Content uploaded by Vivek Sharma
Author content
All content in this area was uploaded by Vivek Sharma on Oct 10, 2022
Content may be subject to copyright.
Technological Advancements in Automated Crop
Pest and Disease Detection: A Review & Ongoing
Research
Vivek Sharma
Dept. of CSE
MNIT
Jaipur, India
2020rcp9048@mnit.ac.in
Ashish Kumar Tripathi*
Dept. of CSE
MNIT
Jaipur, India
ashish.cse@mnit.ac.in
Himanshu Mittal
Dept. of CSE
JIIT NOIDA
Delhi, India
himanshu.mittal@jiit.ac.in
Abstract—Automated crop pests and disease detection have
fateful effects on food safety, leading to significant deterioration
in agriculture products. The effects of crop diseases and pests can
be so severe that a harvest may even be ruined entirely. Therefore,
automatic recognition and diagnosis of crop disease is required
in the agricultural field. However, Fast and accurate crop disease
detection is still a challenging and error-prone task. Earlier,
traditional methods were used to detect abnormalities in crops
caused by fungus, pests and nutritional deficiency. Moreover, in
some cases, it is time-consuming, expensive and impractical. To
overcome these issues, experimental research is being performed
into the use of image processing techniques for crop disease
detection using machine learning, artificial intelligence, deep
learning, generative adversarial networks and the internet of
things. In this study, a comprehensive literature review of current
studies is performed in crop disease and pest recognition using
image processing to extract the features and algorithms used in
prediction studies. In particular, several models have reported
better accuracy on specific data sets. In contrast, in the case
of different data sets or field conditions, the performance of
the models degraded significantly. Despite this, progress has
been encouraging so far. Furthermore, different inputs gained
from the literature indicate that the aforementioned techniques
provide better accuracy in comparison with existing techniques.
Additionally, a detailed study has been performed on several
unresolved challenges to develop a framework for automated
crop pests and disease detection to use in real field conditions.
Index Terms—Machine learning, Deep learning, Generative
adversarial networks, and Internet of things.
I. INT ROD UC TI ON
In the global economy, agriculture plays a important role in
feeding the human and live-stock populations. The agriculture
state in-country is directly related to the quantity and quality
of products, especially crops. According to the Ministry of
Agriculture Farmers Welfare annual report 2020-21, more
than 57.8 % of the total population is directly or indirectly
depends on agriculture in India. Moreover, numerous factors
are responsible for crop production loss, such as pests, weeds,
and disease dysfunctions, are the major prime factors in India
for 20-25 per cent loss of the whole crop production [1]. Over
the past two decades, the growth rate in the agriculture field
is declining tremendously. As a result of this trend, several
challenges such as weather forecasting, rural-urban migration,
population growth, leads to the significant threat to world food
security.
Due to the fact that there will be a slight increase in the
agricultural land in the future, so there is need to improve the
productivity of existing farmland that will be the key in solving
food insecurity. To meet this, it demands faster maturing,
livestock breeding, Planting high-yielding crops and Crop
varieties resistant to drought and disease. Further, considering
a rapid decline in agricultural employment between 2017
and 2030 owing to greater employment in other fields of
the economy, introducing technology in agriculture domain
becomes of paramount importance [2]. Furthermore, with the
advancement of new technologies, the use of chemical pesti-
cides has increased a lot, which has shown a negative effect
on people’s health and polluted the environment. Therefore,
people are moving towards the use of crops grown organically.
However, the government is placing most stricter regulations
to ban the products grown with the usage of chemicals.
Weather forecasting has been an important factor that seeks
the relationship between the live-stock, humans, and growth
of several diseases in crops. Change in climate pattern leads to
occurrence of diseases and pest. Many researches have pointed
out that disease occur mainly when the crops grown for
the first time [3]. Moreover, in certain situations agricultural
experts are not trained enough to deal with new disease and
pest fails to provide support to the farmers.
To address these issues, precision agriculture is gaining
attention among the farmers in order to increase crop pro-
ductivity in a sustainable manner. Precision agriculture is the
technology enabled techniques that covers modern techniques
and decision support system for the proper management of
farms in more controlled and accurate way [4]. Moreover, tech-
niques such as robotics, drones, terrestrial vehicles, devices for
variable spraying of pesticides, data analytics, and navigation
system in tractor using Global Positioning System helps in
increasing the productivity. Furthermore, the application of
image processing in fusion with the current technologies such
as artificial intelligence (AI), machine learning (ML), deep
978-1-6654-6883-1/22/$31.00 ©2022 IEEE.
