Image tampering detection & Localization tool

PEC APPLICATION CONTENT

SEPTEMBER 15, 2023

University of engineering & technology lahore, narowal campus
Department of CS&E
 1

TABLE OF CONTENTS
Abstract................................................................................................................................................2
Outcomes..............................................................................................................................................2
Estimate For Project Proposal............................................................................................................3
Name.....................................................................................................................................................3
Innovations...........................................................................................................................................4
1. Improving Accuracy..................................................................................................................4
2. Tool Development.....................................................................................................................4
3. Providing Plug-Ins.....................................................................................................................4
4. First Proposed Solution in Pakistan............................................................................................5
References............................................................................................................................................5
 2

Abstract
The alarming rise in image tampering and manipulation is a result of the quick adoption of
digital images across various industries, from journalism to e-commerce. Modern, user-
friendly tools make it simple to alter image material for unfair advantage or to spread false
information [1], and digital photographs and videos are not admissible as evidence in courts
of law without trustworthy forensic analysis. For digital information to remain accurate and
for trust to exist in the digital world, image tampering must be addressed. In this regard,
there are two common types of image tampering: copy-move and splicing.
This project offers a comprehensive solution to the issues that image tampering poses on
both the industrial and social fronts. The key objectives of this project include feature
extraction using Image Processing and Convolutional Neural Networks (CNNs), model
training using machine learning methods, and the development of a user-friendly tool with a
Python-based trained model API for real-time image tampering detection and localization
with high accuracy and versatile scope. Additionally, a flexible plugin will be developed to
enable enterprises to easily incorporate this solution into their current systems, thereby
enhancing their digital forensics capabilities.

Outcomes
Here are some potential outcomes of this project:

 Project Objective: Develop an advanced image forgery detection system surpassing

existing tools in accuracy and reliability.
 Advanced Techniques: Utilize state-of-the-art machine learning and computer vision
techniques for improved detection capabilities.
 User-Centric Approach: Prioritize user-friendliness with an intuitive website, mobile
app, and a user-friendly interface.
 Versatile Integration: Create a plugin for seamless integration with websites and
applications, making forgery detection accessible across diverse domains.
 Social Media Impact: Address concerns related to political manipulation, fake news,
and misinformation on social media by detecting forgeries in images related to
politics and news.
 Enhanced Transparency: Contribute to transparency and trust in the digital
landscape by enabling users to verify the authenticity of images in political and news
contexts.
 Forensic Applications: Empower forensic investigators with enhanced forgery
detection capabilities.
 Government Use: Aid government agencies in combatting fraud and ensuring
document integrity, such as detecting tampering in crucial documents like WAPDA
bills.
 Comprehensive Solution: Deliver a comprehensive, accessible, and highly accurate
image forgery detection tool with real-world applicability.
 Ethical Considerations: Consider ethical implications and responsible use of the tool,
especially in sensitive areas like politics and government documents.
 3

By addressing these points, your project aims to make a significant impact in the fields of
digital forensics, misinformation detection, and document integrity verification, contributing
to a more transparent and trustworthy digital environment.

Estimate For Project Proposal

Name Cost Justification

DSLR 100000 The accuracy and effectiveness of our tampering
PKR detection system depend on the quality of the images it
analyzes. A DSLR camera is renowned for its superior
image capture capabilities, producing high- resolution and
high-quality images. These high-quality images serve as
the foundation for our analysis and detection algorithms.
(Nikon D5100 DSLR CAMERA)
TPU 10000 The sheer complexity and computational demands of image
PKR/ analysis and tampering detection require specialized
YEAR hardware acceleration. TPUs are purpose-built for machine
learning tasks, offering significantly faster processing speeds
compared to traditional CPUs or GPUs. This ensures that our
tampering detection algorithms can run efficiently and
swiftly, allowing for real-time or near-real-time analysis of
images.(Cloud TPU v5e)
PYTHON SERVER 5000 A server is crucial for our Image Tampering Detection
PKR/ System due to its essential roles in data storage, processing
MONTH power, and continuous operation. It provides the necessary
infrastructure to manage and analyze large image datasets
efficiently, ensuring the system's reliability and real-time
functionality.

