Special issues on Computer Applications

Identification Of Plant Disease Using Internet Of

Things For Smart Farming

S. Parkavi, Student, S. Sherly, Student, D. B. Ramya Sree, Student,

Dept. Of CSE Dept. Of CSE Dept. Of CSE
Panimalar Institute Of Technology Panimalar Institute Of Technology Panimalar Institute Of Technology
Chennai, India Chennai, India Chennai, India

Dr.T. Kalaichelvi, Professor,

Dept.Of CSE
Panimalar Institute Of Technology
Chennai, India

Abstract— Plant disease detection is the essential thing to

anticipate the reduction in the yield and the proportion of Identification of plant diseases cannot only maximize yield
agricultural products. The study of the plant diseases means the production but also can be supportive of various types of
studies of visually seen markings on the plants. Observing the agricultural methods. India encompasses a big selection of
plant and disease detection of the plant is very risky in the agricultural lands, two-third of the population be able to trust
agricultural field. Hand-operated, it is troublesome for agriculture for their livelihood. It is the basic foundation of
identifying the plant’s disease. Rather, it requires a magnificent the economic development of the country. Agriculture
amount of work, expertise in the plant diseases, and also it needs
provides employment opportunities to the people. An
excessive managing time. An appropriate concern should be
taken in the authentication of the plant, or else it may lead to a important role plays on the salubrity of the plant. When there
productivity decrease in agricultural land. The determination of is a diseased leaf in the plant, the production of farming is
plant defects can be done by image processing. Disease reduced and there is a profit loss. Leaves are necessary for the
identification of the plant involves steps like observing the plant rapid growth of plants and to build up the yield. Appropriate
with its image. In the ancient period, half of the population in monitoring of plant well-being is taken at different stages of
India does agriculture to earn their living. The economy of India plant growth to prevent defected plants. The existence of
is based on agricultural production. Loss of quantity production pests and disease affects the evaluation of crop cultivation and
is due to pests like bacteria, fungus in the plant. As a minimizes crop yield substantially. The present-day system
consequence, the complete supervision of plants is necessary for
depends on naked-eye observation which could be a time-
the production of plants in the agricultural field.
consuming process, it also perceives the demand for
progressing a rapid, profitable, and steady health-monitoring
sensor that would promote advancements in agriculture [10].
Keywords—The Internet of Things (IoT), Plant defect,
Image Processing, Convolutional neural networks(CNN).
Automatic detection of disease is often adopted to detect
disease at the early stages. Many disease management steps
I. INTRODUCTION
are utilized by farmers at proper intervals to lower plant
infections. The existence of defects in the leaf and organic
defects lead to a downturn in the yield. This issue can be
solved using a technically incorporated process of the Internet
of things (IoT) and image processing. Agricultural monitoring
and supervised environment are the functions of the IoT.
Advances of smart farming in agriculture using the IoT
facilitate various processes like loads of information gathered
by sensing element to equip management across the subjective
techniques. Benefits of IoT brings efficient yield, cost
supervision, waste decreases, process computerization, and it
Fig. 1. Smart Farming also upgrades product quality in agriculture. Rigorous in

ISSN: 2393 - 9141 www.internationaljournalssrg.org Page 53

 Special issues on Computer Applications

farming as a whole, rely on image-based detection. it showed good results in research studies. IoT and Image
Implementing these dissections is very helpful for the farmers Processing concepts are used to get the desired outcome of
to obtain a stronger idea in the agricultural field to enhance smart farming. In the hyperspectral proximal sensing method,
their yield. It uses the IoT computing devices, home it is practiced to check the surroundings of the agricultural
appliances, components that are enclosed with electronics, field [2]. For monitoring, plants health optical technologies
network, actuators, and sensors which allows these things to are used. The images at various wavelengths with different
attach and interchange data. To nourish the population of ranges were captured using hyperspectral images [6]. The
India, the production must support IoT technology. The secured outcomes are supportive, reaching a high percentage
requirement for more food has to be reached against of veracity, which outperforms trivial figures, and they can be
challenges such as intense weather conditions and exhaustive practiced as standard tools for peasants to safeguard the
farming practices. Smart farming relating to the IoT tomato plant, from the defects that are caused in the
technologies enhances crop production in the farming agricultural lands [5].
industry.
III. METHODOLOGY
Detection of defects in the plant is the utmost need for farmers
and agricultural exports. There are several variations in image
segmentation and perception system [4]. Separation of images
is done to find the disease accurately. In segregation, it can be
segregated by separating the normal and tainted zone of the
plant by identifying the shade of the leaf. The primary
objective of the detection is to detect plant diseases using the
Internet Of Things. Most of the plant defects take place on a
plant leaf. The difference between diseased and infectious
Fig. 3. Diseased Leaf Image
plants can be measured based on variation in temperature,
humidity, and color. In the upcoming years, systems
incorporated with the IoT and image processing could The IoT, is the framework that incorporates sensing devices
effectively succeed the demand for individuals for observing and cameras to secure the picture of the unhealthy leaf. It
crops [1]. might be separated as an 80-20 proportion for formulating and
resolving the picture for identification. The texture, shade,
pattern, and leaf size are the elements handled for
II. RELATED WORKS examination. Firstly, The procurement of Image, during this
conceptual technique, it expresses the input. The specimen is
recovered by utilizing a camera. The leaf appearance is
extricated and it is collected inside the database for an
additional procedure. Secondly, Image preprocessing, during
this approach the sound is separated, data normalization takes
place. Thirdly, Specimen evaluation, here the segregation of
image is being processed to detect the plant disease. In
segregation, the healthy and unhealthy zone is separated by
the shade of the leaf.

