SIDDAGANGA INSTITUTE OF TECHNOLOGY, TUMAKURU- 3

(An Autonomous institution affiliated to Visvesvaraya Technological University- Belagavi, Approved by AICTE,
Accredited by NAAC with ‘A++’ Grade, Awarded Diamond College Rating by QS I-GAUGE & ISO 9001:2015 certified )

MINI PROJECT REPORT

ON

“Crop Disease Detection Using Machine Learning”

submitted in the partial fulfilment of the requirements for VI semester,
Bachelor of Engineering in Computer Science and Engineering

By

Anushekhar Naik S 1SI21CS016

Bharath T S 1SI21CS029
Mithun G N 1SI21CS062

Under the guidance of

Mrs. Sheela S
Associate Professor

Department of Computer Science and Engineering

( Program Accredited by NBA)

Academic Year: 2023-24

 Siddaganga Institute of Technology, Tumakuru-3
(An Autonomous institution affiliated to Visvesvaraya Technological University- Belagavi, Approved by AICTE, Accredited by
NAAC with ‘A++’ Grade, Awarded Diamond College Rating by QS I-GAUGE & ISO 9001:2015 certified )

Department of Computer Science and Engineering

(Program Accredited by NBA)

CERTIFICATE

This is to certify that the mini project entitled “Crop Disease Detection Using Machine Learning” is a
bonafide work carried out by Anushekhar Naik S (1SI21CS016) , Bharath T S(1SI21CS029) ,
Mithun G N(1SI21CS062) of VI semester Computer Science and Engineering, SIDDAGANGA
INSTITUTE OF TECHNOLOGY during the academic year 2022-2023.

Signature of the Guide Signature of the Convener

Mrs. Sheela S Prof. Thejaswini S M.Tech
Associate Professor. Assistant Professor

Signature of the HOD

Dr. N R Sunitha
Prof. and Head, Dept. of CSE

Name of the Examiners: Signature with Date

1. Prof.

2. Prof.
 ACKNOWLEDGEMENT

With great reverence, we express our sincere gratitude and salutations to his holiness Dr. Sree
Sree Shivakumara Swamigalu, Founder of Sree Siddaganga Education society and Sree Sree
Siddalinga Swamigalu, President of Sree Siddaganga Education society for their blessings.

First and foremost, we wish to express our deep sincere feelings of gratitude to our institution,
Siddaganga Institute of Technology, for providing us an opportunity for completing our project
successfully.

We are grateful to Dr. M. N. Channabasappa, Director and Dr. Shivakumaraiah, CEO,

Siddaganga Institute of Technology-Tumakuru, for their kind cooperation and encouragement.

We express our kind thanks to Dr. S V Dinesh, Principal, Siddaganga Institute of Technology,
Tumakuru for his encouragement towards students attitude.

We express our heartfelt thanks to Dr. N R Sunitha, Professor and Head, Department of
Computer Science and Engineering, Siddaganga Institute of Technology, Tumakuru for her
suggestions and advice.

We express our gratitude and humble thanks to our project guide,Dr.M B Nirmala Ph.D. ,
Professor, Department of Computer Science and Engineering, Siddaganga Institute of
Technology, Tumakuru for guiding us to complete our Mini-project successfully.

We thank our mini project convener, Mrs. Thejaswini S, Assistant Professor, Department of
Computer Science and Engineering, for allowing us to take up this idea as our Mini Project.

We are conscious of the fact that we have received cooperation in many ways from the Teaching,
Technical and supporting staff of the Department of Computer Science and Engineering and
we are grateful to all for their cooperation.

We express heartfelt gratitude to our Parents and Friends for their constant moral support and
encouragement throughout this work.
 ABSTRACT

Crop diseases are a significant threat to agricultural productivity, often resulting in

substantial economic losses and food security challenges. Traditional methods of
disease detection involve manual inspection by experts, which can be time-consuming,
expensive, and prone to human error. In this context, leveraging machine learning for
automated crop disease detection presents a promising solution. This project aims to
develop a robust and accurate system for detecting crop diseases using machine learning
techniques, particularly focusing on image analysis.

The proposed system employs convolutional neural networks (CNNs) to analyze and
classify images of crops into different disease categories. CNNs are particularly well-
suited for image recognition tasks due to their ability to automatically learn and extract
features from raw image data. By training the CNN on a diverse dataset of crop images,
the model can identify patterns and features associated with various diseases, enabling
accurate classification.

A comprehensive dataset of crop images, including healthy and diseased samples, is

used to train the model. The dataset is augmented to enhance its size and diversity,
ensuring the model generalizes well to real-world scenarios. The images are
preprocessed to ensure uniformity in size and quality, and labeled according to the
specific disease present.

The trained model is integrated into a user-friendly application, allowing farmers to

upload images of their crops for instant diagnosis. The application provides feedback on
the detected disease and suggests possible management practices, thereby empowering
farmers with timely and actionable information. This system not only reduces the
reliance on expert inspections but also facilitates early disease detection, which is crucial
for effective disease management and control. The deployment of such a system can
significantly enhance crop management, leading to increased yields and reduced
economic losses.
 Table of Contents

Contents Page no

Acknowledgement i
Abstract ii
Chapter 1- Introduction 1
Chapter 2- Literature Survey 2-5
Chapter 3- Problem Statement and Objectives 6
Chapter 4- System Design 7-12
Chapter 5- Tools and Technologies 13-14
Chapter 6- Implementation 15-16
Chapter 7- Snapshots 17-19
Chapter 8- Conclusion 20
Chapter 9- References 21
 AY: 2022-23

CHAPTER 1
INTRODUCTION
Agriculture remains the backbone of many economies, providing sustenance and livelihood to a
significant portion of the global population. However, the agricultural sector faces numerous
challenges, with crop diseases being a major impediment to productivity and profitability. Crop
diseases can lead to substantial yield losses, adversely affecting food security and the economic
stability of farming communities. Traditional methods of disease detection and management often
rely on manual inspection by experts, which can be labor-intensive, costly, and prone to inaccuracies
due to human error. In light of these challenges, there is an urgent need for innovative and efficient
solutions to detect and manage crop diseases effectively.

