Deepfake Detection Using EfficientNet and XceptionNet
Basma Yasser, Jumana Hani, Salma Elgayar, Omar Amgad, Nourhan Ahmed, Hala M. Ebied, Habiba Amr, Mohamed Salah
Abstract
The increasing prevalence of manipulated media, particularly deepfake videos, poses significant challenges in distinguishing real from fake content. This paper addresses the issue of detecting deepfake videos using advanced CNN architectures such as EfficientNet-B4 and XceptionNet. The FF++ and Celeb-DF (v2) datasets are used to compare real and fake videos. The methodology involves preprocessing the Celeb- DF dataset by extracting frames and isolating faces, training the models, and evaluating their performance using log loss and Area Under the Curve (AUC) metrics. The study shows that both models are effective in accurately classifying real and fake videos and highlights the importance of continuously updating deepfake detection algorithms in response to evolving deepfake generation techniques.