Investigation of Deep Fake Images Using Pre-Trained CNN Frameworks

Abstract

Deep learning technologies, such as neural networks, have tackled complicated issues from large-scale data processing to computer vision and human-level control. Editing photographs with software designed for commercial purposes makes it easy for anyone to manipulate images in order to create fictitious ones. Generative adversarial networks (GANs) are currently being utilised to generate superficially realistic photographs, as opposed to the old methods that were previously used to create phony images. The GANs images are referred to as deep fakes. To the unaided eye, they appear to be real photographs. As a result, it's impossible to spot a phony image produced with GANs. The social network is made less safe as a result of the uploading of these bogus photographs. It is therefore critical that the digital image's legitimacy be detected before it is uploaded. Thus, in this chapter, the authors suggest a few pre-trained deep learning frameworks that may be used to effectively detect deep fake images.