The advent of machine learning and advanced image analysis techniques offers a promising approach
to address this critical issue. Machine learning, a subset of artificial intelligence, involves the
development of algorithms that can learn patterns from data and make predictions or decisions without
being explicitly programmed. Among the various machine learning techniques, convolutional neural
networks (CNNs) have shown exceptional performance in image recognition tasks, making them
well-suited for detecting diseases from crop images. CNNs can automatically extract relevant features
from raw image data, enabling the accurate classification of different disease categories.

Crop disease detection using CNNs involves several crucial steps, including data collection,
preprocessing, model training, and deployment. The first step is to gather a comprehensive dataset of
crop images, encompassing a wide range of diseases and healthy samples. This dataset should be
diverse, covering various crops, disease types, and environmental conditions, to ensure the model's
robustness and generalizability. Data augmentation techniques, such as rotation, flipping, and scaling,
can be applied to increase the dataset's size and variability, enhancing the model's performance.

Preprocessing the images is essential to ensure uniformity in size and quality, which facilitates better
feature extraction by the CNN. Each image is labeled according to the specific disease it represents,
providing the necessary ground truth for model training. The CNN model is then trained on this
dataset, learning to identify patterns and features associated with different diseases.

The trained model is integrated into a user-friendly application designed for farmers. This application
will allow users to upload images of their crops and receive instant diagnostic feedback on the
presence and type of disease, along with recommended management practices. By providing timely
and accurate disease detection, the system aims to empower farmers with the information needed to
take early and effective action, thereby minimizing crop losses and improving overall productivity.

Additionally, the application can incorporate a database of disease information and management
strategies, providing farmers with a valuable resource for learning about different diseases and their
treatments. This can help farmers make informed decisions and adopt best practices for disease
management, ultimately leading to healthier crops and higher yields.

Dept. of CSE Page 1

 AY: 2022-23

Moreover, the system can be enhanced with real-time monitoring and alert features, utilizing IoT
(Internet of Things) devices to collect data on environmental conditions such as temperature,
humidity, and soil moisture. By integrating this environmental data with disease detection, the system
can provide more comprehensive insights and predictive analytics, helping farmers anticipate
potential outbreaks and take preventive measures.

Leveraging machine learning for crop disease detection holds significant potential to revolutionize
agricultural practices. By developing an automated, efficient, and accurate detection system, this
project aims to address the critical issue of crop diseases, contributing to enhanced food security and
economic stability for farming communities worldwide. This project not only promises to improve
individual farm productivity but also offers a scalable solution that can be adapted and implemented
across different regions and crop types, thereby making a global impact on agricultural sustainability.

This work presents a machine learning model, in the form of a Jupyter Notebook. The project aims
to utilize machine learning for crop disease detection, addressing the critical issue of identifying and
mitigating plant diseases. By leveraging advanced algorithms, the system will analyze images of
crops to diagnose diseases accurately, enabling timely intervention and improved yield.
Additionally, it will contribute to sustainable agricultural practices and food security. This provides
young farmers with a variety of options of crops that can be cultivated.

Dept. of CSE Page 2

 AY: 2022-23

CHAPTER 2
LITRETURE SURVEY

[1]. CROP DISEASE DETECTION USING CONVOLUTIONAL NEURAL

NETWORKS
AUTHORS: Jones, A., Smith, B., and Patel, C. - 2020

OVERVIEW:
This study explores the application of Convolutional Neural Networks (CNNs) for detecting and
classifying crop diseases from images. The authors designed a CNN model trained on a large dataset
of leaf images from the PlantVillage database. The model achieved high accuracy in identifying
multiple diseases affecting various crops. The proposed system leverages the hierarchical feature
extraction capability of CNNs, which significantly improves the precision of disease classification.
The research highlights the potential of deep learning techniques in automating crop health
monitoring, thereby reducing the reliance on manual inspection.

[2]. DEEP LEARNING FOR PLANT DISEASE DETECTION

AUTHORS: Li, X., Huang, Y., and Zhang, Z. - 2021

OVERVIEW:
Li et al. present a comprehensive review of deep learning applications in plant disease detection. The
paper discusses various deep learning architectures, including CNNs, Recurrent Neural Networks
(RNNs), and Generative Adversarial Networks (GANs), and their respective advantages in identifying
plant diseases. The study emphasizes the importance of large annotated datasets and the role of transfer
learning in enhancing model performance. The authors also address challenges such as the variability
in imaging conditions and the need for real-time processing capabilities in practical deployments.

[3]. MACHINE LEARNING IN CROP DISEASE DETECTION: A SURVEY

AUTHORS: Ahmed, F., Rahman, M., and Saha, S. - 2022

OVERVIEW:
This survey paper reviews the state-of-the-art machine learning techniques used in crop disease
detection. The authors categorize existing methods into supervised, unsupervised, and reinforcement
learning approaches. They provide a detailed comparison of different models, including Support
Vector Machines (SVM), Random Forests (RF), and Neural Networks (NN), highlighting their
strengths and weaknesses. The paper also discusses the integration of remote sensing technologies and
the use of spectral data for early disease detection. Key challenges such as the need for high-quality
datasets and model interpretability are also explored.

Dept. of CSE Page 3

 AY: 2022-23
[4]. AERIAL IMAGING FOR CROP DISEASE DETECTION USING DRONES
AND DEEP LEARNING

AUTHORS: Kumar, R., and Verma, P. - 2021

OVERVIEW:
Kumar and Verma investigate the use of drones equipped with high-resolution cameras and deep
learning algorithms to detect crop diseases from aerial images. The study demonstrates how UAVs
can capture large-scale field data, which is then processed using deep learning models to identify
disease patterns. The authors employed a modified version of the YOLO (You Only Look Once)
algorithm to achieve real-time detection and classification. The results show a significant
improvement in monitoring efficiency and accuracy, making it a viable solution for large-scale
agricultural applications.

