Table 3. Accuracy comparison of the proposed vs traditional baseline systems for various datasets.

Methods	Datasets
	FFHQ	100K-Faces	Celeb-DF	WildDeepfake
VGG19	84.5	74.12	88.43	89.25
ResNet	88.32	80.11	89.32	86.52
MobileNet	91.15	90.21	90.01	96.75
Proposed SAE-GLSTM-CDGCNN	96.92	97.15	98.12	98.91