Table 1.

Comparison of the patch-wise semantic segmentation method and Deeplab with different resize scales (400, 600, 800). The bold numbers mean that they are the best results.

	PA	MPA	MIoU	FWIoU
Deeplab_400	0.7401	0.3294	0.2514	0.6185
Deeplab_600	0.7508	0.4402	0.2714	0.6259
Deeplab_800	0.7612	0.4684	0.2995	0.6515
Patch-wise	0.7877	0.5978	0.4482	0.6904