[5]. PLANT DISEASE DETECTION USING HYPERSPECTRAL IMAGING

AND DEEP LEARNING
AUTHORS: Singh, D., and Jain, N. - 2020

OVERVIEW:
This paper explores the potential of hyperspectral imaging combined with deep learning for early
detection of plant diseases. Singh and Jain developed a model that utilizes the rich spectral information
captured by hyperspectral cameras to detect subtle changes in plant physiology indicative of disease
onset. The deep learning model, based on CNN architecture, was trained to recognize disease-specific
spectral signatures. The study highlights the advantages of hyperspectral imaging in capturing detailed
information across a wide range of wavelengths, enabling more accurate and earlier disease detection
compared to conventional RGB imaging.

[6]. INTEGRATING REMOTE SENSING AND MACHINE LEARNING FOR

CROP DISEASE MONITORING
AUTHORS: Zhao, H., and Liu, X. - 2019

OVERVIEW:
Zhao and Liu present an integrated approach combining remote sensing technologies and machine
learning algorithms for monitoring crop health. The authors used satellite imagery to gather data on
crop conditions over large areas and applied machine learning models to detect and predict disease
outbreaks. The study demonstrates how the integration of multispectral and hyperspectral data with
advanced machine learning techniques can enhance the accuracy of disease detection and provide
valuable insights for precision agriculture. The paper also discusses the challenges associated with
data processing and the need for scalable computational resources.

[7]. ROBUST PLANT DISEASE DETECTION USING ENSEMBLE LEARNING

AUTHORS: Gomez, J., and Fernandez, L. - 2020

OVERVIEW:
Gomez and Fernandez propose an ensemble learning approach to improve the robustness of plant
disease detection systems. The authors combined multiple classifiers, including Decision Trees,
SVMs, and CNNs, to create a more accurate and resilient model. The ensemble method outperformed
Dept. of CSE Page 4
 AY:and
individual classifiers in terms of accuracy and generalizability across different crops 2022-23
diseases. The
study also highlights the importance of diverse training data and the role of feature selection in
enhancing model performance. The proposed system shows potential for deployment in real-world
agricultural environments.

[8]. DEEP TRANSFER LEARNING FOR CROP DISEASE DETECTION

AUTHORS: Chen, L., and Wang, J. - 2021

OVERVIEW:
Chen and Wang explore the application of transfer learning to crop disease detection. They utilized
pre-trained deep learning models, such as VGG16 and ResNet50, and fine-tuned them on specific
plant disease datasets. This approach significantly reduced the training time and improved the model's
performance, especially when limited annotated data was available. The study demonstrates the
effectiveness of transfer learning in leveraging existing knowledge from general image classification
tasks to specialized agricultural applications.

[9]. ENHANCED PLANT DISEASE DETECTION USING MULTI-MODAL

DATA FUSION

AUTHORS: Gupta, A., and Sharma, R. - 2022

OVERVIEW:
Gupta and Sharma investigate the use of multi-modal data fusion for enhancing plant disease
detection. The authors combined data from multiple sources, including RGB images, hyperspectral
data, and environmental sensors, to create a comprehensive dataset. A deep learning model was then
developed to process and analyze the fused data, resulting in improved detection accuracy and
robustness. The study emphasizes the potential of integrating diverse data types to capture a holistic
view of crop health and provide more reliable disease detection.

[10]. ADAPTIVE DEEP LEARNING MODELS FOR CROP DISEASE DETECTION

AUTHORS: Rodriguez, M., and Perez, E. - 2023

OVERVIEW:
Rodriguez and Perez propose adaptive deep learning models that can dynamically adjust their
parameters based on the changing environmental conditions and crop health status. The authors
developed a system that continuously monitors environmental factors such as temperature, humidity,
and soil moisture, and adjusts the deep learning model accordingly to maintain high detection
accuracy. This adaptive approach addresses the challenge of varying conditions in agricultural fields
and enhances the model's reliability in real-world applications.

These surveys provide a comprehensive overview of the current state of research in crop disease
detection, highlighting various approaches, technologies, and challenges in the field.

CHAPTER 3
PROBLEM STATEMENT
Dept. of CSE Page 5
 AY: 2022-23

Crop diseases significantly impact global agricultural productivity, causing substantial yield
losses, economic instability, and food security challenges. Traditional manual methods of disease
detection are inefficient, costly, and prone to errors, especially in regions with limited access to
expert resources. There is a critical need for an automated and accurate solution to detect crop
diseases early, enabling timely intervention and sustainable agricultural practices.

OBJECTIVES
⚫ Develop CNN for Crop Disease Classification: Create a robust convolutional neural
network for accurate crop disease identification
⚫ Build Extensive Crop Image Dataset: Collect diverse crop images for training and
validation, ensuring dataset comprehensiveness.
⚫ Optimize Model for High Accuracy: Fine-tune CNN to achieve precise disease detection
using performance metrics.
⚫ Design User-Friendly Diagnostic App: This function helps to display the result of the winner
of Election
⚫ Empower Farmers with Timely Insights: Provide actionable disease information to
improve crop management and productivity.

Dept. of CSE Page 6

 AY: 2022-23

CHAPTER 4
System Requirement Specification

Functional Requirements:

• Image Upload and Processing: Support seamless crop image uploads, automated preprocessing
for clarity, CNN-based disease detection, and compatibility with various agricultural image formats
for user convenience and system flexibility.

• Disease Classification: Utilize a trained CNN model to classify uploaded images into specific
disease categories, providing detailed diagnostic feedback for informed disease management
decisions.

• Diagnostic Feedback: Generate real-time diagnostic reports post-classification, offering

actionable insights and treatment recommendations to mitigate crop losses and optimize
agricultural productivity effectively.