2022 International Conference on Computing, Communication, Security and Intelligent Systems (IC3SIS) | 978-1-6654-6883-1/22/$31.00 ©2022 IEEE | DOI: 10.1109/IC3SIS54991.2022.9885605
Authorized licensed use limited to: MALAVIYA NATIONAL INSTITUTE OF TECHNOLOGY. Downloaded on October 10,2022 at 11:38:02 UTC from IEEE Xplore. Restrictions apply.
learning (DL), generative adverserial networks (GAN), and
internet-of-things (IOT) in crop disease detection and pest
recognition has shown the better results. Furthermore the
fusion of technologies has become the hot spot research area
for the researchers to address the issues of accurate and early
detection of diseases and pests [5]. A number of possible
forthcoming revolutions in agriculture technologies such as
digital twins [6], federated learning [7], block chain [8], fog
computing [9], edge computing [10], and robotics [11] here
the accuracy has been reached to perception of human level.
Current research efforts are aimed for the attainment of such
high levels of accuracy in crop disease detection and pest
recognition.
In image processing techniques, computer programs are
used to manipulate and analyze images recorded with a
different types of sensors including infrared imaging devices,
visible light cameras, and detectors with frequency bands.
Moreover, various research has been done in the field of crop
disease and pest recognition in hyper-spectral models [12].
However, these hyper-spectral imaging instruments are very
costly and not easily reachable to extension workers and or-
dinary farmers. Therefore, in the present scenario these image
processing techniques with the fusion of artificial intelligence
technologies employed in the crop disease detection and pest
recognition. Figure 1 depicts the technological support in leaf
pest and disease detection.
1) Image processing techniques can be used to for fast and
accurate predictions in recognizing crop pest and disease
recognition based on stem, flowers, fruits, and images of
leafs.
2) The severity of disease can be determined by measuring
the size of the discolored or deformed area in relation
to the whole fruit, leaf, and flower.
3) Tracking the details about disease progression in plants
that could be difficult to discover by human observers,
such as in identifying the symptoms and the stage of
infection.
4) This also helps the researchers and scientists to under
go the investigation in the lab to find the characteristics
of sowing a new crop cultivation.
5) The information obtained through image processing
techniques can be distributed immediately and inexpen-
sively to people at remote locations.
6) An accurate diagnosis will yield a more efficient use of
pesticides. This will decrease production costs.
7) Extension officers can consult human experts remotely
rather than traveling to individual farms.
II. RE LATE D WO RK
A. Leaf disease recognition using deep learning techniques
Over the decade, convolutional neural networks have de-
termined phenomenal performance as feature classifiers and
extractors in the field of image processing. Furthermore, this
concept has been successfully applied in several areas, it has
recently entered into the field of agriculture to accomplish
Fig. 1. Technological support in leaf pest and disease recognition
Fig. 2. Feature extraction and classification workflow
different tasks such as pest recognition and crop disease
detection, flower and fruit counting, weed detection, and fruit
grading. Particularly, DL has been widely used since 2015
for recognizing leaf diseases detection using image processing
techniques. DL representation method that seeks to find the op-
timal way to represent data optimization techniques instead of
semantic features. As a result of the learning process, features
are extracted automatically rather than manually. Furthermore,
DL constitutes the modern techniques in agriculture that will
be useful for the food industry to grow in robotics, big data,
pest detection, disease diagnosis, marketing and automation.
Figure 2 highlights the overall workflow incorporating the
feature extraction and classification workflow.
For training CNN, large datasets consisting of thousands
of images are required. However, crop disease detection,
such a diverse and massive datasets have not yet mobilized
and availed for the use by different researchers. In order to
create a better CNN classifiers for detecting plant diseases,
transfer learning is currently the most efficient method used
for experimentation. Moreover, Transfer learning is the process
of transforming pretrained CNNs using smaller datasets with
Authorized licensed use limited to: MALAVIYA NATIONAL INSTITUTE OF TECHNOLOGY. Downloaded on October 10,2022 at 11:38:02 UTC from IEEE Xplore. Restrictions apply.
a different distribution than what large datasets earlier used
to build the CNNs from scratch. Moreover, Transfer learning
is the process of transforming pretrained CNNs using smaller
datasets with a different distribution than what large datasets
earlier that are used to build the CNNs . Studies demonstrated
that use of pretrained models CNN on dataset dataset provides
better result for crop disease detection. [13].
