Image Forgery Detection using Deep Learning: A Survey
Zankhana J. Barad, Mukesh M. Goswami
Abstract
The information is shared in form of images through newspapers, magazines, internet, or scientific journals. Due to software like Photoshop, GIMP, and Coral Draw, it becomes very hard to differentiate between original image and tampered image. Traditional methods for image forgery detection mostly use handcrafted features. The problem with the traditional approaches of detection of image tampering is that most of the methods can identify a specific type of tampering by identifying a certain features in image. Nowadays, deep learning methods are used for image tampering detection. These methods reported better accuracy than traditional methods because of their capability of extracting complex features from image. In this paper, we present a detailed survey of deep learning based techniques for image forgery detection, outcomes of survey in form of analysis and findings, and details of publically available image forgery datasets.