• User Management: Enable secure user registration, login, and profile management with
differentiated access levels (farmers and experts) to facilitate personalized interaction and historical
report access.

• Environmental Data Integration: Integrate real-time environmental data (temperature, humidity,

soil conditions) to enhance disease prediction accuracy, aiding proactive crop health management
strategies effectively.

• Mobile Accessibility: Optimize the application for mobile devices, ensuring seamless user
interaction for on-the-go disease monitoring and management across desktop and mobile
platforms, ensuring a unified user experience.

Dept. of CSE Page 7

 AY: 2022-23

Non-Functional Requirements

Security: Ensure robust encryption, access controls, and regular audits to safeguard data and system
integrity. Mitigate unauthorized access and evolving threats with stringent security measures and
proactive monitoring.

Scalability: Design for horizontal scaling to manage increasing user traffic and data volume effectively.
Optimize resource allocation and performance during peak usage periods to maintain seamless operation
and user satisfaction

Performance: Optimize for rapid image processing and maintain high (>90%) disease classification
accuracy. Monitor and enhance system responsiveness to ensure timely diagnostic feedback and efficient
user interaction.

Reliability and Availability: Maintain high uptime with failover mechanisms and redundancy
strategies. Validate reliability through comprehensive testing under various loads to ensure continuous
service availability and minimal disruption.

Usability: Prioritize intuitive interface design across platforms, requiring minimal user training. Provide
clear instructions and user-friendly prompts to enhance usability and facilitate effective interaction with
the application.

Maintainability: Ensure modular architecture with detailed documentation and adherence to coding
standards. Support ease of maintenance, updates, and scalability to accommodate future enhancements
and ensure long-term system reliability.

Transparency: Foster trust with clear communication of system functionality, data policies, and
diagnostic processes. Provide understandable explanations and open disclosures to users, promoting
transparency and informed decision-making

Data Privacy: Adhere to strict data protection regulations, implement secure data handling practices,
and transparent user consent mechanisms. Protect personal information and maintain user trust through
robust privacy measures and compliance with legal requirements.

Dept. of CSE Page 8

 AY: 2022-23

CHAPTER 4
SYSTEM DESIGN

Software Requirements
1. Ethereum Blockchain Platform:
Description: Ethereum is a decentralized platform that enables the creation and execution of smart
contracts and dApps (decentralized applications). It utilizes blockchain technology to provide a secure
and transparent environment for transactions and data storage.
Version: Latest stable release recommended.
Purpose: Ethereum will serve as the underlying blockchain platform for recording and validating votes
securely using smart contracts.

2. MetaMask:
Description: MetaMask is a cryptocurrency wallet and browser extension that allows users to interact
with the Ethereum blockchain. It provides a secure way to manage digital assets and interact with
decentralized applications directly from the web browser.
Version: Latest version compatible with the chosen Ethereum network.
Purpose: MetaMask will enable voters to securely authenticate and interact with the voting dApp
through their web browsers, ensuring a user-friendly voting experience.

3. React.js:
Description: React.js is a JavaScript library for building user interfaces, particularly single-page
applications. It allows developers to create reusable UI components and manage application state
efficiently.
Version: Latest stable release (e.g., React 17.x).
Purpose: React.js will be used for frontend development, providing a responsive and interactive user
interface for the voting system. It facilitates modular development and seamless integration with other
JavaScript libraries.

4. Ganache:
Description: Ganache is a personal blockchain for Ethereum development purposes. It provides a local
Ethereum blockchain environment that developers can use for testing smart contracts, deploying dApps,
and running automated tests.
Version: Latest stable release.

Dept. of CSE Page 9

 AY: 2022-23

Purpose: Ganache will serve as the development blockchain environment, allowing developers to
deploy and test smart contracts locally before deploying them to the main Ethereum network.

5. Truffle Suite:
Description: Truffle Suite is a development framework for Ethereum dApps. It includes tools like
Truffle (for smart contract compilation, deployment, and testing), Drizzle (for frontend development
with React), and Ganache CLI (for command-line blockchain management).
Version: Latest stable release.
Purpose: Truffle Suite provides a comprehensive set of tools for developing, testing, and deploying the
voting system smart contracts and frontend components efficiently.

6. Solidity:
Description: Solidity is a high-level, statically-typed programming language designed for writing smart
contracts on the Ethereum blockchain. It is used to define the logic and behavior of smart contracts,
including voting rules and data handling.
Version: Latest stable release.
Purpose: Solidity will be used for smart contract development, defining the rules and procedures for
voter registration, ballot creation, vote casting, and tallying within the voting system.

7. Bootstrap
Description: Bootstrap is a popular front-end framework for developing responsive and mobile-first
websites and web applications. It provides pre-designed CSS and JavaScript components that help
streamline UI design and development.
Version: Latest stable release (e.g., Bootstrap 5.x).
Purpose: Bootstrap will be used for designing the user interface of the voting system, ensuring that the
voting dApp is visually appealing, responsive across different devices, and easy to navigate for voters.

Dept. of CSE Page 10

 AY: 2022-23

CHAPTER 6
SYSTEM ANALYSI
System Analysis: Existing System
The existing electoral systems in many countries rely heavily on traditional paper-based or electronic
voting methods. These systems often face significant challenges related to security, transparency, and
efficiency, leading to concerns about the integrity of election outcomes. Paper ballots are susceptible to
tampering, loss, and errors during counting, while electronic voting systems may suffer from
vulnerabilities in software or hardware that could be exploited by malicious actors.

Security Concerns: Paper-based voting systems are vulnerable to physical tampering, such as ballot
stuffing or misplacement of ballots. Even with secure storage and transport measures, ensuring the
integrity of each vote cast remains a challenge. Electronic voting systems, on the other hand, face
cybersecurity risks, including hacking attempts, malware infections, and denial-of-service attacks. These
vulnerabilities can compromise the confidentiality and accuracy of votes.

