Leaf Disease Detection Using Image Processing and ML

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 09 Issue: 05 | May 2022 www.irjet.net p-ISSN: 2395-0072
© 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 1361
Leaf Disease Detection Using Image Processing and ML
Prof.(Mrs) Jyotsna V. Barpute1, Ashwini B. Phadatre2, Gauri S. Honrao3
1 Assistant professor of Computer Engineering, Savitribai Phule Pune University
2,3 Pursuing Bachelor of Computer Engineering, Suman Ramesh Tulsiani
Technical Campus Faculty of Engineering Kamshet, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - In most nations, agriculture is the primary
source of income for farmers, and productivity estimation is a
major difficulty for them. Agriculture, together with its linked
industries, is indisputably India's major source of income,
particularly in the country's vast rural areas. For eons,
agriculture has been cultivated in every country. Agriculture
encompasses both the technology and the art of nurturing
plants. Agriculture performed a majorroleinthe development
of human civilization. Agriculture has been done by hand for
ages. Because the world is moving toward new technologies
and implementations, agriculture must also modernize. The
Internet of Things is essential for smart farming. Sensors
connected to the internet of things can provide all kinds of
information about agricultural landscapes. We can achieve
impressive results by implementing deep learning techniques
such as CNN. CNN models are used to detect disease in plants
via plant leaves, and CNNs have shown to be particularly
effective in Machine Vision. Imageprocessing isatechniquefor
performing actions on images in order to improve it and
extract features from them. It is a kind of signal processing in
which the input is an image and the output is an image or
image features.
Key Words: Image Processing, Convolution Neural
Network
1. INTRODUCTION
The modern disease detection approach is built on the basis
optical inspection by specialists, that provides for the
detection and recognition of plant disorders. To do so, an
enormous team of professionals, as well as ongoing
monitoring by specialists, is required, with prices rising as
farms become larger. Aroundthesametime,insomeregions,
farmers lack proper facilities or the knowledge that they'll
need to seek expertise. As a result, even though consulting
specialists demand a significant fee, they also are time -
consuming process. Plant diseases impair agricultural
production volume and quality. Plant disease compromises
the integrity of the leaf, fruit, stem, vegetable, and its
products. This has a big effect on productivity, which has a
big influence on cost. In recent years, machine learning has
gained popularity as a beneficial technique in agriculture.
The concept is essential when it comes to producing,
monitoring, managing, and enhancing productions. Deep
learning is a type of machine learning that uses multiple
layers to transform input into information. It is also used to
handle a variety of complex issues such as image
classification, pattern analysis, and feature extraction. Deep
learning allows you to identify illnessesusingmanydifferent
feature sets. Of these, the traditional handmade approach
and DL features are the most popular feature sets. For
efficiently extractingfeatures,pre-processingsuchaspicture
enhancement, colors modification, and segmentation is
required. Classifiers are applied after feature extraction.
KNN, SVM, decision tree, RF, ANNs, and Deep CNN are
amongst the most widely known classifiers.
2. Proposed Work Objective
The intention of this proposed methodology is by using a
plant leaves image to predict thetypeofdisease.CNN,is fully
unsupervised methodology, is applied to predict the
outcome. This research aims to applyCNN toverifyplantleaf
disease. The goal of this study is to use image recognition to
detect unhealthy plant leaf areas and categorize plant
diseases. One of the most important uses of image
processing is picture recognition, whichisan importanttool
for early diagnosing in crop production.
3. MOTIVATION OF WORK
Farmers tend to judge diseases simply examining at them
through their naked eyes. However, that's not always the
right strategy. Many times, a farmer should approach
specialists for detection of diseases, which is time-
consuming in farmlands. Several applications have been
developed using digital image processing techniques in
various industries such as industrial inspection, medical
imaging, remote sensing, agricultural processing, and so on.

