Using AAEHS-Net as an Attention-Based Auxiliary Extraction and Hybrid Subsampled Network for Semantic Segmentation

Table 9

Results of segmentation on the Massachusetts buildings dataset.

Method	Accuracy	Dice	MIOU	AUC
U-Net	0.9305	0.6931	0.6782	0.9524
MHSU-Net	0.9451	0.7155	0.6781	0.9587
HAD-ResUNet	0.9428	0.7142	0.6802	0.9574
MR-UNet	0.9462	0.7179	0.6798	0.9620
CE-UNet	0.9490	0.7137	0.6811	0.9598
Ours	0.9515	0.7210	0.6835	0.9683

The bold value indicates the best value for the item.