VISVESVARAYA TECHNOLOGICAL UNIVERSITY
BELAGAUM, KARNATAKA -590014
A Seminar Report On
“Classification of Plant Leaf Disease using Machine Learning
and Image Processing”
Submitted in Partial fulfilment of the requirements for the 8th semester
BACHELOR OF ENGINEERING
IN
COMPUTER SCIENCE AND ENGINEERING
By
DEEKSHITHA R
(1SP19CS023)
Under the guidance of:
Ms. SOUJANYA CN
Asst. Professor
Dept of CSE
Department of Computer Science and Engineering
S.E.A. COLLEGE OF ENGINEERING AND TECHNOLOGY
BENGALURU-560049
2022-2023
S.E.A COLLEGE OF ENGINEERING AND TECHNOLOGY
Ekta Nagar, Basavanpura, Virgonagar Post, K. R. Puram, Bengaluru, Karnataka 560059
DEPARTMENT OF COMPUTER SCIENCE AND ENGINEERING
CERTIFICATE
This is to certify the project work entitled “Classification of leaf disease using Machine
learning and image processing” has been successfully carried out by DEEKSHITHA R
(1SP19CS0023) of VIII Sem in partial fulfilment for the award of Bachelor of Engineering in
Computer Science and Technology of the Visvesvaraya Technological University, Belgaum during the
year 2022. The project report has been approved as it satisfied the academic requirement in respect of
theProject work prescribed for the Bachelor of Engineering.
Signature of Guide Signature of HOD Signature of Principal
Prof. Soujanya CN Dr. B. Loganayagi Dr. B Venkata Narayan
Prof and HOD of CSE
Signature of Internal Examiner Signature of External Examiner
ACKNOWLEDGEMENT
Firstly, I thank the Management late Shri A Krishnappa, Chairman SEA
College of Engineering and Technology for Providing the Necessary
infrastructure and creating a good environment.
We would like to thank our respected Dr. BHAGVANT DESHPANDE,
Director of SEA College of Engineering and Technology for the encouragement
and support given by him.
I express my deep sense of gratitude to S.E.A COLLEGE OF ENGINEERING
& TECHNOLOGY, BANGALORE that provided me an opportunity in
fulfilling my most cherished desire of reaching the goal.
I would like to express my profound thanks to our respected and principal
Dr. B Venkata Narayana for the encouragement and support given by him.
I would like to express my sincere thanks to our respected Dr. B. Loganayagi,
HOD of COMPUTER SCIENCE AND ENGINEERING department, for his
assistance and guidance.
I am thankful for the support rendered by my Project guide and coordinator
Prof. Soujanya CN for his valuable suggestions.
I am also obliged, to the faculty members of CSE Department who rendered
their valuableassistance for the Project.
And finally, I would like to express my heart full gratitude to my friends and all
those who haveextended their help throughout my Project.
DEEKSHITHA R
(1SP19CS023)
ABSTRACT
In a developing country like India agriculture plays a noteworthy role. Agricultural intervention
in the livelihood of rural India indulges by about 58%. Among the agricultural products, tomato
is one of the most used crops. Thus, preventing significant loss in quantity and yield of tomato is
majorly dependent on recognition and classification of diseases a tomato plant might possess.
Latest and fostering technologies like Image processing is used to rectify such issues using
different types of techniques and algorithms. Initially, the leaves of a tomato plant get affected,
when plant develops a particular type of disease. In this project, four consecutive stages are used
to discover the type of disease. The four stages include pre- processing, leaf segmentation,
feature extraction and classification. To remove the noise, we aredoing the pre-processing and to
part the affected or damages area of the leaf, image segmentation is used. The k- nearest
neighbors (KNN) algorithm, which is a guided, supervised and advance machine learning
algorithm, is implemented to find solutions for both the problems related to classification and
regression. During the terminal stage, user is recommended with the treatment. Mostly live
plants are adversely affected by the diseases. This paper imparts representation of leaf disease
detection employing image processing thatcan identify drawbacks in tomato plant from images,
based on color, bound and texture to give the brisk and reliable results to the farmer.
