Skip to main content

Table 1 The perception results of our proposed frameworks: where AR indicates the percentage before and after the average area change. PQ indicates the parameter quantification of the methods. Time indicates the run time of each batch size.“ –” means equal values for the same model

From: Exploring spatiotemporal changes in cities and villages through remote sensing using multibranch networks

Datasets

Model

P(%)

R(%)

F(%)

AP(%)

PQ(M)

Time(s)

LEVIR-CD

VGG-LR

63.54

65.19

64.35

12.33

4.12

2.01

ChangeNet

64.98

67.56

66.24

13.08

6.19

3.81

FDCNN

67.49

68.95

68.21

14.59

6.01

4.92

CD-UNet++

68.55

70.07

69.30

15.38

5.54

4.53

UNetLSTM

69.28

71.05

70.15

17.18

12.01

7.32

SRCDNet

74.83

80.62

77.62

20.59

10.11

6.07

ESCNet

77.07

84.16

80.45

23.44

17.22

11.88

FGCN [29]

78.51

79.44

78.97

24.94

19.72

14.43

PToP CNN [26]

79.92

81.57

80.74

26.26

20.02

15.84

KPCAMNet

81.56

86.07

83.75

27.48

21.49

16.09

Ours

84.38

92.04

88.04

31.43

18.14

11.95

SZAB

VGG-LR

33.24

35.52

34.34

5.19

ChangeNet

34.28

37.16

35.66

5.86

FDCNN

35.12

38.05

36.52

6.27

CD-UNet++

36.94

38.99

37.93

6.88

UNetLSTM

37.52

39.65

38.56

7.22

SRCDNet

40.69

42.28

41.47

9.44

ESCNet

42.14

44.37

43.23

10.87

FGCN [29]

43.49

45.74

44.57

11.04

PToP CNN [26]

44.23

46.07

45.13

11.26

KPCAMNet

44.52

46.01

45.25

11.32

Ours

46.15

64.27

53.72

13.49