Table 2.

Comparison with the different approaches in terms of the evaluation of VDS metrics on Pointer-10K⁴⁹.

Method	Split	AP	AP50	AP75	APM	APL	AR	AR50	AR75	ARM	ARL
	validation	0.804	0.870	0.823	0.921	0.979	0.882	0.922	0.894	0.923	0.978
VDN49	test	0.720	0.777	0.744	0.879	0.914	0.819	0.860	0.835	0.881	0.918
	validation	0.777	0.839	0.799	0.928	0.976	0.879	0.913	0.887	0.911	0.972
Anam-Net61	test	0.657	0.719	0.676	0.888	0.902	0.821	0.869	0.832	0.891	0.933
	validation	0.685	0.747	0.702	0.896	0.967	0.856	0.894	0.869	0.900	0.971
UNet62	test	0.622	0.671	0.637	0.875	0.897	0.801	0.850	0.816	0.867	0.909
	validation	0.789	0.793	0.802	0.909	0.979	0.873	0.913	0.876	0.918	0.975
UNet++63	test	0.759	0.811	0.777	0.906	0.958	0.853	0.888	0.867	0.901	0.964
	validation	0.040	0.063	0.038	0.164	0.417	0.227	0.312	0.229	0.162	0.415
SegNet64	test	0.032	0.056	0.030	0.132	0.392	0.219	0.309	0.219	0.126	0.390
	validation	0.790	0.858	0.809	0.912	0.969	0.872	0.912	0.883	0.938	0.963
Enet65	test	0.702	0.767	0.711	0.831	0.938	0.802	0.844	0.809	0.834	0.942
	validation	0.803	0.871	0.821	0.910	0.980	0.876	0.918	0.893	0.925	0.981
LEDNet66	test	0.718	0.774	0.743	0.880	0.910	0.811	0.853	0.829	0.884	0.918
	validation	0.817	0.872	0.829	0.930	0.969	0.887	0.921	0.893	0.927	0.977
improved Unet++67	test	0.751	0.814	0.761	0.878	0.930	0.877	0.913	0.890	0.907	0.938
	Validation	0.815	0.862	0.830	0.928	0.975	0.876	0.919	0.892	0.930	0.980
improved U$_{}^2$-Net68	test	0.747	0.818	0.770	0.877	0.947	0.850	0.880	0.857	0.915	0.950
	validation	0.813	0.871	0.832	0.929	0.976	0.889	0.925	0.893	0.933	0.979
AFFormer69	test	0.753	0.815	0.778	0.881	0.916	0.853	0.891	0.856	0.905	0.953
	validation	0.815	0.865	0.835	0.933	0.973	0.891	0.917	0.898	0.943	0.981
EfficientSAM70	test	0.747	0.819	0.773	0.886	0.921	0.855	0.886	0.863	0.903	0.956
	validation	0.820	0.873	0.839	0.936	0.981	0.892	0.925	0.901	0.940	0.981
PIN(Ours)	test	0.763	0.826	0.781	0.897	0.937	0.856	0.893	0.863	0.902	0.958

Significant values are in bold.