. 2025 Jan 2;15:397. doi: 10.1038/s41598-024-84680-x

Table 1.

Quantitative comparison on three complex datasets: HKU-IS³⁶, ECSSD³⁷ and DUTS-TE³⁸.

Method	HKU-IS					ECSSD					DUTS-TE
Method
PiCANet	0.904	0.840	0.043	0.919	0.950	0.917	0.867	0.047	0.935	0.952	0.869	0.756	0.051	0.860	0.920
BASNet	0.909	0.889	0.032	0.928	0.952	0.916	0.904	0.037	0.942	0.951	0.866	0.803	0.047	0.860	0.903
SCRN	0.916	0.876	0.034	0.934	0.956	0.927	0.899	0.038	0.950	0.037	0.885	0.803	0.040	0.888	0.925
LDF	0.919	0.904	0.027	0.939	0.958	0.924	0.915	0.034	0.950	0.954	0.892	0.845	0.034	0.898	0.930
U2Net	0.916	0.890	0.031	0.935	0.954	0.928	0.910	0.033	0.951	0.957	0.861	0.804	0.044	0.873	0.911
ITSD	0.917	0.894	0.031	0.934	0.960	0.925	0.911	0.035	0.947	0.959	0.885	0.824	0.041	0.883	0.929
CTDNet	0.922	0.909	0.027	0.941	0.961	0.925	0.915	0.032	0.950	0.956	0.893	0.847	0.034	0.897	0.935
VST	0.928	0.897	0.030	0.937	0.968	0.932	0.910	0.034	0.944	0.964	0.896	0.828	0.037	0.877	0.939
RCSB	0.919	0.909	0.027	0.938	0.959	0.922	0.916	0.033	0.944	0.955	0.881	0.840	0.034	0.889	0.925
EDN	0.921	0.900	0.029	0.938	0.959	0.928	0.915	0.034	0.948	0.959	0.883	0.822	0.041	0.881	0.922
ICON	0.915	0.895	0.032	0.935	0.957	0.919	0.905	0.036	0.945	0.953	0.878	0.822	0.043	0.883	0.924
OLER	0.920	0.911	0.042	0.940	0.960	0.927	0.924	0.030	0.953	0.959	0.889	0.852	0.0332	0.896	0.934
MENet	0.927	0.917	0.023	0.948	0.965	0.928	0.920	0.031	0.955	0.956	0.905	0.870	0.028	0.912	0.944
ELSANet	0.923	0.916	0.025	0.935	0.963	0.929	0.926	0.030	0.943	0.960	0.893	0.856	0.034	0.882	0.934
EMSNet	0.921	0.912	0.025	0.940	0.960	0.926	0.921	0.031	0.948	0.955	0.893	0.892	0.031	0.896	0.933
DC-Net	0.924	0.909	0.027	0.942	0.963	0.924	0.913	0.034	0.949	0.953	0.896	0.852	0.035	0.899	0.935
CANet	0.920	0.939	0.027	0.941	0.961	0.928	0.951	0.032	0.950	0.958	0.895	0.898	0.033	0.899	0.937
Ours	0.934	0.920	0.023	0.948	0.972	0.939	0.931	0.026	0.957	0.969	0.907	0.866	0.030	0.904	0.948

For Inline graphic and , higher and lower scores indicate better results, respectively. denotes max E-measure, denotes max F-measure. The best and second-best results are shown in bold and italics, respectively. The symbol ’–’ indicates the results of the model are unavailable.