Deep learning techniques for deep fake identification: A review

Abstract

Deep learning has been a hugely successful approach which is utilized in a wide diversity of domains such as Natural Language Processing, Computer Vision, and Machine Learning One of the most controversial applications of Deep Learning is the creation of Deep fakes-synthetically generated videos that closely mimic real individuals, often making the transition from genuine to fabricated footage virtually undetectable to the human eye. To deal with Deep fakes, numerous deep learning-based techniques have been developed to detect such manipulations. This paper presents a comprehensive analysis of deep fake generation and detection systems, emphasizing learning-based approaches. The study provides an in-depth review of the latest detection methods, highlighting their respective strengths and limitations. Furthermore, we examine state-of-the-art techniques used to identify Deep fakes in social media content, positioning our analysis as a valuable resource for academic research. By detailing the most recent methodologies and datasets, this work facilitates meaningful comparisons with prior studies and advances understanding in the rapidly evolving field of deep fake detection.