Detection of DeepFakes Using Local Features and Convolutional Neural Network

Abstract

Fake photos and films are nothing unique. Humans have already been manufacturing counterfeit to fool or amuse. This trend has only grown with the broad acceptance on the web. But today, instead of photographs getting modified by modifying programs such as Photoshop or films being skillfully altered, there's a new generation of machine-made fakes, so they may make it more difficult for humans to differentiate reality from imagination in the future. DeepFakes seem to be the most common type of "synthetic media," which consists of pictures, audio, and film that look to be being made using traditional methods but were produced using advanced software. In this chapter, we have examined some articles to understand what DF is and its applications. Also explained are the generation techniques (autoencoder and GANs (GANs)) and the detection techniques (CNN and local features).