TABLE OF CONTENTS
Abstract I
Declaration II
Acknowledgement III
Table of Contents IV
List of Figures VI
List of Tables VII
Chapter No. Chapter Name Page No.
1 Introduction 1
1.1 Introduction 1
2 Machine Learning 3
2.1 Introduction 3
2.2 Types of Machine Learning 3
3 System Architecture 5
3.1 Steps for System Architecture 5
4 Execution Procedure 7
4.1 Procedure 7
4.2 Detail Design 12
iv
4.3 System Design 12
5 Conclusion 15
References 16
v
INTRODUCTION
CHAPTER 1
In countries like India, it is of utmost importance to bring technological advancement in the fields related
to crop productivity. Research initiatives and tentative study process in the important domain of qualitative
farming is focused towards improving the yield and food crop standard at low cost, with greater monetary
outcome. Agricultural building model stands as a result of a compound interlinking of soil with seeds and
chemicals used to enhance growth. Vegetable and fruits exist as one of the present significant agricultural
achieved outputs. In directive for getting surplus and effective worthy products, a product value examination
andimprovement has always been importantly imperative. Diseases are disablement to the conventional
state of the plant that translates or hinders its important roles such as transpiration, photosynthesis,
fertilization, pollination,germination etc. They distorting diseases are spawned by pathogens like, fungi,
bacteria and viruses, because of unfavorable environmental situations. Accordingly, the preliminary stage
for diagnosing of plant disease is a significant task. Farmers need periodic monitoring by professionals
which might be prohibitively costly and time absorbing. Thence, lookingfor quick, less costly and precise
ways to smartly detect the diseases from the indicators that look to be on the plant leaf is of great pragmatic
importance. In our study we are proposing a system which can be used to identify the particular type of
disease a tomato leaf might have. It is of major concern to identify the type of disease an important crop
like tomato can have, by implementing upbringing technologies like image recognition, which represent
the application functioning visually and it is also an important reason for making digital technologies
popular.
Many people and technological groups are involved in the field of agriculture to increase the yield
and throughput. There has been various techniques used in the past to solve problems related to disease
spread in atomato plant. With the advancement in technology tomato plant disease detection have become
easier and more precise. In our system a different approach, i.e. KNN algorithm is used for the same.
Various kind of methods have been used recently to determine the type of plant disease. Some of these
involves analysis and study of chemical analysis method to determine plant diseases and ways which are
indirect by implementing physical techniques, like spectroscopy of the leaf and imaging to get information
related to properties of tomato plant. Following this, the merits of the project contrasted with the existing
technologies are related to the underlying points. The system avoids the process involved in gathering inputs
for studying them in the laboratory, becauseof pre-existing images taken in place of the plant diseases. It
examines the chances where a particular plant is concurrently simulated with higher than one pest or disease
in the unchanged recorded input. The outlook
Classification of plant leaf disease using ML and image pre-processing Introduction
images apprehended by various cameras with diverse resolutions, like mobile phone and the other available
cameras devices. The project is systematically pact with different conditions related to illuminations, the size
of actors in an image, and surrounding distinction, etc. holding across the neighboring part of that particular
plant. It imparts feasible functioning approach that is able to manoeuvre in the domain by not using costly
andcomplex and compound technologies. The economy of India is mainly depending on agricultural,
throughout this area, plant leaf disease identification and classification/recognition plays an important role.
The quantity, quality or production of the respective agricultural field mainly depends on the proper care, not
providing proper care can cause large effectson plants.
Most plant diseases are caused by fungi, bacteria, and viruses. Plant disease identification by visual way
is more laborious task and at the same time less accurate and can be done only in limited areas. Whereas if
automatic detection technique is used it will take less efforts, less time and provide better accuracy. In plants,
some general diseases are Bacterial spot, Early blight, Late blight, Powdery mildew, mold, Septoria leaf spot,
brown and yellow spots, early and late scorch and other fungal, viral and bacterial diseases.
We have surveyed various types of plant diseases and their detection techniques. There are two main
approaches for plant disease detection Machine learning techniques and Deep learning techniques. Image
processing is used for measuring affected area of disease, and to determine the difference in the color of the
affected area. The main aim of this paper is to provide survey on various plant disease detection systems that
can improve agricultural production.
