Deepfake Detection and Classification of Images from Video: A Review of Features, Techniques, and Challenges

Abstract

The proliferation of deepfake technology poses significant challenges to the integrity and authenticity of visual content in videos, raising concerns about misinformation and deceptive practices. In this paper, we present a comprehensive review of features, techniques, and challenges related to the detection and classification of deepfake images extracted from videos. Existing literature has explored various approaches, including feature-based methods, machine learning algorithms, and deep learning techniques, to mitigate the adverse effects of deepfake content. However, challenges persist, such as the evolution of deepfake generation methods and the scarcity of diverse datasets for training detection models. To address these issues, this paper reviews related work on approaches for deepfake image detection and classification and synthesises these approaches into four categories - feature extraction, machine learning, and deep learning. The findings underscore the importance of continued research efforts in this domain to combat the harmful effects of deepfake technology on society. This study provides recommendations for future research directions, emphasizing the significance of proactive measures in mitigating the spread of manipulated visual content.