TABLE 4.

Performance comparison of the DRIVE dataset.

Method	Year	SE	SP	ACC	F1	AUC
U-Net Ronneberger et al. (2015)	2015	0.7776	0.9867	0.9681	0.8108	0.9766
R2UNet Alom et al. (2019)	2018	0.7792	0.9813	0.9556	0.8171	0.9784
DUNet Jin et al. (2019)	2019	0.7963	0.9800	0.9566	0.8237	0.9802
AG-Net Zhang et al. (2019)	2019	0.8100	0.9848	0.9692	—	0.9856
IterNet Li et al. (2020)	2019	0.7921	0.9874	0.9699	0.8244	0.9861
NFN+ Wu et al. (2020)	2020	0.7796	0.9813	0.9582	0.8295	0.9830
RVSeg-Net Wang et al. (2020b)	2020	0.8107	0.9845	0.9681	—	0.9817
SCS-Net Wu et al. (2021)	2021	0.8289	0.9838	0.9697	—	0.9837
SA-UNet Guo et al. (2021b)	2021	0.8264	0.9823	0.9687	0.8224	0.9861
CAR-UNet Guo et al. (2021a)	2022	0.8135	0.9849	0.9699	—	0.9852
FR-UNet Liu et al. (2022)	2022	0.8356	0.9837	0.9705	0.8316	0.9889
AGC-Net (Our)	2023	0.8251	0.9844	0.9704	0.8301	0.9881

The value in bold is the highest value under that metric.