Table 2 Quantitative comparison including IoU, HD, mAP and F1 in three datasets.

		U-Net	PraNet	CaraNet	HarDNet	CENet	DCRNet	CTO	XBound	H2Former	Ours
BUSI	IoU\(\uparrow \)	0.785	0.824	0.828	0.826	0.819	0.799	0.813	0.811	0.796	0.846
	HD\(\downarrow \)	8.71	8.23	8.05	8.23	8.16	8.34	8.49	8.60	8.58	7.78
	mAP\(\uparrow \)	0.646	0.709	0.716	0.712	0.695	0.675	0.696	0.687	0.664	0.746
	F1\(\uparrow \)	0.842	0.873	0.876	0.879	0.867	0.856	0.864	0.874	0.848	0.896
ISIC 2017	IoU\(\uparrow \)	0.821	0.842	0.823	0.848	0.811	0.829	0.843	0.848	0.856	0.862
	HD\(\downarrow \)	16.51	15.84	16.73	15.78	16.10	16.19	15.81	15.48	15.48	15.39
	mAP\(\uparrow \)	0.829	0.828	0.799	0.832	0.783	0.814	0.827	0.832	0.842	0.856
	F1\(\uparrow \)	0.891	0.906	0.894	0.904	0.881	0.885	0.906	0.906	0.905	0.917
X-ray	IoU\(\uparrow \)	0.937	0.941	0.940	0.941	0.940	0.942	0.942	0.946	0.946	0.949
	HD\(\downarrow \)	15.70	14.89	15.00	14.94	15.36	15.44	14.90	14.94	14.41	14.31
	mAP\(\uparrow \)	0.922	0.925	0.921	0.928	0.907	0.912	0.928	0.928	0.927	0.936
	F1\(\uparrow \)	0.966	0.965	0.968	0.969	0.961	0.960	0.966	0.972	0.973	0.973

\(\uparrow \) and \(\downarrow \) denote larger and smaller is better, respectively. The italic represents the best results, bold represents the second best results.

Quick links

Search