SFRNet: Feature Extraction-Fusion Steganalysis Network Based on Squeeze-and-Excitation Block and RepVgg Block

Table 3

Performance comparisons between proposed network and several state-of-the-art models on S-UNIWARD and WOW at two different payloads.

Steganography	CNN model	0.2 bbp	0.4 bpp
S-UNIWARD	Yedroudj-Net	63.0	78.1
	SRNet	67.4	81.6
	Zhu-Net	71.5	84.7
	DFSE-Net	65.9	78.5
	SFRNet	72.5	89.6
WOW	Yedroudj-Net	72.3	83.1
	SRNet	75.4	86.9
	Zhu-Net	76.7	88.2
	DFSE-Net	75.3	85.1
	SFRNet	76.8	87.9