Fig. 2. Detecting Sensor

IV. CONVOULUTIONAL NEURAL NETWORKS

Plant detection of the diseased area is an integral topic, which

has been examined throughout a prolonged period of time and
it is propelled by the demand to generate the well-being of
crops in the farming land. For analysis and disease
identification various conventional machine learning
procedures have been used in the past years. Several leaf
detection, classification of neural network algorithms, and
image processing are used recently. Convolutional neural
networks, are used to recognize shrub defects supervision and

ISSN: 2393 - 9141 www.internationaljournalssrg.org Page 54

 Special issues on Computer Applications

Fig. 4. Structure Of Convolutional Neural Networks with different detectors like fertilizer, a buzzer for enemies,
and to monitor the condition of soil for a specific kind of crop
growth. However, a few technical and business issues are
required to be forwarded in smart farming for the IoT- based
Deep convolutional neural networks are related in numerous productions and functions for the manufacturer. In upcoming
distinct dominion, as illustrations of end to end studies. It years, the offered system will generate a way for the farmers
contributes a mapping between input and output for the in the agricultural lands for a good yield. The forthcoming
detection of the plant disease [9]. It is used to extricate wider works will be concentrated on upgrading the present
information from the specimen. The construction of the outcomes, and an auspicious operation will be broadened.
convolutional neural networks comprises a convolutional, a
pooling, a reshaping, a dense, and dense output layers. The
dense output layer, which accomplishes the desired label for
REFERENCES
the input information [7]. Different CNN architectures are
accessible regarding the detection of the plant disease are
AlexNet, GoogLeNet, VGGNet. With the development of [1] Tichkule. S.K, Gawali. P.D.H, 2016, Plant disease
detection using image processing techniques.Green
CNN, the architectures are emerged [8]. The system of CNN Engineering and Technologies.
incorporates an expansion stage where the feedforward and [2] Meroni. M, Rosini. M, Picchi. V, Panigada. C, Cogliati.
backpropagation will be functioned in the reduction of the S, Nali. C, Colombo. R, Asse Assessing Steady-state
misconception. Observation of the convolutional neural Fluorescence and PRI from Hyperspectral Proximal
network on its precision can be used in the large scale image Sensing as Early Indicators of Plant Stress: The Case of
Ozone Exposure, Sensors 2008.
recognizing process [3]. [3] K.Simonyan and A. Zisserman, “Very deep convolutional
networks for large-scale image recognition,” in 2015.
V. LIMITATIONS [4] P. L. Cheong and S. D. Morgera, “Iterative methods for
restoring noisy images,” Institute of Electrical and
Electronics Engineers Transactions on Acoustics, Speech
To detect the infection of the leaf, Image Processing is and Signal Processing.
applied. Sunlight is the principle factor that influences the [5] M. Brahimi, K. Boukhalfa, and A. Moussaoui, “Deep
learning for tomato diseases: classification and symptoms
outcome. In the night, the pictures cannot be caught. In the visualization,” Applied Artificial Intelligence, in, 2017.
daytime, daylight is too sharp to capture the image. There is [6] Nagasubramanian K, Jones S, Singh A.K, Singh A,
another condition, when the light gets reflected on the leaf, the Ganapathysubramanian B, Sarkar S. Explaining
color or the diseased region of the leaf is not recognized by the hyperspectral imaging-based plant disease identification:
camera, at that point results may vary. Another primary 3D CNN and saliency maps.
concern is the power line network to the server, if there is no [7] Paoletti M, Haut J, Plaza J, Plaza A, “ A new deep
convolutional neural network for fast hyperspectral image
electricity, then the entire system will not work. classification”, Isprs J. Photogramm. Remote Sens. 2018.
[8] A. Krizhevsky, I. Sutskever, and G. E. Hinton, “Imagenet
classification with deep convolutional neural networks,”
VI. CONCLUSION in Proceedings of the 26th Annual Conference on Neural
Information Processing Systems (NIPS '12), Lake Tahoe,
Nev, USA, December 2012.
In this work, the robust deep learning-based detector is [9] Sharada P. Mohanty, David P. Hughes and Marcel
examined to observe the defected plant. The image processing Salathe, “Using Deep Learning for Image-Based Plant
and IoT introduce a hands-on and suitable solution for Disease Detection”, Front. Plant Sci., 2016.
observing the defected region of the plant, which in case [10] Cristina Davis, Sindhuja Sankaran, Ashish Ratna Mishra
illustrates a major comparable difference. For the “A review of advanced techniques for detecting plant
diseases”, Computers and electronics in agriculture, June
forthcoming advancements, it can be advanced in an 2010.
enormous system. Likewise, the process can be coordinated

ISSN: 2393 - 9141 www.internationaljournalssrg.org Page 55