Table 7. Comparison of different backbones.

Dataset	Backbone	Overall accuracy	Mean F1-Score	mIoU
LoveDA	ResNet18	94.66	74.02	61.35
	DenseNet121	95.16	76.25	63.81
	Swin Transformer	95.08	75.98	63.55
	MobileNetV3	94.78	74.22	61.62
	LDCNet	95.42	77.41	65.27
CITY_OSM	ResNet18	90.04	84.70	73.49
	DenseNet121	90.38	85.14	74.32
	Swin Transformer	90.29	85.03	74.19
	MobileNetV3	89.91	84.37	73.01
	LDCNet	90.49	85.28	74.39

Notes.

Bold values represent the best performance indicators.