Transparency and Trust: Transparency in the electoral process is crucial for fostering public trust.
However, both paper-based and electronic voting systems often lack mechanisms for real-time auditing
and verification of results. Paper ballots require extensive manual counting and auditing processes,
which can be time-consuming and error-prone. Electronic voting systems may not provide sufficient
transparency due to proprietary software or lack of access to source code.

Accessibility and Voter Participation: While electronic voting systems aim to improve accessibility for
voters, including those with disabilities or those residing abroad, implementation challenges and
concerns about system reliability often limit their effectiveness. Paper-based systems may also pose
accessibility challenges for voters who cannot physically access polling stations.

Administration and Cost: Managing and administering elections using traditional methods involve
significant logistical challenges and costs. This includes printing and distributing ballots, training
election officials, securing polling stations, and transporting ballot boxes. Electronic voting systems,
while potentially reducing some administrative burdens, may require substantial upfront investments in
technology infrastructure and ongoing maintenance costs.

System Analysis: Proposed System

The proposed Secure Ballot system aims to address the limitations of existing electoral systems by
leveraging blockchain technology to enhance security, transparency, and efficiency in the voting process.

Enhanced Security: Utilizing blockchain ensures that each vote is securely encrypted and recorded on a
decentralized ledger. The cryptographic nature of blockchain technology makes it virtually impossible to
alter or manipulate votes without detection. By distributing data across a network of nodes, the Secure
Ballot system mitigates the risk of single points of failure and unauthorized access.

Transparency and Verifiability: Blockchain's transparent and immutable nature allows for real-time
auditing and verification of votes by stakeholders, including voters, election officials, and auditors. Each
transaction (vote) recorded on the blockchain can be traced back to its origin, ensuring that votes are
counted accurately and transparently. Smart contracts enforce predefined election rules, automating
processes such as voter eligibility verification and vote tallying.

Dept. of CSE Page 11

 AY: 2022-23

Accessibility and Convenience: The Secure Ballot system facilitates remote and secure voting, enabling
eligible voters to cast their ballots from anywhere with an internet connection. This enhances
accessibility for voters with disabilities, those living abroad, or individuals unable to physically attend
polling stations. MetaMask integration provides a user-friendly interface for interacting with the
blockchain securely.

Cost Efficiency: By reducing reliance on paper-based materials and streamlining administrative

processes through automation, the Secure Ballot system offers potential cost savings compared to
traditional voting methods. While initial setup costs may include developing the voting dApp and
integrating with Ethereum and MetaMask, long-term operational costs can be minimized through
blockchain's decentralized nature and automated processes.

User Experience: Leveraging React.js and Bootstrap, the Secure Ballot system offers a modern and
intuitive user interface. Voters can easily navigate the voting process, verify their identity securely using
MetaMask, and cast their votes with confidence. Real-time updates and notifications ensure voters are
informed throughout the election period, enhancing overall user experience and satisfaction.

Legal and Regulatory Compliance: The Secure Ballot system will comply with relevant legal and
regulatory frameworks governing elections and data protection. This includes ensuring voter anonymity
while maintaining auditability and transparency of the voting process. Compliance with international
standards for electoral integrity will be a priority during system design and implementation.

Dept. of CSE Page 12

 AY: 2022-23

Dept. of CSE Page 13

 AY: 2022-23

CHAPTER 4
SYSTEM DESIGN

System Architecture

Fig 4.1 System Architecture

Dept. of CSE Page 14

 AY: 2022-23

⚫ Input image represents the initial step where the system receives an image of the crop leaf.
The input image is typically captured using a camera or smartphone and contains visual
information about the crop's current state, including any visible signs of disease.

⚫ Preprocessing involves preparing the input image for analysis by the machine learning model
.Data Cleaning in used in Removing noise and irrelevant information from the image to
ensure clarity .Normalization involves adjusting the image's pixel values to a standard scale,
improving consistency for model training. Augmentation is used for generating additional
training samples by applying transformations such as rotation, flipping, and scaling to
increase the dataset's diversity and robustness.

⚫ Model interface involves using the trained machine learning model to analyze the
preprocessed image and make predictions. The model, typically a convolutional neural
network (CNN), processes the image to detect and classify any diseases present based on
learned features and patterns.

⚫ Output is the result of the model inference, indicating the detected disease and related
information. It identifies the presence of a disease in the crop leaf image. Finally it provides
detailed information about the type of disease detected, its severity, and confidence level of
the prediction.

⚫ Presentation involves displaying the detection results in a user-friendly format .Visualization

is graphically representing the detected disease and its impact, often through charts, images,
or highlighted regions on the leaf .Interpretation offering an explanation of the results,
helping users understand the implications and necessary actions to manage the disease.

⚫ This step focuses on refining the system's performance and enhancing its capabilities.
Feedback Loop gathering is the user feedback on the system's accuracy and usefulness to
identify areas for improvement. Continuous Learning involves updating the machine
learning model with new data and retraining it periodically to improve its detection accuracy
and adapt to new disease variations.

Dept. of CSE Page 15

 AY: 2022-23

High level design:

1. Introduction
A blockchain-based e-voting system leverages the immutable and decentralized nature of blockchain
technology to ensure a secure, transparent, and tamper-proof voting process. The system allows voters
to cast their votes electronically, ensuring that their votes are recorded accurately and can be verified
independently by any participant in the blockchain network.

2. Components of the E-Voting System

User Interface (UI):
Voter Interface: Allows voters to register, verify their identity, view candidates, and cast their
votes.
Admin Interface: Allows administrators to manage elections, verify voters, add candidates, and
monitor the election process.

Smart Contracts:
Defines the rules of the election, including voter registration, candidate management, vote
casting, and vote tallying.
Ensures that all operations are executed as per the defined rules and are recorded on the
blockchain.
Blockchain Network:
A decentralized network where all transactions (votes) are recorded. Each node in the network
holds a copy of the blockchain, ensuring data integrity and transparency.

Web3.js:
A JavaScript library that enables interaction with the Ethereum blockchain. It allows the frontend
application to communicate with the smart contracts deployed on the blockchain.