Kawasaki et al. [14] proposed a novel CNN architecture for
disease detection in cucumber leaf such as Zucchini yellow
mosaic and yellow spot viruses infection. The study shows
that data augmentation has more contribution in recognising
the performance than enlarging the several training epochs.
Furthermore, the researchers addressed different architectures
to improve the classification using several augmentation strate-
gies.
Durmusßet al. [15] demonstrated the work done in
proposing a extended work by training AlexNet, CNNs on
SqueezeNet using the platform nvidia jetson. However, there
is a slight decrement in the accuracy achieved by Alexnet
in comparison to the workdone in where TITAN X GPU
is used. Furthermore, the results indicate that crop disease
detection algorithms could be experimented to run in real
time. However, this study shows that in embedded applications
in order to attain high performance initially train the model
on a traditional GPU and then deploy the training model
to embedded system. The study conducted by Brahimi et
al. [13]the comparison of googlenet and alexnet CNNs from
the starting point and for the detection of nine tomato leaf
disease using transfer learning. Furthermore, two classifiers
are used such as SVM and RF in training the network.
The study demonstrate that pre-trained CNNs shows bettter
performance than the CNNs from the scratch and CNN models
in comparison to SVM and RF classifiers performed better.
Furthermore these networks are very critical in image disease
detection and knowing this is essential to minimize CNN
models. These also highlighted that combination of different
CNN classifiers produced higher recognition accuracy and this
will be helpful to detect symptoms of disease from high-
resolution images.
The studies of different authors [16] presented the alterna-
tive approach was taken by applying DL in object detectors.
Furthermore, Single Shot Multi-box detector and faster region-
based CNNs architectures are used to classify and find the
region containing diseases which are based upon the features
inside the bounding box. In order to use these detectors
with CNNs, different architectures were investigated such
as ResNetXt-101, ResNet-101, AlexNet, GoogLeNet, ZFNet,
VGG-16. Barbedo et al. [17] highlighted that in order to
classify leaf diseases using DL, image of individual spots and
lesions is require rather than entire leaves. However, there
are still some issues are not resolved is the problems with
exact automatic background removal and segmentation of the
images into individual lesions. Further, the compact deep CNN
system used to implemented detect plant diseases on mobile
phones. Furthermore, CNN models can be deployed on mobile
devices, which will make this technology more accessible and
Fig. 3. Geneative Adverserial Network
beneficial to farmers.
B. Leaf disease recognition using generative adversarial net-
work
In the past half decade generative adversarial net-
works(GANs) has been introduced in the field of generating
synthetic images. Furthermore, CNNs has been widely used
in the leaf disease detection and pest recognition. However,
CNNs have been applied in this fields and proven to be
a effective approach, despite this the major issue has been
overlooked that is of limited training dataset. This has led
to the problem of over fitting of data. Moreover, with the
evolution of GANs based methods prediction accuracy has
been increased and resolved the problem of over fitting of
data when training data is limited. In [18] Goodfellow et al.
GANs are mainly used to address the data scarcity problem.
The structure of GANs consists of two networks such as
discriminator and a generator. In this the training data dis-
tribution is captured by the generator whereas a discriminator
determines the probability that whether an image came from
the generator or from training data. Moreover,the goal is to
increase the capacity of the generator to fool the discriminator,
which is trained to distinguish real images from synthetic ones.
Using this method, a dedicated GANs is required which will
generate synthetic images, and that is used to train leaf disease
and pest identification system. Figure 3 depicts the different
types of generative adversarial networks applied in varieties
of agricultural applications such as crop disease, fruit, and
vegetables.
Arsenovic et al. [19] addressed the generation of crop
images artificially using Generative Adversarial Networks
(GANs). Furthermore several variants of GAN architectures
as been proposed in past few years such as CGAN, DCGAN,
ProGAN, and StyleGAN. The conditional GAN (CGAN), is
used to represent multimodal data generation in a better way.
