Table 5.

Performance comparison of our model and the classical segmentation models on the ALL-IDB dataset.

Models	DC (%)	TPp (%)	FPp (%)	FN (%)	AP (%)
U-Net	86.64	84.62	0.68	17.28	83.42
ATT-UNet	87.12	85.07	0.62	16.64	84.27
DCAN	88.34	86.25	0.56	15.88	84.93
Mask R-CNN	92.24	89.76	0.41	9.56	88.96
YOLACT	90.53	87.83	0.47	12.30	86.58
StarDist	92.89	90.04	0.29	8.17	89.42
SPCNet (ours)	93.67	90.68	0.18	6.33	90.09