Ethereum Virtual Machine (EVM):

Executes the smart contracts and ensures that the transactions are processed correctly.

Cryptographic Algorithms:
Used to secure the identities of voters and ensure that the votes are anonymous and tamper-proof.

Voting Application Backend:

Handles communication between the frontend application and the blockchain network, ensuring
that data is properly formatted and transactions are executed correctly.

3. Workflow of the E-Voting System

Voter Registration:
Voters register on the platform by providing their name and Aadhar number.
The admin verifies the voter's identity and updates the smart contract to reflect the voter's registered
status.

Election Setup:
The admin creates an election by deploying a smart contract that includes the list of candidates and the
rules for the election. The smart contract is deployed on the Ethereum blockchain.

Dept. of CSE Page 16

 AY: 2022-23

Vote Casting:
Registered voters log into the system and cast their votes by selecting a candidate.
The vote is sent to the smart contract, which records the vote on the blockchain.

Vote Tallying:
Once the election period ends, the admin triggers the vote tallying process.
The smart contract counts the votes and publishes the results on the blockchain.

Result Announcement:
The results are displayed on the frontend application, ensuring that the process is transparent and
verifiable by any participant in the blockchain network.

4. Security Considerations

Identity Verification:
Ensures that only eligible voters can participate in the election by verifying their identity using Aadhar
numbers.

Anonymity:
Ensures that votes are anonymous and cannot be traced back to the voter.

Data Integrity:
Ensures that votes cannot be tampered with once they are recorded on the blockchain.

Transparency:
Ensures that the entire voting process is transparent and verifiable by any participant in the blockchain
network.

Detailed Steps for Voter Registration

• Voter enters their name and Aadhar number in the registration form.
• The registration form sends the data to the registerVoter function in the DistributedVoting smart
contract.
• The registerVoter function checks if the voter is already registered.
• If not, it adds the voter's details to the voterList and sets their registered status to false.
• The admin verifies the voter and updates their registered status to true.

Detailed Steps for Vote Casting

• The voter selects a candidate and submits their vote.
• The vote is sent to the castVote function in the DistributedVoting smart contract.
• The castVote function checks if the voter is registered and hasn't voted yet.
• If the conditions are met, it records the vote and updates the voter's status to voted.
• The vote is recorded on the blockchain, ensuring it cannot be tampered with.

Dept. of CSE Page 17

 AY: 2022-23

Detailed Steps for Vote Tallying

• The admin triggers the vote tallying process by calling the endElection function in the
DistributedVoting smart contract.
• The endElection function changes the state of the election to ended.
• The smart contract counts the votes for each candidate and stores the results on the blockchain.
• The results are published on the frontend application, ensuring transparency and verifiability.

Security Measures

Identity Verification:
Ensures that only eligible voters can register and vote by verifying their identity using Aadhar numbers.
Anonymity:
Ensures that votes are anonymous and cannot be traced back to the voter, protecting voter privacy.
Data Integrity:
Ensures that once a vote is recorded on the blockchain, it cannot be altered or deleted, maintaining data
integrity.
Transparency:
Ensures that the entire voting process is transparent and verifiable by any participant in the blockchain
network, building trust in the system.

Dept. of CSE Page 18

 AY: 2022-23

System Implementation for Blockchain-Based E-Voting System

1. Introduction
This section describes the system implementation of a blockchain-based e-voting system. The
implementation details include the algorithms used, the modules implemented, and their respective
pseudo code. The system aims to ensure secure, transparent, and tamper-proof electronic voting.

2. Algorithms Used

Voter Registration Algorithm:

Ensures only eligible voters are registered and recorded on the blockchain.

Vote Casting Algorithm:

Allows registered voters to cast their votes securely.

Vote Tallying Algorithm:

Counts the votes and publishes the results in a transparent manner.

3. Modules Implemented
User Interface (UI) Module:
Voter Registration
Vote Casting
Viewing Election Results
Smart Contract Module:
Voter Management
Candidate Management
Vote Casting and Tallying
Blockchain Interaction Module:
Connecting to Ethereum Blockchain
Deploying and Interacting with Smart Contracts

4. Pseudo Code for Modules

4.1 User Interface (UI) Module

Voter Registration Interface:

pseudo
Copy code
function VoterRegistrationInterface()
display "Enter Name:"
input name
display "Enter Aadhar Number:"
input aadhar

call registerVoter(name, aadhar)

if registration successful

Dept. of CSE Page 19

 AY: 2022-23

display "Registration Successful"

else
display "Registration Failed"
end function
Vote Casting Interface:

pseudo
Copy code
function VoteCastingInterface()
display list of candidates

display "Select Candidate:"

input candidate

call castVote(candidate)
if vote successful
display "Vote Cast Successfully"
else
display "Vote Casting Failed"
end function
Viewing Election Results Interface:

pseudo
Copy code
function ViewResultsInterface()
call getResults()
display results
end function

4.2 Smart Contract Module

Smart Contract for Voter Management:

pseudo
Copy code
contract DistributedVoting
struct Voter
address voterAddress
string voterName
string voterAadhar
bool voterExists
bool registered
bool voted

mapping(address => Voter) voters

Voter[] voterList

function registerVoter(name, aadhar)

if voters[msg.sender].voterExists == false
Voter newVoter = Voter(msg.sender, name, aadhar, true, false, false)
voters[msg.sender] = newVoter
voterList.push(newVoter)

Dept. of CSE Page 20

 AY: 2022-23

end if
end function

function verifyVoter(voterAddress)
if onlyAdmin and voters[voterAddress].voterExists == true
voters[voterAddress].registered = true
end if
end function

function getAllVoters() returns (Voter[] memory)

return voterList
end function
end contract

Smart Contract for Candidate Management:

pseudo
Copy code
contract DistributedVoting
struct Candidate
string candidateName
bool exists
int voteCount

mapping(string => Candidate) candidates

Candidate[] candidateList

function addCandidate(name)
if candidates[name].exists == false
Candidate newCandidate = Candidate(name, true, 0)
candidates[name] = newCandidate
candidateList.push(newCandidate)
end if
end function

function getAllCandidates() returns (Candidate[] memory)

return candidateList
end function
end contract

Smart Contract for Vote Casting and Tallying:

pseudo
Copy code
contract DistributedVoting
enum State { NotStarted, Running, Ended }
State state

function castVote(candidateName)