In our study, StyleGAN generated leaf images with the best
results at the 256 ×256 image pixel resolution. These GAN
networks were not successful in training on field images due
to the busy background, a problem that remains unresolved
Authorized licensed use limited to: MALAVIYA NATIONAL INSTITUTE OF TECHNOLOGY. Downloaded on October 10,2022 at 11:38:02 UTC from IEEE Xplore. Restrictions apply.
even after more training on field images. However,on the
test set networks trained on GAN created images achieved
about 1 percent better accuracy than networks trained only on
natural images. Furthermore, a new two-stage CNN known
as Plant DiseaseNet (PDNet) was also introduced [19]. Also
stage one (PDNet-1) used a detector YOLOv3 with feature
extractor Alexnet for predicting bounding box in the leaf of
crops. Moreover, a new plant diseasenet known as the stage
two (PDNet-2) consist of softmax layer, a CNN architecture
of 32 layers, pooling layer for averaging globally, and 42-way
layer fully connected. The map score attained by PDNET-1
is 0.9165 while in case of PDNet-2 attained is 93.67 percent
crop disease recognition accuracy. It should be noted that, the
GANs holds a lot of potential that can be used to generate
the training images artificially and very useful to resolve the
problem of scarcity of data.
C. Leaf disease recognition using Machine Learning
Machine learning is one of the most prominent approach for
the crop pest and disease recognition. Crop disease recognition
is one the challenging problem in agricultural field. Further-
more, several models have been addressed and validated so
far. From the studies it has been concluded that crop disease
recognition is not a trivial task, instead it has to pass through
the several complicated steps. In the present scenario, crop
disease recognition models can recognise the actual disease in
the crops, but a better prediction is still desirable. ML is a
branch of AI and computer Vision which focuses on learning,
just like the way humans learn. ML is useful in determining
the correlation and pattern and knowledge discovery from
the data sets. Furthermore, several models are trained on the
datasets, and output is based upon the past experience. In order
to built predictive models several features are required, such
as determining the parameters of the model using historical
data in the training phase. While in case of testing phase,
for the performance evaluation the historical data can be used
which has not been used in training phase. A ML models
can be predictive or descriptive, it depends upon the research
problems and questions. Furthermore, in order to make the
future predictions predictive models whereas to extract some
information from the sampled data collected and explain what
has happened descriptive models are used. Further,in the study,
it has been found that in order to built high performance
predictive models in machine learning different challenges
need to be faced. Furthermore, it is very difficult to select
the right model to solve the challenging problem at hand, and
to handle the large volume of data in the underlying platforms.
As ML now became an integral part for the precision agri-
culture, where crop pest and disease recognition can be done.
This also presented a tool for discrimination and detection
of healthy plants and smut fungus during the crop growth.
Ebrahimi et el. presented a new technique for the automatic
pest detection using SVM classification [20]. Further, they
also proposed a novel method for screening and detection of
disease bakanae in rice seedlings. Furthermore, identification
of crop disease and pest recognition using image processing
in fusion with machine learning algorithms have been applied
in the sample crops such as in fruit leaves, brinjal leaf, potato,
peanut leaf and groundnut leaf.
D. Leaf disease recognition using internet of things
With the advancement of technology in last one decade is
witnessing the new era of computing that is shifting from
traditional approaches to the intelligent ones. The internet
of thing (IOT) constitutes a recent modern technique use
in different applications such as health , agriculture, and
in business and many more [21] with large potential and
promising results. With this the domain the field of agriculture
is getting immensely fortified. The IOT has been introduced
in agriculture sector to improve the quality and quantity in
this field. This can only be achieved through the early disease
recognition before the harvesting of crops. With the emergence
of IoT, major controlling resources have been added in the
area of disease detection in plant phenology, assisting in
preventing disease outbreaks. Pandiyan et al. [22], developed a
novel platform for analysing data from heterogeneous IoT with
advanced segment extraction. Different levels such as, service
level, platform level, and connectivity level were used to
perform the several tasks namely data aggregation, automatic
identity identification and transmission. The study has been
on leaf gestures for the better identification of diseased leaves
done. Zhao et al. [23] proposed a novel learning system
that can automatically recognize the crop diseases form the
cluttered background.
Kale et al. [24] presented support system in decision making
in smart farming with the help of smart fertilizers. Moreover,
lack of judgment lead to the inappropriate decision. In this
study, genetic algorithm and IOT were used in system design.
Further, in designing the system for plant disease detection an
improved genetic algorithm with extreme learning machines
classifiers along with IOT was proposed.
Pawara et al. [25], addressed the different disease such as
spot in fruit & leaf spot, bacterial blight in pomegranate.