Dept of CSE, SEACET 2022-2023 2
CHAPTER 2
MACHINE LEARNING
Machine learning (ML) is the study of computer algorithms that improve automatically through experience. It
is seen as a subset of artificial intelligence. Machine learning algorithms build a model based on sample data,
known as "training data", in order to make predictions or decisions without being explicitly programmed to do
so. Machine learning algorithms are used in a wide variety of applications, such as email filtering and computer
vision, where it is difficult or unfeasible to develop conventional algorithms to perform the needed tasks. A
subset of machine learning is closely related to computational statistics, which focuses on makingpredictions
using computers; but not all machine learning is statistical learning. The study of mathematical optimization
delivers methods, theory and application domains to the field of machine learning. Data miningis a related field
of study, focusing on exploratory data analysis through unsupervised learning. In its application across business
problems, machine learning is also referred to as predictive analytics.
Machine Learning involves computers discovering how they can perform tasks without being explicitly
programmed to do so. It involves computers learning from data provided so that they carry out certain tasks. For
simple tasks assigned to computers, it is possible to program algorithms telling the machine how to execute all
steps required to solve the problem at hand; on the computers part, no learning is needed. For more advanced
tasks, it can turn out to be more effective to help the machine develop its own algorithm, rather than having
human programmers specify every needed setup. The discipline of machine learning employs various
approaches to teach computers to accomplish tasks where no fully satisfactory algorithm is available. In cases
where vast numbers of potential answers exist, one approach is to label some of the correct answers as valid.
This can then be used as training data for the computer to improve the algorithms it uses to determine correct
answers.
Dept of CSE, SEACET 2022-2023 3
Classification of plant leaf disease using ML and image pre-processing Machine learning
A Machine Learning system learns from historical data, builds the prediction models and whenever it receives
new data, predicts the output for it. The accuracy of predicted output depends upon the amount of data, as the
huge amount of data helps to build a better model which predicts the output more accurately. Suppose we
have a complex problem, where we need to perform some predictions, so instead of writing a code for it, we
just need to feed the data to generic algorithms, and with the help of these algorithms, machine builds the
logic as per the data and predict output. Machine learning has changed our way of thinking about the problem.
The block diagram 2.1 explains the working of machine learning algorithm.
Figure 2 1: Working of Machine Learning Algorithm.
2.2 : Types of Machine learning
➢ Supervised learning: In this type of machine learning, data scientists supply algorithms with labeled
training data and define the variables they want the algorithm to assess for correlations. Both the input
and the output of the algorithm is specified.
➢ Unsupervised learning: This type of machine learning involves algorithms that train on unlabeled
data. The algorithm scans through data sets looking for any meaningful connection. The data that
algorithms train on as well as the predictions or recommendations they output are predetermined.
➢ Semi-supervised learning: This approach to machine learning involves a mix of the two preceding
types. Data scientists may feed an algorithm mostly labeled training data, but the model is free to
explore the data on its own and develop its own understanding of the data set.
➢ Reinforcement learning: Data scientists typically use reinforcement learning to teach a machine to
complete a multi-step process for which there are clearly defined rules. Data scientists program an
algorithm to complete a task and give it positive or negative cues as it works out how to complete a
task. But for the most part, the algorithm decides on its own what steps to take along the way.
Dept of CSE, SEACET 2022-2023 4
CHAPTER 3
SYSTEM ARCHITECTURE
Architecture Structure depicts how the system interacts and the control flows from one point in the
cycle to another. There we will discuss the hardware control flow of the system right from capturing
from the image to detection of the disease and displaying it.
Figure 3.1: Shows the hardware and software control flow of the system as a whole.
3.1 Steps for System Architecture
1. Image Capturing through Camera
The images of the diseased plant leaf are captured via a mobile or a digital camera. The image captured
is then transferred to the system and it is followed by the process involving the image processing stage.
The resultant image is given as an input to this software system. This image forms our testing data
which will determine the accuracy of our system.
