Abbreviated View of Deepfake Videos Detection Techniques
Mohammed Akram Younus, Taha Mohammed Hasan · 2020
In the era of technological advances and a qualitative breakthrough in the artificial intelligence field and deep neural networks, a new age of hyper-realistic digital videos forgery called DeepFake has been born, with that new technology, it is difficult to distinguish between real videos and fake ones which are uploaded daily on various websites across the Internet. Many open-source DeepFake creation methods have risen, leading to a growing number of synthesized media clips over the internet. There are many efficient fast methods and techniques which have been designed to detect and spot such phenomenon. Background Comparison, Temporal Pattern Analysis, Eye blinking, Facial Artifacts, Mesoscopic Analysis, and Pose Estimation are some of those techniques. Some of these approaches designed to detect and identify the video forgery without any prior enlightenment concerning the videos under analysis. The primary scope of this study is to provide an abbreviate review of these methodologies of a method-comparison study that has been presented to assist the researcher's evaluation of such studies.