Fake image detection through automatic fact checking

• Main Author: See-To, Junwei
• Other Authors: Mao Kezhi
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2023
• Subjects: Engineering::Electrical and electronic engineering
• Online Access: https://hdl.handle.net/10356/166830
• Institution: Nanyang Technological University

With the development of sophisticated image manipulation techniques, the prevalence of fake images in today's digital world is becoming a major concern. Researchers have investigated various methods for detecting fake images in response, with machine learning algorithms yielding promising results. This study investigates the efficacy of using machine learning to detect fake images generated by various image manipulation methods. The author aims to use Convolutional Neural Networks (CNNs), Vision Transformers (ViT) and Generative Adversarial Networks (GANs) to analyze a diverse dataset of manipulated images generated using various techniques such as image splicing, copy-move forgery, and deep learning-based image synthesis. The developed networks would be trained using supervised learning techniques, and their performance in detecting manipulated images would be evaluated. This project emphasizes the importance of detecting fake images using various methods, focusing on experimenting with various neural network architectures and training techniques to detect fake images generated by various types of image manipulation methods. It lays the groundwork for future research into developing more robust and accurate image forensic machine learning models.