2. Dataset Collection and Creation
The dataset is a collection of huge amounts of data that is unprocessed and needs to be trained in
order to yield some useful information out of it. The dataset used in our system is collection of images
of various plants leaf’s that either are healthy or fall into one of the diseased categories
Dept of CSE, SEACET 2022-2023 5
Classification of plant leaf disease using ML and image pre-processing System architecture
➢ Early Blight
➢ Bacterial Spot
➢ TYLCV
Fig 3.2: Tomato Leaf
The database needs to be huge in order to ensure maximum accuracy and precision of our disease
identification model. We have a total of 200 leaf images: 50 each for healthy, early blight, bacterial
spot, and TYLCV. All these images are resized and refined to one specific quality and dimensions so
as to have a uniform dataset.
3. Digital Image Processing
It is a whole big mechanism comprising of various stages and algorithms. If we look from the
entire process as a whole, the function of this stage is to:
1. Perform training of the images that are pre-composed and collected by us and stored in the
form of our dataset.
2.Perform testing on the image of the plant leaf captured by us to know if the leaf is diseased
or not. The dataset collected forms the training data which is trained against our image
processing model. This model is then saved and used to test the images taken in by the camera.
The Web interface comes into requirement when we need a user interface wherein a user needs
to upload the captured picture into the front-end and the model is pre-trained by the dataset of
images in the back-end. The result is generated on the user interface without needing the user
to navigate between training and testing phase of the system. This interface renders easy and
smooth flow of control and the user does not need to know the entire mechanism behind the
Plant Disease Identification System.
Dept of CSE,SEACET 2022-2023 6
CHAPTER 4
EXECUTION PROCEDURE
4.1 PROCEDURE
The Digital Image Processing is a combination of a number of steps and algorithms that go together
in a controlled flow manner. The flowchart as shown below depicts the various stages that the
images goes through before the final result is produced.
Figure 4.1: Execution flow
1. PRE-PROCESSING
This step involves processing the test image in order to bring it to the size, color and quality of the
images comprising our dataset. This involves various stages through which the image goes. These
stages are:
➢ Image Resizing: The dimensions of the image are brought equal to the dimensions of the training
images by using the imresize ( )‟ method in MATLAB. Image resizing is a crucial step as
the pixel values may change if the dimensions of all the training as well as the testing images are not
the same.
Dept of CSE, SEACET 2022-2023 7
Classification of plant leaf disease using ML and image pre-processing Execution Procedure
➢ Smoothening: The smoothening of the image renders the pixel values to gradually even out to allthe
points of the image so as to allow a smooth image. Along with this the image also gets converted from
colored to grayscale image using the function RGB2GRAY ( )”.
➢ Noise Filtering: The noise is the unwanted extras that are present in the images that make feature
determination and extraction difficult. Thus the process of noise filtering involves removal
or averaging of the pixel values that add noise to the image. The process used in our system to ensure
noise removal is “Median Filter”.
Figure 4.1.1: Pre-processing
2. Feature Extraction
Feature Extraction is a method used for dimensionality reduction which helps in representing into a
compact feature vector, the parts of the image which are interesting. This process results to be very
helpful when the sizes of images are large and for the faster image matching and retrieval, reduced
feature representations are required to complete tasks quickly.
The Feature Extraction techniques we use in our system are as this:
➢ Histogram of Oriented Gradients (HOG):
It captures characteristics by calculating the occurrence of the gradient orientation. The HOG as
shown in Fig. 5 is used to divide the image into unique parts and computes a histogram of gradient
orientation representation over the images. HOG are enforced in huge object recognition fields like
recognition based on face and in our project, for plant leaf recognition.
Dept of CSE, SEACET 2022-2023 8
Classification of plant leaf disease using ML and image pre-processing Execution Procedure
The histogram of oriented gradient is being practiced in various study centers with amazing and
outstanding performance and working in fields like pedestrian detection. A fast arithmetical approach
was build and many unique elements that are responsible the HOG‟s performance were calculated
to build a HOG descriptor consisting of fine scale gradient.