4. RELATED WORK
Uday Pratap Singh, Siddharth Singh Chouhan –
Multilayer Convolution Neural Network for the
Classification of Mango Leaves Infected by Anthracnose
Disease || March 27, 2019.
Mango trees, specifically the fruits and the leaves
are highly affected bythefungal diseasenamedAnthracnose.
The main goal of this work is to develop appropriate and
effective methods for diagnosing the disease and its
symptoms, and thus advocate the appropriate system.
Dr. T. Vijayakumar, Mr. R. Vinothkanna – Mellowness
Detection of Dragon Fruit Using Deep Learning Strategy
|| April 9, 2021.
They use RESNET 152, a deep learning
convolutional neural network, to identify dragon fruit
maturity, and this proposed system identifies dragon fruit
maturity and harvest time. They trained the model using
Python and TensorFlow. The developed structure was
trained with images of dragon fruit at various stages of
maturity and tested for convergence range and
misclassification using 100 new data points.
Snehal A. Lale, Dr. V. K. Shandilya – A Survey on
ImplementationofDeep DictionaryLearningandCoding
Network for Plant Disease Detection in Agricultural
Field. || May 2021.
Here, the concepts of deep learning and
dictionary learning are combined. The advantages of
dictionarylearninganddeep learningmethodsarecombined
in the DDLCN framework proposed for deep learning. This
dictionary learning and coding layer can fill the
convolutional layer of a standard deeplearningarchitecture.
Muhammad E. H. Chowdhury, Tawsifur Rahmen, Amit
Khandakar – Tomato Leaf Disease DetectionUsingDeep
Learning Technique || April 30, 2021.
The overall methodology of the tomato leaf disease
detection is to classify healthy and unhealthy leaves of
tomato plants the classification is done using pre-trained
networks like- ResNet18, MobilenetV2, InceptionV3, and
DenseNet201.
Ashwini C, Anusha B, Divyashree B. R, Impana V – Plant
Leaf Disease Detection Using Image Processing || 2020.
In the proposed system, they use image processing
to identify the color features of the leaves for detecting
diseases, viruses and also provides prevention to the
particular disease.
5. PROPOSED SYSTEM APPROACH
i. PROBLEM STATEMENT
India is mainly an agricultural country. 70% of the rural
population is still dependent on agriculture and 82% of the
farmers are subsistence farmers. However, leaf infections
cause significant crop loss. Leaf infections can occur due to
environmental conditions such as heavy rains, rapid
temperaturechanges,impropercare,andheavypesticideuse.
Automatic detection of plant diseases is an important
research theme that enables early detection of diseases and
viruses.
ii. SYSTEM ARCHITECTURE
The act of obtaining an image from a source, usually a
physical equipment such as a camera, is known as visual
acquisition. Image preprocessing turns image data in such a
way that Deep Learning algorithms can solve. After that, the
image is split into subgroups to help minimize the image's
complexity and make subsequent operations easier. The
database includes a huge number of sample image files that
include both disease and non-disease photos. The features
will be removed based on Picture feature extraction
techniques recommendations, and the image file will be
stored in database.
Sr.
No.
Title of Paperand
Year
Methodology Findings
1. Multilayer CNN for
the classificationof
Mango leaves
infected by
Anthracnose
Disease.
CNN, Image
Processingand
Plant
Pathology
Comparing results
with the other state
approaches PSO,
SVM, and RBFNN
2. Mellowness
Detection of
dragon fruit using
DL strategy.
Mellowness
detection in
dragon fruit
using deep
convolution
neural
network.
The proposed
method had higher
learning accuracies.
3. A survey on the
implementation of
deep dictionary
learning and
coding networkfor
plant disease
detection in the
agriculture field.
Dictionary
learning,
DDLCN, Haar
Wavelet
algorithm.
The algorithm's
concept is to
compare past data
with upcoming data.
4. Tomato leaf
disease detection
using DLtechnique
Automatic
disease
detection,deep
learning, CNN.
When compared to
other architectures,
DenseNet201 was
found to be better at
extracting features
from images.
5. Plant Disease
Detection using
image processing.
Pre-processing
segmentation,
SVM Classifier
This paper proposed
using a combination
of form, texture, and
gray level to detect
illness in leaves.