Furthermore, a sensor based and hidden markov model model
has been introduced in model system design with the several
parameters were taken into the consideration such as leaf wet-
ness, air humidity, and soil wetness. Sharma, V. and Tripathi,
A.K. [26] highlighted the systematic survey of meta-heuristic
algorithms in IoT based application. Moreover, presented a
real time monitoring system for the collection of environ-
mental data with cloud storage and IOT for crop disease
identification and detection. For classifying the environmental
data support vector machine regression was used. Further,
demonstrated a prototype for smart farming capable of disease
detection and diagnosis in plant using web enabled system
that is IOT based. In this study, septoria plant disease was
investigated for experimentation. The overall summary of the
technologies, type of crop, disease identification, and algo-
rithm are described in Table I. Here, different abbreviation’s
namely DL stands for DL, ML stands for machine learning,
IOT stands for internet of technology.
Authorized licensed use limited to: MALAVIYA NATIONAL INSTITUTE OF TECHNOLOGY. Downloaded on October 10,2022 at 11:38:02 UTC from IEEE Xplore. Restrictions apply.
TABLE I
CLASSIFICATION OF ALGORITHMS BASED ON APPLICATIONS
Technologies Type of crop Disease detected Algorithm Ref
M L Apple Early blight, Apple rust Radial bias function neaural netwrok (RBFN) [27]
M L Sugrcane Red rot, leaf spot, mosaic virus Support Vector Machine (SVM) [28]
M L Tulsi Maturity of leaves in tulsi Multi layer perceptron [29]
M L Vine Powdery mildew One Class Classifier [30]
D L Tomato Early blight Convolutional neural network based transfer learn-
ing models such as resnet
[31]
D L Potato Late blight , Early blight Deep Convolutional neural network (DCNN) [32]
D L Rice sheath rot , bacterail blight Deep neural network (DNN) with jaya optimized
algorithm
[16]
IoT Banana sigatoka disease GLCM and RFC [33]
IOT Pomegranate Fruit spot, bacterail blight HMM and sensor based model [25]
IOT Rice bacterail blight Drone based IOT architecture [34]
III. FUTURE RESEARCH DIRECTIONS
Aforementioned technologies is anticipated to increase the
potential of agricultural industry to feed the forthcoming
generations. Integration of different technologies is the urge
of the modern practices in agriculture field. A number of
possible forthcoming revolutions in agriculture technologies
such as digital twins [6], federated learning [7], block chain
[8], fog computing [9], edge computing [10], and robotics
[11]. These technologies play an important role for the futur-
istic adaptations in the several diversified dimensions namely
harvesting of crops, fruit counting, weed detection & man-
agement, forestry, livestock farming, fertilization & irrigation,
scouting of crops, and pest surveillance. In the recent times, a
majority of work has been done in the crop disease and pest
identification using image identification. However, few work
has been done using the satellite images thought lack of dataset
is still a major constraint in this field. Historical dataset can be
a game changes in the effective estimation the growth of crops,
fertility status, and salinity issues. These challenges can be
easily managed using the aforementioned techniques. Digital
twin technology can be a new revolutionized field which can
efficiently helps in the 3D view with different shapes of plant
growth. However, dataset is still a major challenge to perform
the experimental work using these techniques. Another one is
the federated learning is a new approach in machine learning
field which works in a decentralized manner, hence reduces
the load of the different severs when large amount of data
is trained. However, efficient GPUs are required to perform
the experimental work which are widely not available. These
technologies can trigger the different nascent perspectives
which can be in the future work for precision agriculture.
IV. DISCUSSION AND CONCLUSION
This documents demonstrates the comprehensive review
of research efforts done in agricultural domain using image
processing techniques. In the ongoing research a survey have
been performed for the crop pest and disease identification
system using DL models, GAN, ML, and IOT. Furthermore,
the accurate diagnosis depends on the type of technology used,
hence fusion of multi modals can led to the development of
such system. Lately, these technologies are capable enough
to use in several areas such as, health, sentiment analysis and
object detection. In the agricultural field, different technologies
have been addressed such as land monitoring, crop yield pre-
diction, disease detection, and crop identification. Moreover,
several fusions have been employed in the agricultural field
to improve the early crop disease detection. Many of these
fusion modals have shown the promising results with the
high potential for the early prediction of pest and disease.