Figure 4.1.2: Histogram of Oriented Gradients (HOG) with its features
➢ Gray Scale Co-occurrence Matrix (GLCM):
The texture filter is implemented to show and calculate the view of texture which is statistical found
on the image histogram. These are used to provide appropriate result about the texture of an image
but it unable give data related to the shape that is the spatial relationship between the pixels
of a given image. Feature extraction requires reducing the number of resources needed to regulate a
huge set of information.
The extraction of GLCM is based on a statistical texture analysis; this texture is figured out from the
numerical distribution of observed mix of intensities at the specified position to each other in the
image. The GLCM feature is useful in motion estimation of videos and real time pattern of
Recognizing operations such as Military and Medical Applications.
Dept of CSE,SEACET 2022-2023 9
Classification of plant leaf disease using ML and image pre-processing Execution Procedure
Figure 4.1.3: GLCM
The output of this phase is a vector or a matrix representing various gradient descent and
various useful features that further get analyzed and classified.
3. Classification
The final stage of our Image Processing Phase is the training of the dataset and testing of the images
against the trained model. The algorithm used in this classification model is the KNN. KNNalgorithm
can be explained as supervised machine learning algorithm as it can be implemented to
find solutions of both the classification and regression problems. This algorithm begins with making
an assumption that the things with more similarity are existing in close proximity, or we can also say
that similar things are near to each other.
In KNN algorithm we first start with the loading of the data and then initializing K to the selected
number of neighbors, then for each example in the data we find the distance between the
Dept of CSE,SEACET 2022-2023 10
Classification of plant leaf disease using ML and image pre-processing Execution Procedure
query example and the given current example from the given data. Then the data collected is sorted
orderly from smallest to largest according to the distances, later we pick the first k entries from the
sorted collection and take the labels of the selected k entries.
If there is regression then the algorithm returns mean and in case for the classification the mode
value is returned. Hence our images go through this algorithm and get classified according to the
disease category they belong to. It is a very clean and precise process which produces accurate results.
Figure 4.1.4: KNN Classification Algorithm
Dept of CSE,SEACET 2022-2023 11
Classification of plant leaf disease using ML and image pre-processing Execution Procedure
4.2 DETAILED DESIGN
Once the requirements document for the software to be developed is available, the softwaredesign phase
begins. While the requirement specification activity deals entirely with the problem domain, design is the
first phase of transforming the problem into a solution. In the design phase, the customer and business
requirements and technical considerations all come together to formulatea product or a system.
The design process comprises a set of principles, concepts and practices, which allow a software
engineer to model the system or product that is to be built. This model, known as design model, is assessed
for quality and reviews before a code is generated and tests are conducted. The design model provides
details about software structures, architecture, interfaces and components which are required to implement
the system. This chapter discusses the design elements that are required to develops software design model.
4.3 System Design
System design is the process of planning a new system to compliment or all together replace the old system.
The purpose of the design phase is the first step in moving from the problem domainto the solution domain.
The design of the system is the critical aspect that affects the quality of theaspects of the system into physical
aspects of the system. It is the process of defining thearchitecture, modules, interfaces, and data for a system
to satisfy specified requirements. System design could be seen as the application of system theory product
development. There is some overlap with the disciplines of system analysis, system architecture, and system
engineering.
Dept of CSE,SEACET 2022-2023 12
Classification of plant leaf disease using ML and image pre-processing Execution Procedure
Figure 4.3: System Design
4.3.1 High Level Design
High-level design (HLD)explains the architecture that would be used for developing a software product. The
architecture diagram provides an overview of an entire system, identifying the main components that would be
developed for the product and their interfaces. The HLD uses possibly nontechnical to mildly technical terms
that should be understandable to the administrators of the system.In contrast, low-level design further exposes
the logical detailed design of each of these elements for programmers.
A high-level design provides an overview of a system, product, service or process. Such an overviewhelps
supporting components be compatible to others. The highest-level design should briefly describeall platforms,
systems, products, services and processes that it depends on and include any important changes that need to be
made to them. In addition, there should be brief consideration of all significant commercial, legal,
environmental, security, safety and technical risks, issues and assumptions. The idea is to mention every work
area briefly.
