Table 1.

Quantitative comparisons.

	NJUD datasets				NLPR datasets				STERE datasets				SIP datasets
	MAE	FM	SM	EM	MAE	FM	SM	EM	MAE	FM	SM	EM	MAE	FM	SM	EM
PCF	0.059	0.872	0.877	0.924	0.044	0.841	0.874	0.925	0.064	0.86	0.875	0.925	0.071	0.838	0.842	0.901
MMCI	0.079	0.852	0.858	0.915	0.059	0.815	0.856	0.913	0.068	0.863	0.873	0.927	0.086	0.818	0.833	0.897
CPFP	0.053	0.877	0.879	0.926	0.036	0.867	0.888	0.932	0.051	0.874	0.879	0.925	0.064	0.851	0.85	0.903
DMRA	0.051	0.886	0.886	0.927	0.031	0.879	0.899	0.947	0.066	0.847	0.835	0.911	0.085	0.821	0.806	0.875
D3Net	0.041	0.9	0.9	0.95	0.025	0.897	0.912	0.953	0.046	0.891	0.899	0.938	0.063	0.861	0.86	0.909
UCNet	0.043	0.895	0.897	0.936	0.025	0.903	0.92	0.956	0.039	0.899	0.903	0.944	0.051	0.879	0.875	0.919
SSF	0.043	0.896	0.899	0.935	0.026	0.896	0.914	0.953	0.044	0.89	0.893	0.936	0.053	0.88	0.874	0.921
S2MA	0.053	0.889	0.894	0.93	0.03	0.902	0.915	0.95	0.051	0.882	0.89	0.932	0.054	0.884	0.878	0.92
CoNET	0.047	0.892	0.895	0.937	0.031	0.887	0.908	0.945	0.04	0.904	0.908	0.948	0.063	0.867	0.858	0.913
cmMS	0.044	0.897	0.9	0.936	0.027	0.896	0.915	0.949	0.042	0.891	0.895	0.937	0.061	0.871	0.867	0.091
DisenFuse	0.052	0.897	0.889	0.914	0.035	0.895	0.9	0.933	0.054	0.887	0.883	0.915	0.068	0.866	0.859	0.899
ICNet	0.051	0.903	0.895	0.901	0.028	0.919	0.922	0.945	0.054	0.897	0.891	0.911	0.063	0.882	0.864	0.903
CMWNet	0.046	0.913	0.903	0.923	0.029	0.913	0.917	0.941	0.043	0.911	0.905	0.93	0.062	0.89	0.867	0.909
BBSNet	0.039	0.926	0.916	0.937	0.026	0.921	0.923	0.948	0.046	0.901	0.896	0.928	0.056	0.892	0.874	0.912
CDNet	0.038	0.919	0.913	0.94	0.024	0.925	0.93	0.954	0.041	0.909	0.903	0.938	0.06	0.888	0.862	0.905
DCF2	0.038	0.917	0.903	0.941	0.023	0.917	0.921	0.956	0.037	0.915	0.905	0.943	0.052	0.9	0.873	0.921
DANet	0.048	0.88	0.891	0.932	0.029	0.903	0.915	0.953	0.048	0.881	0.892	0.93	0.054	0.884	0.878	0.92
A2dele	0.052	0.872	0.868	0.914	0.031	0.875	0.89	0.937	0.043	0.885	0.885	0.935	0.07	0.834	0.829	0.889
PGAR	0.045	0.905	0.906	0.94	0.028	0.898	0.918	0.948	0.044	0.893	0.903	0.936	0.059	0.877	0.875	0.914
DFM-Net	0.042	0.91	0.906	0.947	0.026	0.908	0.923	0.957	0.045	0.893	0.898	0.941	0.051	0.887	0.883	0.926
DSA2F	0.039	0.917	0.904	0.937	0.024	0.916	0.918	0.952	0.039	0.91	0.897	0.942	0.057	0.891	0.862	0.911
HAINet	0.038	0.92	0.909	0.931	0.025	0.917	0.921	0.952	0.038	0.919	0.909	0.938	0.048	0.916	0.886	0.925
SSL	0.038	0.923	0.909	0.939	0.025	0.923	0.922	0.96	0.039	0.914	0.904	0.939	0.046	0.909	0.888	0.927
Ours	0.039	0.914	0.902	0.936	0.022	0.922	0.924	0.958	0.036	0.914	0.905	0.941	0.046	0.904	0.886	0.926
Rank	5	7	10	10	1	3	2	2	1	3	3	5	1	3	2	2

	TOP3	Average metrics
		MAE	FM	SM	EM
PCF	0/16	0.060	0.853	0.867	0.919
MMCI	0/16	0.073	0.837	0.855	0.913
CPFP	0/16	0.051	0.867	0.874	0.922
DMRA	0/16	0.058	0.858	0.857	0.915
D3Net	1/16	0.044	0.887	0.893	0.938
UCNet	1/16	0.040	0.894	0.899	0.939
SSF	0/16	0.042	0.891	0.895	0.936
S2MA	0/16	0.047	0.889	0.894	0.933
CoNET	2/16	0.045	0.888	0.892	0.936
cmMS	0/16	0.044	0.889	0.894	0.728
DisenFuse	0/16	0.052	0.886	0.883	0.915
ICNet	0/16	0.049	0.900	0.893	0.915
CMWNet	1/16	0.045	0.907	0.898	0.926
BBSNet	3/16	0.042	0.910	0.902	0.931
CDNet	5/16	0.041	0.910	0.902	0.934
DCF2	7/16	0.038	0.912	0.901	0.940
DANet	0/16	0.045	0.887	0.894	0.934
A2dele	0/16	0.049	0.867	0.868	0.919
PGAR	0/16	0.044	0.893	0.901	0.935
DFM-Net	4/16	0.041	0.900	0.903	0.943
DSA2F	1/16	0.040	0.909	0.895	0.936
HAINet	9/16	0.037	0.918	0.906	0.937
SSL	10/16	0.037	0.917	0.906	0.941
Ours	11/16	0.036	0.914	0.904	0.940
Rank	1	1	3	3	3

For MAE, the lower, the better. On the contrary, for FM, SM and EM, the higher, the better. The second row from the bottom refers to the evaluated scores of our proposed method and the last row refers to the rank of our proposed method.