Multistrengthening Module-Based Salient Object Detection

Table 2

Quantitative evaluation on five datasets.

Method			ECSSD						HKU-IS
			1,000 images						4,447 mages
			MAE				Sm		MAE				Sm
BASNet			0.037		0.88		0.921		0.032		0.895		0.946
F3Net			0.033		0.925		0.927		0.028		0.91		0.953
GateNet			0.041		0.941		0.917		0.035		0.928		0.909
ITSD			0.04		0.939		—		0.035		0.927		—
LDF			0.034		0.93		0.925		0.027		0.914		0.954
MINet			0.036		0.943		0.947		0.03		0.932		0.955
AFNet			0.042		0.935		0.914		0.036		0.923		0.905
EGNet			0.044		0.941		0.913		0.034		0.929		0.91
PoolNet			0.038		0.945		—		0.03		0.935		—
Ours			0.028		0.985		0.993		0.012		0.968		0.987
Method	DUT-OMRON					DUTS-TE				PASCAL-S
	5,168 images					5,019 images				850 images
	MAE			Sm		MAE			Sm	MAE				Sm
BASNet	0.056	0.756		0.869		0.048		0.791	0.884	0.076		0.775		0.847
F3Net	0.053	0.766		0.87		0.035		0.84	0.902	0.062		0.84		0.859
GateNet	0.061	0.794		0.82		0.045		0.87	0.869	0.07		0.882		0.855
ITSD	0.063	0.813		—		0.042		0.877	—	—		—		—
LDF	0.051	0.773		0.873		0.034		0.855	0.91	0.06		0.848		0.865
MINet	0.057	0.794		0.864		0.039		0.877	0.912	0.065		0.882		0.898
AFNet	0.057	0.797		0.826		0.046		0.862	0.866	0.071		0.868		0.85
EGNet	0.056	0.826		0.813		0.043		0.88	0.866	0.076		0.863		0.848
PoolNet	0.053	0.833		—		0.036		0.892	—	0.065		0.88		—
Ours	0.014	0.942		0.976		0.013		0.945	0.981	0.069		0.969		0.983