Table 2.

Quantitative results on different SOD datasets

		ECSSD [54]				HKU-IS [25]				DUT-OMRON [35]				DUT-TE [47]
No.	model	Eφ ↑	Sα ↑	$F_w^{\beta }$ ↑	M ↓	Eφ ↑	Sα ↑	$F_w^{\beta }$ ↑	M ↓	Eφ ↑	Sα ↑	$F_w^{\beta }$ ↑	M ↓	Eφ ↑	Sα ↑	$F_w^{\beta }$ ↑	M ↓
1	BAN [39]		0.924	0.923	0.035	0.950	0.913	0.900	0.032	0.892	0.879	0.815	0.040	0.860	0.832	0.746	0.059
2	CPD [51]	0.925	0.918	0.917	0.037	0.944	0.905	0.891	0.034	0.887	0.869	0.805	0.043	0.866	0.825	0.747	0.056
3	EGNet [64]			0.920	0.037	0.948	0.916	0.898	0.032	0.891		0.815	0.039	0.867
4	SCRN [10]	0.926		0.918	0.037	0.949	0.916	0.896	0.034	0.888	0.885	0.809	0.040	0.863	0.837	0.746	0.056
5	F3Net [48]		0.924				0.917								0.838	0.766	0.053
6	MINet [37]			0.924				0.909		0.898	0.884	0.828		0.865	0.833	0.755	0.056
7	ITSD [68]			0.895			0.917	0.899	0.031	0.895	0.885	0.804	0.041	0.863	0.840		0.061
8	Ours	0.924									0.886		0.038

The best two results are highlighted in and respectively. The training set is DUTS-TR [47] for all the methods