notesum.ai

Published at November 27

Deep Fourier-embedded Network for Bi-modal Salient Object Detection

cs.CV

Released Date: November 27, 2024

Authors: Pengfei Lyu¹, Xiaosheng Yu¹, Chengdong Wu¹, Jagath C. Rajapakse¹

Aff.: ¹Not specified

Arxiv: http://arxiv.org/abs/2411.18409v1

Datasets	Metric	CSRNet	MIDD	OSRNet	SwinNet	ADF	TNet	MGAI	LSNet	IFFNet	HRTransNet	CAVER	WaveNet	DFENet_m	DFENet_b
		TCSVT21	TIP21	TIM22	TCSVT22	TMM23	TMM23	TCSVT23	TIP23	TIM23	TCSVT23	TIP23	TIP23	DFENet_m	DFENet_b
VT821	$Em\uparrow$	0.908	0.895	0.896	0.926	0.842	0.919	0.913	0.911	0.918	0.929	0.929	0.929	0.934	0.934
	$Sm\uparrow$	0.885	0.871	0.875	0.904	0.810	0.899	0.891	0.878	0.907	0.906	0.898	0.912	0.910	0.912
	$wF_{\beta}\uparrow$	0.821	0.760	0.801	0.818	0.626	0.841	0.824	0.809	0.849	0.849	0.846	0.863	0.862	0.866
	$F_{\beta}\uparrow$	0.830	0.804	0.813	0.847	0.716	0.842	0.829	0.825	0.848	0.853	0.854	0.857	0.865	0.866
	$\mathcal{M}\downarrow$	0.038	0.045	0.043	0.030	0.077	0.030	0.031	0.033	0.029	0.026	0.026	0.024	0.026	0.024
VT1000	$Em\uparrow$	0.925	0.933	0.935	0.947	0.921	0.937	0.935	0.935	0.947	0.945	0.949	0.952	0.956	0.956
	$Sm\uparrow$	0.918	0.915	0.926	0.938	0.910	0.929	0.929	0.925	0.938	0.938	0.938	0.945	0.945	0.945
	$wF_{\beta}\uparrow$	0.878	0.856	0.891	0.894	0.804	0.895	0.893	0.887	0.912	0.913	0.912	0.921	0.921	0.923
	$F_{\beta}\uparrow$	0.877	0.822	0.892	0.896	0.847	0.889	0.885	0.885	0.900	0.900	0.906	0.909	0.896	0.909
	$\mathcal{M}\downarrow$	0.024	0.027	0.022	0.018	0.034	0.021	0.021	0.023	0.017	0.017	0.016	0.015	0.015	0.015
VT5000	$Em\uparrow$	0.905	0.897	0.908	0.942	0.891	0.927	0.915	0.915	0.938	0.945	0.935	0.940	0.946	0.948
	$Sm\uparrow$	0.868	0.868	0.875	0.912	0.864	0.895	0.883	0.877	0.905	0.912	0.900	0.912	0.915	0.917
	$wF_{\beta}\uparrow$	0.797	0.763	0.807	0.846	0.722	0.840	0.815	0.806	0.856	0.870	0.849	0.865	0.871	0.877
	$F_{\beta}\uparrow$	0.811	0.801	0.823	0.865	0.778	0.846	0.824	0.825	0.864	0.871	0.856	0.867	0.877	0.882
	$\mathcal{M}\downarrow$	0.042	0.043	0.040	0.026	0.048	0.033	0.034	0.037	0.028	0.025	0.028	0.026	0.025	0.024
VI-RGBT1500	$Em\uparrow$	-	0.870	0.917	0.948	-	0.943	0.938	0.935	-	0.947	0.949	0.951	0.957	0.955
	$Sm\uparrow$	-	0.762	0.872	0.901	-	0.894	0.881	0.883	-	0.899	0.900	0.909	0.913	0.910
	$wF_{\beta}\uparrow$	-	0.665	0.811	0.863	-	0.855	0.834	0.835	-	0.870	0.868	0.874	0.883	0.882
	$F_{\beta}\uparrow$	-	0.765	0.836	0.882	-	0.869	0.862	0.859	-	0.892	0.881	0.879	0.895	0.895
	$\mathcal{M}\downarrow$	-	0.069	0.044	0.031	-	0.034	0.038	0.037	-	0.030	0.029	0.027	0.026	0.027
Backbone		ESPNet	VGG	VGG	Swin-B	VGG	ResNet	ResNet	Mobile	Segfor	HRFormer+	ResNet	Wave-	CDFFor	CDFFor
		v2	-16	-16		-16	-50	-50	netv2	mer	ResNet-18	-101	MLP	mer-m36	mer-b36
Params(M)		4.6	52.4	42.4	198.7	83.1	87.0	86.8	5.4	-	58.9	93.8	80.7	149.2	264.6
FLOPs(G)		4.2	257.9	15.6	124.3	77.6	39.7	94.2	1.2	-	17.3	63.9	64.0	139.7	238.0