Table 5. Performance comparison of the proposed methods with different datasets in terms of the error detection rate.

DF detection methods	Datasets
	FFHQ	100K-Faces	Celeb-DF	WildDeepfake
VGG19	13.11	18.32	12.3	11.4
ResNet	13.4	15.2	12.1	11.2
MobileNet	11.2	14.22	11.65	9.21
Proposed SAE-GLSTM-CDGCNN	5.91	6.01	7.12	5.1