Our findings demonstrates that these technologies provides
better performance than other traditional techniques.Further
in existing literature some research gaps are highlighted for
future work in this area. In the current time, research efforts
have been made to overcome the problem of limited data set
using generative adversarial networks. Further efforts should
also made to develop the models for backdrop removal and
assimilate distinct types of data namely disease incidence
history, topographical location, and weather forecasting data
in order to increase reliability and efficiency in the disease
detection systems. However, identification of syndrome in
disease that occurs on the different part of plants such as
stem has not been highlighted by the researchers. The overall
benefits of using these technologies is to apply these in the
field of agriculture, to encourage the researchers to the move
towards the advancement of sustainable farming for increasing
the food production.
REFERENCES
[1] T. Deshpande, State of agriculture in india, PRS Legislative Research
53 (8) (2017) 6–7.
[2] C. Schrder, Employment in european agriculture: Labour costs, flexibil-
ity and contractual aspects (2014).
[3] A. Johannes, A. Picon, A. Alvarez-Gila, J. Echazarra, S. Rodriguez-
Vaamonde, A. D. Navajas, A. Ortiz-Barredo, Automatic plant disease
diagnosis using mobile capture devices, applied on a wheat use case,
Computers and electronics in agriculture 138 (2017) 200–209.
[4] R. Schmaltz, What is precision agriculture, Agfundernews,
https://agfundernews. com/what-is-precision-agriculture. html (2017).
[5] L. C. Ngugi, M. Abelwahab, M. Abo-Zahhad, Recent advances in image
processing techniques for automated leaf pest and disease recognition–a
review, Information processing in agriculture 8 (1) (2021) 27–51.
[6] W. Li, D. Zhu, Q. Wang, A single view leaf reconstruction method based
on the fusion of resnet and differentiable render in plant growth digital
twin system, Computers and Electronics in Agriculture 193 (2022)
106712.
[7] A. Durrant, M. Markovic, D. Matthews, D. May, J. Enright, G. Leontidis,
The role of cross-silo federated learning in facilitating data sharing in the
agri-food sector, Computers and Electronics in Agriculture 193 (2022)
106648.
Authorized licensed use limited to: MALAVIYA NATIONAL INSTITUTE OF TECHNOLOGY. Downloaded on October 10,2022 at 11:38:02 UTC from IEEE Xplore. Restrictions apply.
[8] V. Hassija, S. Batra, V. Chamola, T. Anand, P. Goyal, N. Goyal,
M. Guizani, A blockchain and deep neural networks-based secure
framework for enhanced crop protection, Ad Hoc Networks 119 (2021)
102537.
[9] F. M. R. Junior, R. A. Bianchi, R. C. Prati, K. Kolehmainen, J.-P.
Soininen, C. A. Kamienski, Data reduction based on machine learning
algorithms for fog computing in iot smart agriculture, Biosystems
Engineering (2022).
[10] S. S. Gill, A manifesto for modern fog and edge computing: Vision,
new paradigms, opportunities, and future directions, in: Operationalizing
Multi-Cloud Environments, Springer, 2022, pp. 237–253.
[11] D. T. Fasiolo, L. Scalera, E. Maset, A. Gasparetto, Recent trends in
mobile robotics for 3d mapping in agriculture, Advances in Service and
Industrial Robotics: RAAD 2022 120 (2022) 428.
[12] T. Chen, J. Zhang, Y. Chen, S. Wan, L. Zhang, Detection of peanut leaf
spots disease using canopy hyperspectral reflectance, Computers and
electronics in agriculture 156 (2019) 677–683.
[13] M. Brahimi, K. Boukhalfa, A. Moussaoui, Deep learning for tomato
diseases: classification and symptoms visualization, Applied Artificial
Intelligence 31 (4) (2017) 299–315.
[14] Y. Kawasaki, H. Uga, S. Kagiwada, H. Iyatomi, Basic study of au-
tomated diagnosis of viral plant diseases using convolutional neural
networks, in: International symposium on visual computing, Springer,
2015, pp. 638–645.
[15] H. Durmus¸, E. O. G¨
unes¸, M. Kırcı, Disease detection on the leaves
of the tomato plants by using deep learning, in: 2017 6th International
Conference on Agro-Geoinformatics, IEEE, 2017, pp. 1–5.
[16] P. Jiang, Y. Chen, B. Liu, D. He, C. Liang, Real-time detection of
apple leaf diseases using deep learning approach based on improved
convolutional neural networks, IEEE Access 7 (2019) 59069–59080.
[17] J. G. A. Barbedo, Plant disease identification from individual lesions and
spots using deep learning, Biosystems Engineering 180 (2019) 96–107.
