Digital Image Forgery Detection

This project proposes an image forgery detection method using CNN, capable of delivering high accuracy and a clear explanation for each forged instance. In the past few years, image forgery has increased drastically, owing to the easy availability of image editing tools and techniques, including morphing. To mitigate this increasing menace and mitigate the effect of tampered content, the current project presents a framework for detecting digital image forgery, employing Convolutional Neural Networks (CNN) in conjunction with generative AI tools. The proposed framework classifies images as either forged or original and gives the reason for its classification using Google Gemini, which is interfaced through a Flask-based application. Unlike conventional detection methods, this method not only yields high accuracy reaching 96% on the dataset tested but also increases interpretability by giving the reason behind forgery predictions. This solution is intended to help users better recognize tampered images, thereby enhancing trust in digital content.