Deepfake Generation and Detection - An Exploratory Study
Diya Garg, Rupali Gill
Abstract
Deepfakes generated through algorithms based on deep learning have obtained a lot of interest recently. Deepfakes are utilized to manipulate content (audio, video and image) with high realism. Deepfakes have been influenced using artificial intelligence to make it look like someone is saying or doing something that they never actually said or did. Deepfakes can be used for malicious purposes, such as to tarnish someone's reputation or to influence public opinion. Researchers are developing new methods to detect deepfakes, as it is a challenging to distinguish between real and fake content. Deep learning is a powerful tool that is usable to develop both deepfake generation and detection methods. This article provides an extensive review of the existing state of research in creation and detection of deepfakes. It covers the diverse techniques for creation and detection, and existing benchmark datasets.