Scientific Reports

Table 1 Metrics on the DUTS-TE, DUT-OMRON, HKU-IS, ECSSD, SOD, and PASCAL-S test sets were calculated.

From: SU²GE-Net: a saliency-based approach for non-specific class foreground segmentation

Model				DUTS-TE					DUT-OMRON					HKU-IS
Swin	TTA	Edge	GCT	MAE\(\downarrow\)	MaxF\(\uparrow\)	MeanF\(\uparrow\)	\(F_{\beta }\uparrow\)	S-measure\(\uparrow\)	MAE\(\downarrow\)	MaxF\(\uparrow\)	MeanF\(\uparrow\)	\(F_{\beta }\) \(\uparrow\)	S-measure\(\uparrow\)	MAE\(\downarrow\)	MaxF\(\uparrow\)	MeanF\(\uparrow\)	\(F_{\beta }\) \(\uparrow\)	S-measure\(\uparrow\)
\(\checkmark\)				0.034	0.912	0.843	0.881	0.904	0.051	0.866	0.78	0.813	0.865	0.031	0.947	0.898	0.93	0.928
\(\checkmark\)	\(\checkmark\)			0.033	0.908	0.849	0.88	0.905	0.051	0.865	0.788	0.814	0.865	0.029	0.947	0.904	0.930	0.929
\(\checkmark\)	\(\checkmark\)	\(\checkmark\)		0.033	0.908	0.849	0.88	0.904	0.051	0.863	0.787	0.812	0.864	0.029	0.947	0.905	0.93	0.929
\(\checkmark\)	\(\checkmark\)	\(\checkmark\)	\(\checkmark\)	0.032	0.912	0.855	0.883	0.906	0.050	0.865	0.790	0.815	0.864	0.028	0.948	0.907	0.931	0.930
(Ours:SU²GE-Net)				0.032	0.912	0.855	0.883	0.906	0.050	0.865	0.790	0.815	0.864	0.028	0.948	0.907	0.931	0.930
U²-Net				0.053	0.862	0.794	0.812	0.853	0.059	0.829	0.753	0.768	0.832	0.036	0.929	0.887	0.903	0.903
P2T-vgg				0.041	0.892	0.840	0.856	0.882	0.057	0.831	0.764	0.777	0.837	0.029	0.942	0.91	0.924	0.920
P2T-resnet				0.035	0.898	0.858	0.872	0.892	0.049	0.839	0.784	0.795	0.849	0.027	0.943	0.916	0.929	0.923
MSIN				0.037	0.884	0.828	0.825	0.884	0.055	0.810	0.756	0.738	0.833	0.028	0.935	0.908	0.899	0.920
SCRN				0.040	0.888	0.809	0.803	0.885	0.056	0.811	0.746	0.72	0.056	0.033	0.935	0.897	0.878	0.917
EGNet-R				0.039	0.889	0.815	0.816	0.887	0.053	0.815	0.756	0.738	0.053	0.031	0.935	0.901	0.887	0.918
BASNet				0.048	0.859	0.791	0.803	0.866	0.056	0.805	0.756	0.751	0.056	0.033	0.93	0.898	0.890	0.908
RCSB				0.035	0.889	0.840		0.881	0.049	0.809	0.752		0.835	0.027	0.938	0.909		0.919
DC-Net				0.035	0.899	0.852		0.896	0.053	0.827	0.772		0.849	0.027	0.942	0.909		0.924

Model				ECSSD					SOD					PASCAL-S
Swin	TTA	Edge	GCT	MAE\(\downarrow\)	MaxF\(\uparrow\)	MeanF\(\uparrow\)	\(F_{\beta }\) \(\uparrow\)	S-measure\(\uparrow\)	MAE\(\downarrow\)	MaxF\(\uparrow\)	MeanF\(\uparrow\)	\(F_{\beta }\) \(\uparrow\)	S-measure\(\uparrow\)	MAE\(\downarrow\)	MaxF\(\uparrow\)	MeanF\(\uparrow\)	\(F_{\beta }\) \(\uparrow\)	S-measure\(\uparrow\)
\(\checkmark\)				0.032	0.958	0.917	0.942	0.937	0.086	0.883	0.82	0.852	0.830	0.058	0.902	0.840	0.864	0.881
\(\checkmark\)	\(\checkmark\)			0.030	0.957	0.921	0.942	0.937	0.083	0.883	0.828	0.851	0.828	0.057	0.901	0.844	0.865	0.881
\(\checkmark\)	\(\checkmark\)	\(\checkmark\)		0.030	0.957	0.921	0.943	0.937	0.083	0.879	0.829	0.852	0.827	0.057	0.901	0.844	0.865	0.880
\(\checkmark\)	\(\checkmark\)	\(\checkmark\)	\(\checkmark\)	0.028	0.959	0.925	0.945	0.939	0.083	0.882	0.834	0.854	0.827	0.055	0.901	0.847	0.865	0.881
(Ours:SU²GE-Net)				0.028	0.959	0.925	0.945	0.939	0.083	0.882	0.834	0.854	0.827	0.055	0.901	0.847	0.865	0.881
U²-Net				0.041	0.947	0.907	0.922	0.915	0.119	0.859	0.772	0.785	0.770	0.084	0.865	0.797	0.808	0.829
P2T-vgg				0.034	0.956	0.925	0.938	0.928	0.102	0.871	0.808	0.818	0.797	0.065	0.888	0.837	0.848	0.860
P2T-resnet				0.032	0.953	0.927	0.938	0.927	0.098	0.871	0.818	0.828	0.799	0.062	0.887	0.845	0.855	0.864
MSIN				0.033	0.947	0.924	0.911	0.925						0.064	0.882	0.842	0.821	0.857
SCRN				0.037	0.950	0.918	0.899	0.927						0.065	0.890	0.839	0.816	0.867
EGNet-R				0.037	0.947	0.920	0.903	0.925						0.075	0.878	0.831	0.807	0.853
BASNet				0.037	0.942	0.879	0.904	0.916						0.077	0.863	0.781	0.800	0.837
RCSB				0.034	0.944	0.916		0.922						0.059	0.875	0.826		0.860
DC-Net				0.034	0.949	0.913		0.924						0.066	0.874	0.814		0.857

Higher values of MaxF, MeanF, \(F_{\beta }\), and \(S-measure\), and lower values of MAE, indicate better performance.
Optimal outcomes are highlighted in bold.

Back to article page

Search

Advanced search

Quick links