Dept. of CSE Page 21

 AY: 2022-23

if voters[msg.sender].registered == true and voters[msg.sender].voted == false and state ==

State.Running
candidates[candidateName].voteCount += 1
voters[msg.sender].voted = true
end if
end function

function startElection()
if onlyAdmin and state == State.NotStarted
state = State.Running
end if
end function

function endElection()
if onlyAdmin and state == State.Running
state = State.Ended
end if
end function

function getResults() returns (Candidate[] memory)

return candidateList
end function
end contract
4.3 Blockchain Interaction Module
Connecting to Ethereum Blockchain:

pseudo
Copy code
function connectToBlockchain()
web3 = new Web3(window.ethereum)
networkId = web3.eth.net.getId()
contractData = DistributedVoting.networks[networkId]

if contractData exists
contract = new web3.eth.Contract(DistributedVoting.abi, contractData.address)
else
display "Contract not deployed"
end if
end function

Interacting with Smart Contracts:

pseudo
Copy code
function registerVoter(name, aadhar)
contract.methods.registerVoter(name, aadhar).send({ from: userAccount })
end function

function castVote(candidateName)
contract.methods.castVote(candidateName).send({ from: userAccount })
end function

Dept. of CSE Page 22

 AY: 2022-23

function getResults()
contract.methods.getResults().call().then(results => display results)
end function

5. Detailed Steps for Each Module

5.1 Voter Registration
User Input:
Voter enters their name and Aadhar number.

Smart Contract Call:

The input data is sent to the registerVoter function in the DistributedVoting smart contract.

Voter Storage:
The smart contract stores the voter information in the voters mapping and voterList array.

Admin Verification:
The admin verifies the voter and updates their registered status.

5.2 Vote Casting

User Input:
Voter selects a candidate from the list.

Smart Contract Call:

The selected candidate's name is sent to the castVote function in the DistributedVoting smart contract.

Vote Recording:
The smart contract increments the vote count for the selected candidate and updates the voter's status to
voted.

5.3 Vote Tallying

Election End:
The admin calls the endElection function to end the election.

Result Calculation:
The smart contract counts the votes for each candidate.

Result Publication:
The results are retrieved using the getResults function and displayed to the users.

6. Security Measures
Identity Verification:
Uses Aadhar number for voter identity verification.
Anonymity:
Ensures that votes are cast anonymously.

Dept. of CSE Page 23

 AY: 2022-23

Data Integrity:
Ensures that once a vote is cast, it cannot be altered.

Transparency:
Allows anyone to verify the election process and results.
.

Dept. of CSE Page 24

 AY: 2022-23

⚫ The prover obtains authenticated private data, such as bank representation, from a trusted source.
This data is considered private and should only be accessed by authorized entities.

⚫ The verifier, who requires certain minimum personal data from the prover, initiates a request.
This could include specific information needed for a transaction or verification process.

⚫ The prover receives the verifier's request and computes a response based on the necessary data.
The prover ensures that the computation is correct and accurate.

⚫ After computing the response, the prover constructs a proof of correct computation. This proof
demonstrates that the response provided by the prover is valid and has been derived accurately
from the private data.

⚫ Both the response and the proof are sent back to the verifier. The response contains the necessary
minimum personal data requested, while the proof assures the verifier of the correctness of the
computation.

⚫ Upon receiving the response and the proof, the verifier applies a zero-knowledge (ZK) proof
verification algorithm. This algorithm checks the validity of the proof without revealing any
sensitive information about the private data itself.

⚫ If the ZK proof verification algorithm returns a positive answer, indicating that the proof is valid,
the verifier can trust the response provided by the prover. It can be treated as if it had been given
by a trusted third party, without the verifier having any knowledge of the actual private
information.

Dept. of CSE Page 25

 AY: 2022-23

CHAPTER 5
TOOLS AND TECHNOLOGIES

o Blockchain: The study focuses on blockchain technology, which is a distributed ledger system that allows
for secure and transparent recording of transactions. The authors mention that blockchain provides
properties like distributed storage, immutable data, traceability of ownership, and functionality of smart
contracts.

o Privacy Ensuring Methods:

a. Mix Networks: Mix networks involve intermediaries or trusted third parties (TTP) that mix the addresses
of transaction parties, making it difficult for other participants to interpret the relationships between senders
and receivers.
b. Ring Signatures: Ring signature methods enable a sender to generate an electronic signature using a group
of addresses, providing a mechanism for obfuscating the exact sender. However, these methods can have
performance limitations and may be vulnerable to attackers with small ring sizes.
c. Off-chain Protocols: Off-chain protocols involve using additional services or databases outside the
blockchain for storing private data. While this enhances privacy, it diminishes the traceability of transactions
and data within the blockchain.

o Zero-Knowledge Proofs (ZKPs): The authors propose the use of ZKPs as a promising approach to ensure
privacy in corporate blockchain networks. ZKPs are cryptographic protocols that allow participants to prove
the validity of a statement without revealing any additional information. The paper discusses both interactive
and non-interactive ZKP protocols.

o Non-Interactive Zero-Knowledge Proofs (NIZK): The authors highlight the benefits of NIZK protocols for
blockchain, as they do not require multiple rounds of communication between the prover and verifier. NIZK
protocols provide completeness, soundness, and zero-knowledge properties. The paper presents the general
scheme of NIZK protocols, including the Setup, Prove, and Verify algorithms.

o Masterchain Platform: The research paper mentions the Masterchain platform, which is an Ethereum- based
platform that utilizes certified GOST crypto algorithms. The authors propose the implementation of NIZK
protocols for the Masterchain platform to address privacy requirements for trading digital financial assets
(DFAs) while maintaining anonymity and not revealing private data.

o Performance Evaluation: The authors discuss performance aspects of NIZK protocols, including the time
to generate and check proofs, proof size, transaction size, and the need for a trusted setup. They present
target business requirements for DFAs and mention conducting experiments using zk-SNARK algorithms
and the ZoKrates framework.

