Extensive Analysis of Deep Learning-based Deepfake Video Detection

Deepfake is the practice of replacing an existing image or video with someone else’s likeness. Currently, the spread of face-swapping deepfake strategies is increasing, producing a considerable range of naturalistic fake videos that cause danger to everyone. Due to the issue made by deepfake, recognizing between real and fake videos becomes a major issue. Deep learning is an efficient and useful approach for detecting deepfake videos and images. Research in recent years has focused on understanding how deepfake works and a number of deep learning-based approaches that are developed to do so. The aim of this study is to give an in-depth look at how several architectures work, along with the challenges encountered when detecting deepfake videos. It examines the existing deepfake detection methods using machine learning and deep learning approaches.