Detection Facial Forensics Based on Deep learning approaches: Comprehensive Literature Review

Abstract

Facial Forensics (FF) is one of the most active research problems in computer vision and digital image forensics, with a wide variety of practical and commercial applications, including identification, access control, and interaction between people with smart devices. However, identifying a face raises serious questions about individual liberties and raises ethical issues. In recent years, important methods, algorithms, approaches, and databases have been proposed for research FF without constraints. There are two approaches namely: the 2D approach has reached a certain level of maturity and has reported very high recognition rates; the 3D approach has been helping to reduce such ambient conditions. It has been proposed as an alternative to the problems mentioned above. The advantage of the 3D dataset is that it is invariant to the pose and lighting conditions, which improves the efficiency of the detection systems. However, the 3D dataset is a bit sensitive to changes in facial expressions. This paper presents FF technologies, currently advanced methods, and future directions. It focuses specifically on the latest database of 2D and 3D facial forensics that utilized for last five years. Furthermore, it focuses on deep learning (DL) approaches due to excellent performances. Also, potential directions for research in facial forensics are presented to provide the reader with a point of reference for topics worthy of consideration in the facial forensic field.