arun, “Identification of plant leaf
learning techniques,” in Proceedings of the 2020 International diseases using a nine-layer deep convolutional neural net-
Conference on Computer Science and Software Engineering work,” Computers & Electrical Engineering, vol. 76,
(CSASE), IEEE, April 2020. pp. 323–338, 2019.
[4] J. Chen, D. Zhang, and Y. A. Nanehkaran, “Identifying plant [20] P. Jiang, Y. Chen, B. Liu, D. He, and C. Liang, “Real-time
diseases using deep transfer learning and enhanced detection of apple leaf diseases using deep learning approach
12 Journal of Food Quality
based on improved convolutional neural networks,” IEEE
Access, vol. 7, pp. 59069–59080, 2019.
[21] A. Kumar and M. Vani, “Image based tomato leaf disease
detection,” in Proceedings of the 2019 10th International
Conference on Computing, Communication and Networking
Technologies (ICCCNT), July 2019.
[22] M. Z. Hasan, “Recognition of jute diseases by leaf image
classification using convolutional neural network,” in Pro-
ceedings of the 2019 10th International Conference on Com-
puting, Communication and Networking Technologies
(ICCCNT), July 2019.
[23] M. Agarwal, “FCNN-LDA: a faster convolution neural net-
work model for leaf disease identification on apple’s leaf
dataset,” in Proceedings of the 2019 12th International Con-
ference on Information & Communication Technology and
System (ICTS), July 2019.
[24] S. Huang, “Development and validation of a deep learning
algorithm for the recognition of plant disease,” in Proceedings
of the IEEE 21st International Conference on High Performance
Computing and Communications; IEEE 17th International
Conference on Smart City; IEEE 5th International Conference
on Data Science and Systems (HPCC/SmartCity/DSS), IEEE,
August 2019.
[25] V. Singh, “Sunflower leaf diseases detection using image
segmentation based on particle swarm optimization,” Arti-
ficial Intelligence in Agriculture, vol. 3, pp. 62–68, 2019.
[26] J. Sujithra and M. F. Ukrit, “A review on crop disease
identification and classification through leaf images,” Euro-
pean Journal of Molecular & Clinical Medicine, pp. 1168–1183,
2020.
[27] J. F. Montecchia, M. I. Fass, I. Cerrudo et al., “On-field
phenotypic evaluation of sunflower populations for broad-
spectrum resistance to Verticillium leaf mottle and wilt,”
Scientific Reports, vol. 11, no. 1, pp. 11644–11714, 2021.
[28] H. Q. Cap, “A deep learning approach for on-site plant leaf
detection,” in Proceedings of the 2018 IEEE 14th International
Colloquium on Signal Processing & its Applications (CSPA),
March 2018.
[29] X. Zhang, Y. Qiao, F. Meng, C. Fan, and M. Zhang, “Iden-
tification of maize leaf diseases using improved deep con-
volutional neural networks,” IEEE Access, vol. 6,
pp. 30370–30377, 2018.
[30] de Luna, G. Robert, E. P. Dadios, and A. A. Bandala, “Au-
tomated image capturing system for deep learning-based
tomato plant leaf disease detection and recognition,” in
Proceedings of the TENCON 2018-2018 IEEE Region 10
Conference, October 2018.
[31] F. Ajaz, M. Naseem, S. Sharma, M. Shabaz, and G. Dhiman,
“COVID-19: challenges and its technological solutions using
IoT,” Current Medical Imaging Formerly Current Medical
Imaging Reviews, vol. 18, no. 2, pp. 113–123, 2022.
[32] Z. Yan, Y. Yu, andM. Shabaz, “Optimization research on deep
learning and temporal segmentation algorithm of video shot
in basketball games,” Computational Intelligence and Neu-
roscience, vol. 2021, Article ID 4674140, 10 pages, 2021.