Dept of CSE,SEACET 2021-2022 13
Classification of plant leaf disease using ML and image pre-processing Execution Procedure
Figure 4.3.1: High Level Design
➢ This project is an automated system to detect and classify powdery mildew using
RandomForest classification techniques.
➢ First, the infected image is given as input to the application then we convert the RGB
images into BGR images.
➢ Then in segmentation the image is converted into gray image.
➢ Features are extracted by applying global feature descriptor. Global features are extracted from
the image using three feature descriptors namely:
o Color: Color Channel Statistics (Mean, Standard Deviation) and
ColorHistogram
o Texture: Haralick Texture, Local Binary Patterns (LBP)
o Shape: Hu Moments.
➢ Finally, the extracted features are evaluated using Random Forest classifiers and the
disease is detected.
Dept of CSE, SEACET 2022-2023 14
CHAPTER 5
CONCLUSION
The proposed methodology in the following tomato plant leaf disease detection system focus on
generating an advance and efficient system which makes the process of creating high yield of tomato
much more easier for the farmers. The project aims to detect the most common diseases occurring on a
tomato leaf, namely early blight, bacterial spot and curl using image processing technique under
upbringing technology i.e. machine learning. In easier terms, the farmer will be able to accurately detect
the type of disease a particular plant is having using the image of the plant.
In this study, we describe the comparison of our system with preexisting systems with proper
methodology and implementation. The proposed systems functionality is better than existing disease
detection system as it is able to generate a more accurate and precise result with easier and faster
implementation. It aims to make the life of farmers easier. The system can be a boon to the agricultural
sector as it advances the crop production and management process, as agriculture is of the major reason
to facilitate growth of per capita income of our country.
REFERENCES
[1] [1 Sankaran, S.; Mishra, A.; Ehsani, R. “A review of advanced techniques for detecting plant
diseases”. Comput.Electron. Agric. 2010, 72, 1–13. [CrossRef]
[2] Chaerani, R.; Voorrips, R.E. “Tomato early blight (Alternariasolani): The pathogens, genetics, and
breeding for resistance”, J. Gen. Plant Pathol. 2006, 72, 335–347. [CrossRef]
[3] Alvarez, A.M.” Integrated approaches for detection of plant pathogenic bacteria and diagnosis of
bacterial diseases”, Annu. Rev. Phytopathol. 2004, 42, 339–366. [CrossRef] [PubMed]
[4] Gutierrez-Aguirre, I.; Mehle, N.; Delic, D.; Gruden, K.; Mumford, R.; Ravnikar, M. “Real-time
quantitative PCR based sensitive detection and genotype discrimination of Pepino mosaic virus”,.
J. Virol. Methods 2009, 162, 46–55. [CrossRef] [PubMed]
[5] Martinelli, F.; Scalenghe, R.; Davino, S.; Panno, S.; Scuderi, G.; Ruisi, P.; Villa, P.; Stropiana, D.;
Boschetti, M.; Goudart, L.; et al. “Advanced methods of plant disease detection. A review”
,Agron.Sust. Dev. 2015, 35, 1–25. [CrossRef]
[6] Bock, C.H.; Poole, G.H.; Parker, P.E.; Gottwald, T.R. “Plant Disease Sensitivity Estimated
Visually, by Digital Photography and Image Analysis, and by Hyperspectral Imaging”. Crit. Rev.
Plant Sci. 2007, 26, 59–107.
[7] Down To Earth's print edition dated April 1-15, 2019
[8] Lefeuvre, Pierre, Darren P. Martin, Gordon Harkins, Philippe Lemey, Alistair JA Gray, Sandra
Meredith, Francisco Lakay et al. "The spread of tomato yellow leaf curl virus from the Middle
East to the world." PLoS pathogens 6, no. 10 (2010).
[9] India Agriculture GDP report from Ministry of Statistics and Programme Implementation (MOSPI)
[10] L. S. Puspha Annabel, T. Annapoorani and P. Deepalakshmi, “Machine Learning for Plant Leaf
Disease Detection and Classification – A Review,” 2019 International Conference on
Communication and Signal Processing (ICCSP), Chennai, India, 2019, pp. 0538-0542.
16
