Table 7 Test metric results for each semantic segmentation method for the Pavia University and Kennedy Space Center datasets while employing patching for different patch sizes. For Deep CNN Ensemble T = 10, for EECNN and TCNN-E-ILS T = 20. An explanation of the metrics can be found in "Experiment configurations" section

PU (k=2)

KSC (k=2)

HybridSN

99.59 ± 0.05

99.37 ± 0.05

99.50 ± 0.05

96.62 ± 0.1

96.10 ± 0.1

96.46 ± 0.1

5 x 5

U-Net

99.60 ± 0.05

99.40 ± 0.05

99.52 ± 0.1

96.90 ± 0.1

96.30 ± 0.1

96.52 ± 0.1

CEU-Net

98.61 ± 0.1

97.9 ± 0.1

98.00 ± 0.1

96.97 ± 0.1

96.29 ± 0.1

96.54 ± 0.1

HybridSN

99.58 ± 0.05

99.56 ± 0.05

99.38 ± 0.05

97.31 ± 0.1

96.54 ± 0.1

97.00 ± 0.1

10 x 10

U-Net

99.54 ± 0.05

99.10 ± 0.05

99.40 ± 0.05

95.24 ± 0.1

94.72 ± 0.1

94.69 ± 0.1

CEU-Net

99.59 ± 0.05

99.12 ± 0.05

98.00 ± 0.1

99.10 ± 0.1

98.57 ± 0.1

98.97 ± 0.1

HybridSN

99.57 ± 0.05

99.45 ± 0.05

99.47 ± 0.05

95.32 ± 0.1

93.81 ± 0.1

94.78 ± 0.1

15 x 15

U-Net

99.34 ± 0.05

99.50 ± 0.05

99.10 ± 0.05

92.1 ± 0.1

90.94 ± 0.1

91.19 ± 0.1

CEU-Net

99.59 ± 0.05

99.12 ± 0.05

98.00 ± 0.1

97.7 ± 0.1

96.57 ± 0.1

97.43 ± 0.1

EECNN

99.34 ± 0.06

99.30 ± 0.04

99.27 ± 0.07

N/A

N/A

N/A

25 x 25

EECNN

N/A

N/A

N/A

N/A

N/A

N/A

27 x 27

CNN Ensemble

94.98

N/A

92.04

N/A

N/A

N/A

33 x 33

TCNN-E-ILS

89.62

85.14

86.51

99.27 ± 0.36

98.87 ± 0.64

99.19 ± 0.41