PLAYSTORE APP 5000 We are deploying our Image Tampering Detection System
DEPLOYMENT PKR/ app on the Play Store to make our innovative solution
MONTH accessible to a wider audience. This platform offers a user-
friendly and widely recognized distribution channel,
ensuring that our project's outcomes can benefit a broader
user base while facilitating easy updates and maintenance

PRINTER 40000 In our Image Tampering Detection System FYP, a printer is a

PKR vital tool to enhance the practicality and effectiveness of our
project. It enables us to produce physical copies of
potentially tampered images for detailed examination,
ensuring that subtle alterations are not overlooked.
Moreover, printed documentation and visual aids provide
tangible records that aid in project evaluation, user
interaction, and academic presentations, contributing
significantly to the overall success of our project.(HP LASER
Jet M111a Mono Printer)
WEBSITE 5000 Our website serves as a valuable online hub for our
HOSTING PKR/ community, providing easy access to essential information,
MONTH resources, and updates. It's a convenient platform for
communication, engagement, and sharing knowledge,
enhancing our online presence, and fostering meaningful
connections with our audience.
 4

Innovations
1. Improving Accuracy
Accuracy of a Machine Learning model plays an important role in the desired
outcome. Especially in the supervised machine learning model. As we are dealing
with detection of tempered images, hence we are dealing with a dataset which
included a large no of images. Tempered Image Detection includes variants of
datasets and models with accuracies near to 97% but when those models were
tested in real time environment their accuracies dropped down to 57% [2]. The
accuracy which has been achieved till now is 50% on 500 images and maximum 65%
on 1000 images. But in this project, we are going to achieve more than 90%[2]. In
this way we are going to achieve higher accuracy which will lead to innovation in our
project.
We are providing a single solution for both forgery detection domains i.e.
Copy/Move and Splicing. In this way we are going to provide a hybrid solution
for our users.

2. Tool Development

This tool will be available for users to use our services. It will have following benefits:
 Accessibility
This tool makes our project accessible to a wider audience. Users can easily
download and use this tool on their smartphones or computers, allowing for broader
adoption. This will be huge innovation for users.
 User-Friendly Interface
This tool will provide a user-friendly interface that will help our users to use it easily
there will not be any requirement for a user to have technical expertise. Even a lay
man can use it. Moreover, User friendly interface attracts more users and will
increase the usability.
 Convenience
Users can use the proposed tool conveniently whenever they need to detect and
localize tempered images. This can be especially valuable for individuals or
organizations that frequently deal with image integrity and security.
 Monetization
Depending on our tool audience and features, we could monetize this tool through
various means, such as selling it on app stores or offering premium features through
in-app purchases and services.
 Showcasing Innovation
By creating this exceptional tool and showcasing it, we will demonstrate innovation
and practical application of our tempered image detection and localization
technology, which can be valuable for academic and professional recognition.
 5

3. Providing Plug-Ins

Providing plugins or integrations for websites to use can offer several benefits, both for your
project and for website owners and developers who integrate your tools. A plug-in, which
can also be called an add-on or an extension, is third-party software that adds new functions
to a host program on a computer, without altering the host program. Here are some
advantages:

 Increased Visibility: Plugins can increase the visibility of our tool. When website
owners integrate our tool, it becomes accessible to their user base, potentially
exposing THIS project to a larger audience.
 Higher Adoption Rate: By providing a plug-in for our tool to be used on different
websites we will get an enhanced adoption rate. In this way our tool will be
accessible to a vast number of users.
 Data Collection and Feedback: Integrations can provide valuable feedback and data.
Website owners may provide insights into how our tool performs in various real-
world scenarios, helping us refine and improve our technology. Moreover, when
users use our tool on their websites using our plug-ins, we will get more and more
data and our model will train on that data as well, which will improve the accuracy of
our model. Moreover, we will get to know about user’s feedback which will help us
to find flaws in our tool and eventually we will solve those defects.
 User Engagement: Website owners may integrate our tool to enhance user
engagement and trust. For example, e-commerce platforms may use our image
tampering detection tool to verify product images, improving the shopping
experience.

4. First Proposed Solution in Pakistan

Our solution stands out not only for its effectiveness but also for being the first of its kind to
originate in Pakistan. This means that our algorithms, features, and datasets have been
tailored to the specific needs of our country, ensuring a more accuracy. approach to image
tampering detection

By pioneering this technology within Pakistan, we aim to empower law enforcement

agencies, and businesses with a reliable tool to safeguard the integrity of their visual
content. This not only enhances security but also bolsters trust in digital media across the
nation.

References

1. Zheng, L., Y. Zhang, and V.L. Thing, A survey on image tampering and its detection in real-
world photos. Journal of Visual Communication and Image Representation, 2019. 58: p. 380-
399.
2. Abdulqader, M.F., A.Y. Dawod, and A.Z. Ablahd, Detection of tamper forgery image in security
digital mage. Measurement: Sensors, 2023. 27: p. 100746.
6