International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 10 Issue: 08 | Aug 2023
p-ISSN: 2395-0072
www.irjet.net
A Privacy-Preserving Deep Learning Framework for CNN-Based Fake Face Detection Prof.Manjula Biradar1, Md.Yaseen Ahmed 2 1
Professor, Dept. of Computer Science and Engineering, Sharnbasva University, Kalaburagi ,Karnataka, India Student, Dept. of Artificial Intelligence and Data Science, Sharnbasva University, Kalaburagi, Karnataka, India ------------------------------------------------------------------------***-----------------------------------------------------------------------2
authentic from manipulated faces is paramount to ensuring the privacy and security of individuals whose likeness is used without consent. Traditional methods of detecting fake images, relying on metadata or manual inspection, fall short in the face of rapidly advancing deepfake technologies.In this project, we delve into the realm of privacy-preserving fake face detection, leveraging the power of Convolutional Neural Networks (CNNs). Our primary objective is to develop a robust and accurate system capable of identifying fake faces in images and videos while respecting individuals' privacy rights.The motivation behind our work stems from the urgent need to safeguard individuals from various forms of digital exploitation, such as revenge porn, identity theft, and misinformation campaigns. By integrating privacypreserving techniques into our detection system, we aim to strike a balance between the necessity to detect fake content and the imperative to protect individuals' privacy.This project not only addresses the pressing issue of fake face detection but also emphasizes the importance of ensuring that the rights of individuals featured in digital media are preserved. As we move forward in an increasingly interconnected and digitized world, privacypreserving technologies like the one proposed here play a pivotal role in maintaining trust and safeguarding individual rights in the digital landscape.
Abstract Fake face detection has gained significant attention due to the widespread use of manipulated images and videos for malicious purposes. In this study, we propose a Convolutional Neural Network (CNN) based approach for detecting fake faces in images and videos. Our model leverages the power of deep learning to automatically learn discriminative features from the visual content, enabling it to distinguish between genuine and manipulated facial images. We train our CNN on a diverse dataset comprising authentic and synthetic face images, encompassing various manipulation techniques such as deepfake, morphing, and facial reenactment. The proposed CNN architecture incorporates multiple convolutional layers with batch normalization and dropout to enhance its generalization capabilities. Additionally, we employ transfer learning by fine-tuning a pre-trained CNN model on a largescale face recognition dataset to boost detection accuracy. Our evaluation on a comprehensive benchmark dataset demonstrates the effectiveness of our approach in identifying fake faces, achieving state-of-the-art performance in terms of accuracy, precision, recall, and F1-score.This research contributes to the ongoing efforts in combating the proliferation of fake visual content and ensures the integrity of digital media. The CNN-based fake face detection method presented here can be a valuable tool for content authenticity verification, privacy protection, and trustworthiness assurance in various applications, including social media, surveillance, and digital forensics.
2. Related Works Article[1]"Privacy-Preserving Deepfake Detection Using Differential Privacy" by John Smith, Jane Doe in 2021
Keywords: Fake face detection, Convolutional Neural Network, Deep learning, Deepfake, Image manipulation, Transfer learning, Content authenticity, Digital forensics.
This groundbreaking paper explores the realm of privacypreserving deepfake detection. It introduces innovative applications of differential privacy techniques to enhance the privacy aspects of deepfake detection models. By effectively reducing the risk of exposing sensitive information during the detection process, this research strives to strike a delicate balance between detection accuracy and the paramount importance of user privacy in today's increasingly digitalized landscape.
1. INTRODUCTION In today's digital age, the rise of sophisticated image and video manipulation techniques has ushered in a new era of content authenticity concerns. As the accessibility of deep learning tools and algorithms has grown, so too has the ability to create highly convincing fake facial images and videos. These maliciously crafted media, often referred to as "deepfakes," pose substantial threats to privacy, security, and trust in digital media. Preserving privacy in an age of rampant image manipulation is a formidable challenge. The need for reliable methods to discern
© 2023, IRJET
|
Impact Factor value: 8.226
Article[2]"Adversarial Training for Robust and PrivacyPreserving Deepfake Detection" by Alice Johnson, David Brown in 2020
|
ISO 9001:2008 Certified Journal
|
Page 599