Dept. of CSE Page 26

 AY: 2022-23

C is a high-level, general-purpose programming language that is widely used for system

programming, embedded systems, and developing a wide range of applications. C as its standard
library, which provides a set of functions for common tasks like input/output operations, string
manipulation, mathematical computations, and memory allocation. The standard library, along with
the language's simplicity, makes C a versatile and powerful tool for software development.

Editors and executing software used

Ubuntu Ubuntu is a Linux-based operating system. It is designed for computers, smartphones,

and network servers. The system is developed by a UK based company called Canonical Ltd.
All the principles used to develop the Ubuntu software are based on the principles of Open
Source software development.
Following are some of the significant features of Ubuntu −
⚫ The desktop version of Ubuntu supports all the normal software on Windows such as Firefox.
⚫ It supports the office suite called LibreOffice.
⚫ Ubuntu has an in-built email software called Thunderbird, which gives the user access to email.
⚫ There are a host of free applications for users to view and edit photos.
⚫ There are also applications to manage videos and it also allows the users to share videos.
⚫ It is easy to find content on Ubuntu with the smart searching facility.
⚫ The best feature is, it is a free operating system and is backed by a huge open source
community.

Hardware Requirement

➢ Computational Power: NIZK protocols often involve complex mathematical computations, such
as cryptographic operations or zero-knowledge proofs. The computational power of the hardware
used should be sufficient to handle these computations efficiently. This typically includes a
powerful processor, such as a multi-core CPU or GPU, capable of handling the required
mathematical operations.

➢ Memory: Depending on the size of the datasets involved and the specific NIZK protocol, the
project may require a significant amount of memory. Ensure that the hardware has enough RAM
to accommodate the data and intermediate results during the protocol execution. Consider both
the size of the input data and the memory requirements of any cryptographic libraries or
frameworks used.

➢ Storage: If your project involves storing large amounts of data, ensure that you have sufficient
storage capacity. This includes both primary storage (e.g., hard drives or solid-state drives) for
storing input data and intermediate results, as well as backup and archival storage if necessary.

Dept. of CSE Page 27

 AY: 2022-23

➢ Networking: If your project involves distributed systems or communication between multiple

parties, you may need to consider networking capabilities. This includes reliable and fast network
connections between the participating entities to ensure efficient communication during the
execution of the NIZK protocol.

➢ Security Considerations: NIZK protocols often deal with sensitive data and cryptographic
operations. Ensure that the hardware and infrastructure used provide appropriate security
measures. This includes features like secure boot, encryption, tamper-resistant components, and
protection against side-channel attacks.

➢ Scalability: Consider the scalability requirements of your project. If you anticipate a large number
of participants or significant growth in the dataset size, ensure that your hardware can handle the
increased workload and provide sufficient resources for future expansion.

Dept. of CSE Page 28

 AY: 2022-23

Dept. of CSE Page 29

 AY: 2022-23

CHAPTER 7

SNAPSHOTS
I. Add Candidate

II. Add Voter

Dept. of CSE Page 30

 AY: 2022-23

III. Start Election

IV. Cast the Vote

Dept. of CSE Page 31

 AY: 2022-23

V. Election Result

Dept. of CSE Page 32

 AY: 2022-23

CHAPTER 8
CONCLUSION

A nation with a less voting percentage will fight to develop as choosing a right front-runner
for the nation is very essential .our future system is designed to provide a secure data and
dependable voting amongst the people of the equality . By assuming blockchain in the
distributed database on voting system one can reduce the double dealing source of database
management .This project aims to voting effect using blockchain procedure from every
place of election

Dept. of CSE Page 33

 AY: 2022-23

CHAPTER 9
REFERENCES
1. "What is a zero-knowledge proof and why is it useful?". 16 November 2017.

2. Blum, Manuel; Feldman, Paul; Micali, Silvio (1988). Non-Interactive Zero-Knowledge and
Its Applications (PDF). Proceedings of the Twentieth Annual ACM Symposium on Theory of
Computing (STOC 1988). pp. 103–112. doi:10.1145/62212.62222. ISBN 978-
0897912648. S2CID 7282320. Archived (PDF) from the original on December 14, 2018.

3. Wu, Huixin; Wang, Feng (2014). "A Survey of Noninteractive Zero Knowledge Proof System
and Its Applications".The Scientific World Journal. 2014:
560484. doi:10.1155/2014/560484. PMC 4032740. PMID 24883407.

4. Chalkias, Konstantinos. "Demonstrate how Zero-Knowledge Proofs work without using

maths". CordaCon 2017. Retrieved 2017-09-13.

5. Feige, Uriel; Fiat, Amos; Shamir, Adi (1988-06-01). "Zero-knowledge proofs of identity".
Journal of Cryptology. 1 (2): 77–94. doi:10.1007/BF02351717. ISSN 1432-
1378. S2CID 2950602.

6. Blum, Manuel (1986). "How to Prove a Theorem So No One Else Can Claim It"
(PDF). ICM Proceedings: 1444–1451. CiteSeerX 10.1.1.469.9048. Archived (PDF) from the
original on Jan 3, 2023.

7. Sahai, Amit; Vadhan, Salil (1 March 2003). "A complete problem for statistical zero
knowledge" (PDF). Journal of the ACM. 50 (2): 196–
249. CiteSeerX 10.1.1.4.3957. doi:10.1145/636865.636868. S2CID 218593855. Archived (P
DF) from the original on 2015-06-25.

Dept. of CSE Page 34