Table 1 Comparison results of multiple models on the Massachusetts dataset. The best values are highlighted in bold and the second-best values are underlined.

From: DSWFNet: dual-branch fusion of spatial and wavelet features for road extraction from remote sensing images

Methods

Params (M)

FLOPs (G)

P (%)

Recall (%)

F1 (%)

IoU (%)

D-linkNet14

31.10

48.62

76.48

79.91

78.16

64.15

SDUNet16

2.47

234.72

76.57

79.74

78.12

64.10

REF-LinkNet50

34.29

63.12

78.77

76.98

77.86

63.75

SFFNet29

34.18

59.77

78.27

80.31

79.28

65.67

OARENet51

33.50

3.81

75.86

81.66

78.65

64.82

LCMorph30

71.90

587.72

78.04

78.73

78.39

64.45

DenseDDSSPP-DeepLabV3+52

29.45

107.99

72.42

79.33

75.72

60.93

AEFNet53

20.24

51.09

76.68

79.11

77.88

63.77

DSWFNet-Ours

61.33

18.47

79.41

79.73

79.57

66.07