Fig.1 System Architecture of Proposed System
6. MATHEMATICAL MODEL
Let ‘S’ be the system.
 Where,
S = I, O, P, Fs, Ss
Where,
 Set of input Set of Output
 P = Set of technical process
 Fs = Set of Failure State
 Ss = Set of Success State
 Identify the input data I1, I2,…… In
I = Input Data
 Identify the Process as P
P = Image Processing, Segmentation, Feature
Extraction, Classification
 Identify the Failure state as Fs
Fs = If data is not clear, not successfully loaded, If
not predicted, or it take more time that it should
 Identify the Success State as Ss
P = Accurate prediction within given time.
7. ALGORITHM
On this Proposed system, the CNN set of rules isemployed.A
Convolution Neural Network is a deep learningtechnology
which makes use of trainable weights and biases to offer
trainable weights and biases to various factors in apicture,
permitting it to characterise amongst them. The variety of
pre-processing achieved via way of means of a ConvNet
ismuch much less than that required via way of means of
different category techniques. A ConvNet`s structure is
motivated via way of means of thevisual cortex's shape and
is comparable to the human brain's connecting community
of neurons. Individual neuronscan best react to stimuli with
in receptive field, that's a constrained region of the visible
field. To span the wholevisual field, some of equal fieldsmay
be layered on pinnacle of 1 another
8. CONCLUSIONS
The use of convolutional neural networks to predict plant
diseases is the focus of this proposed system. This
technology helps reduce the time and cost of manual
forecasting. Based on the results, we can conclude that the
convolutional neural network has a high level of accuracy in
perturbation detection. You can extend this task to create
real-time applications that can recognize multiple plant
species instead of just one. To segment the leaf area, certain
systems use scaling and gauss filtering for image
preprocessing, followed by a CNN classification algorithmto
detect leaf disease.
REFERENCES
[1] Challa Karthik, Nagul Ulaganathan - Application for
Plant’s Leaf Disease Detection using Deep Learning
Techniques || volume: 07 Issue: 08 | Aug 2020
[2] Saraswathi. E, and Usha Kiruthika - Study on stages of
diseases detection in a plant using Deep Convolutional
Neural Network (CNN) in Agriculture || 20 April 2021
[3] Ashwini C, Anusha B, Divyashree B R, Impana V, Nisarga
S P -Plant Disease Detection using Image Processing ||
ISSN: 2278-0181 || 2020
[4] Uday Pratap Singh Siddharth Singh - Multilayer
Convolution Neural Network for the Classification of
Mango Leaves Infected by AnthracnoseDisease||March
27, 2019
[5] Snehal A. Lale, Dr. V. K. Shandilya- A survey on
implementationofdeeplearningdictionarylearning and
coding network for plant disease detection || May 2021
[6] Punam Bedi, PushkarGole-plantdiseasedetectionusing
hybrid model based on convolution autoencoder and
convolution neural network. || 8 May 2021
[7] Sethy, P.K.; Barpanda, N.K; Rath, A.K.; Behra, S.K- Deep
feature-based rice leaf disease identification using
support vector machine || 2020
[8] Kamlesh Golhani a, Siva K. Balasundram – A review of
neural networks in plant disease detection using
hyperspectral data || May 2018
[9] Badge S, Patil S, Patil P- Artificial neural network-based
plant leaf disease detection || 2015
[10] Patil, J.K, & Kumar – Advances in image processing for
detection of plant disease. Journal of advanced
bioinformatics application and research 2 || 2011

[11] Chen, J; Chen, J; Zhao, Zhang – Using deep transfer
learning for image-based plant disease identification ||
2020
[12] Zeng, W.; Li, M – Crop leaf disease recognition based on
self-attention convolutional neural network || 2020
[13] Asifullah Khan, Anabia Sohail,UmmeZahoora – Asurvey
of the recent architectures of deep convolutional neural
networks.
[14] Ahmed k, Shahidi, Irfanul Alam – Rice leaf disease
detection using machine learning techniques. || 2019
[15] Ramesh, S., Vydeki, D – Recognition, and classificationof
paddy leaf disease using optimizeddeep neural network
with Jaya algorithm. || 2020
[16] Sanga, S.L., Machuve, D., Jomanga – Mobile-based deep
learning models for Banana disease detection. || 2020
[17] Tewari, V.K., Pareek, C.M., Lal – Image processing based
real-time variable-rate chemical spraying system for
disease control in paddy crop. || 2020
[18] Tiwari, D., Ashish, M., Gangwar, N., Sharma, A., Patel –
Potato leaf disease detection using deep learning. ||
2020
[19] J. G. M. Esgario, R. A. Krohling, and J. A. Ventura – Deep
learning for classification and severity estimation of
coffee leaf biotic stress. || July 2020
[20] Shima Ramesh, Mr.Ramachandra Hebbar,Nivenditha M,
Pooja R, Prasad Bhat – Plant disease detection using
Machine Learning. || April 2018
[21] Hao Tang, Hong Liu, Wei Xiao, and Nicu Sebe – When
dictionary learning meets deep learning: Deep
Dictionary Learning and coding network for Image
Recognition with a limited date. || 2020

Leaf Disease Detection Using Image Processing and ML

More Related Content

Similar to Leaf Disease Detection Using Image Processing and ML (20)

More from IRJET Journal (20)

Recently uploaded (20)

Leaf Disease Detection Using Image Processing and ML