Table 5.

The results of each model on the Postdam dataset.

Method	Backbone	Per-Class F1 Score(%)					MIOU	OA	F1
		Impervious Surface	Building	Low Vegetation	Tree	Car
Deeplabv3+	Resnet50	92.35	96.77	85.22	86.79	93.58	——	89.74	90.94
FCN	Resnet50	90.98	94.10	81.25	87.58	76.80	——	88.66	86.14
FCN+SE [48]	Resnet50	90.43	93.95	81.33	87.50	63.33	——	88.27	83.31
FCN+BAM [49]	Resnet50	90.77	94.01	81.54	87.78	71.76	——	88.62	85.17
FCN+CBAM [50]	Resnet50	90.86	94.03	81.16	87.63	76.26	——	88.64	85.99
FCN+GloRe [51]	Resnet50	90.57	93.99	81.28	87.49	70.09	——	88.41	84.68
PSP-Net	Resnet50	91.61	96.30	86.41	86.84	91.38	——	89.45	90.51
DA-Net	Resnet50	91.61	96.44	86.11	88.04	83.54	——	89.72	89.14
UZ_1 [48]	——	89.30	95.40	81.80	80.50	86.50	——	85.80	86.70
UFMG_4 [52]	——	90.80	95.60	84.40	84.30	92.40	——	87.90	89.50
LA-Net	Resnet50	93.05	97.19	87.30	88.04	94.19	——	90.84	91.95
RE-Net	Resnet50	93.33	96.30	86.63	85.94	96.11	84.93	91.37	91.66