A Computer Vision-Based Approach To Detect Fake Images

Abstract

Image forensics aims to detect the manipulation of digital images. These days, user friendly image editing tools, like Adobe Photoshop, have made it easy for anyone to manipulate images. Image editing techniques develop over time. These techniques may also be intentionally applied to certain images to maliciously manipulate them to convince the users that those are real images. These techniques are in wide use in areas that play a central role in our everyday lives like in politics, social networking sites, eCommerce websites and in the sales of expensive artworks. Currently, splicing detection, copy-move detection, contrast enhancement detection, facial morphing detection, fake colorization ,and image inpainting detection are attracting significant attention from researchers. There is currently, no single system that can detect these specific forgeries. This motivates the need for a publicly available application that can verify the integrity of images obtained anywhere on the internet.We have developed a standalone web application that takes in an image as its input and performs the detection techniques on the image and provides the users with a simple yet informative output on whether the image is digitally manipulated or not and also shows the specific regions of forgery.