AI vs. AI: Deep Learning Approaches for Detecting Deepfake Content
Sarthak Kulkarni, Dinesh Kumar Vishwakarma, Virender Ranga · 2024
The proliferation of deepfake content has raised significant concerns due to its potential for misuse in spreading misinformation and undermining trust in digital media. This review paper explores the implementation of deep learning techniques during deepfake detection regarding the effectiveness of different algorithms and models regarding synthetic media generation using advanced AI methods. The paper is an in-depth analysis of some key benchmark datasets among the likes of FakeAVCeleb, FaceForensics++, DFDC, FaceForensics, and LAV-DF, essential for training and evaluation of deepfake detection techniques. This paper compares the strengths and weaknesses of these models differently in different scenarios using FACTOR, EfficientNetB4 + EfficientNetB4ST + B4Att + B4AttST, Cross-Efficient Vision Transformation, XceptionNet, BA-TFD deep learning models over these datasets. However, it deals directly with the current improvements in the field that tackle the varying and growing challenges we face due to increasingly sophisticated generative technologies. It is through this comprehensive review that this paper will provide better insights into deepfake detection mechanisms and the contributions made toward the development of further robust and scalable solutions toward the threats imparted by deepfake content.