[18] I. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley,
S. Ozair, A. Courville, Y. Bengio, Generative adversarial nets, Advances
in neural information processing systems 27 (2014).
[19] M. Arsenovic, M. Karanovic, S. Sladojevic, A. Anderla, D. Stefanovic,
Solving current limitations of deep learning based approaches for plant
disease detection, Symmetry 11 (7) (2019) 939.
[20] M. Ebrahimi, M.-H. Khoshtaghaza, S. Minaei, B. Jamshidi, Vision-
based pest detection based on svm classification method, Computers
and Electronics in Agriculture 137 (2017) 52–58.
[21] A. K. Tripathi, K. Sharma, M. Bala, A. Kumar, V. G. Menon, A. K.
Bashir, A parallel military-dog-based algorithm for clustering big data
in cognitive industrial internet of things, IEEE Transactions on Industrial
Informatics 17 (3) (2020) 2134–2142.
[22] S. Pandiyan, M. Ashwin, R. Manikandan, K. R. KM, A. R. GR, Hetero-
geneous internet of things organization predictive analysis platform for
apple leaf diseases recognition, Computer Communications 154 (2020)
99–110.
[23] Y. Zhao, L. Liu, C. Xie, R. Wang, F. Wang, Y. Bu, S. Zhang, An effective
automatic system deployed in agricultural internet of things using multi-
context fusion network towards crop disease recognition in the wild,
Applied Soft Computing 89 (2020) 106128.
[24] A. P. Kale, S. P. Sonavane, Iot based smart farming: Feature subset
selection for optimized high-dimensional data using improved ga based
approach for elm, Computers and Electronics in Agriculture 161 (2019)
225–232.
[25] S. Pawara, D. Nawale, K. Patil, R. Mahajan, Early detection of
pomegranate disease using machine learning and internet of things,
in: 2018 3rd International Conference for Convergence in Technology
(I2CT), IEEE, 2018, pp. 1–4.
[26] V. Sharma, A. K. Tripathi, A systematic review of meta-heuristic
algorithms in iot based application, Array (2022) 100164.
[27] S. S. Chouhan, A. Kaul, U. P. Singh, S. Jain, Bacterial foraging
optimization based radial basis function neural network (brbfnn) for
identification and classification of plant leaf diseases: An automatic
approach towards plant pathology, IEEE Access 6 (2018) 8852–8863.
[28] K. Renugambal, B. Senthilraja, Application of image processing tech-
niques in plant disease recognition, International Journal of Engineering
Research & Technology 4 (3) (2015) 919–923.
[29] G. Mukherjee, A. Chatterjee, B. Tudu, Morphological feature based
maturity level identification of kalmegh and tulsi leaves, in: 2017 Third
International Conference on Research in Computational Intelligence and
Communication Networks (ICRCICN), IEEE, 2017, pp. 1–5.
[30] X. E. Pantazi, D. Moshou, A. A. Tamouridou, Automated leaf disease
detection in different crop species through image features analysis and
one class classifiers, Computers and electronics in agriculture 156 (2019)
96–104.
[31] A. S. Chakravarthy, S. Raman, Early blight identification in tomato
leaves using deep learning, in: 2020 International Conference on Con-
temporary Computing and Applications (IC3A), IEEE, 2020, pp. 154–
158.
[32] M. Al-Amin, D. Z. Karim, T. A. Bushra, Prediction of rice disease from
leaves using deep convolution neural network towards a digital agricul-
tural system, in: 2019 22nd International Conference on Computer and
Information Technology (ICCIT), IEEE, 2019, pp. 1–5.
[33] R. D. Devi, S. A. Nandhini, R. Hemalatha, S. Radha, Iot enabled efficient
detection and classification of plant diseases for agricultural applications,
in: 2019 International Conference on Wireless Communications Signal
Processing and Networking (WiSPNET), IEEE, 2019, pp. 447–451.
[34] N. Kitpo, M. Inoue, Early rice disease detection and position mapping
system using drone and iot architecture, in: 2018 12th South East Asian
Technical University Consortium (SEATUC), Vol. 1, IEEE, 2018, pp.
1–5.
Authorized licensed use limited to: MALAVIYA NATIONAL INSTITUTE OF TECHNOLOGY. Downloaded on October 10,2022 at 11:38:02 UTC from IEEE Xplore. Restrictions apply.
