Table 6 Classification accuracy (in percent) of different on Pavia University dataset with five training samples per class

From: Multiscale superpixel depth feature extraction for hyperspectral image classification

Class names

2D-CNN

MCMs

HybridSN

SSFTT

CEGCN

MSSGU

SBD

SHDA

AMGCFN

MSDFE

Asphalt

66.48(14.04)

18.11(10.45)

30.63(20.54)

65.73(10.14)

90.34(3.56)

81.41(13.38)

63.79(8.71)

78.29(7.09)

73.05(12.95)

94.67(3.21)

Meadows

71.67(11.33)

79.33(7.71)

65.93(14.43)

76.12(10.25)

84.49(7.23)

81.91(10.69)

68.59(15.25)

74.63(10.51)

82.12(4.30)

98.63(1.23)

Gravel

63.11(11.14)

92.35(4.31)

76.08(17.27)

81.43(7.71)

88.85(12.90)

91.52(7.79)

76.03(19.38)

78.93(9.14)

88.05(8.24)

98.88(0.74)

Trees

91.53(7.76)

98.46(1.13)

63.99(12.53)

85.08(6.97)

87.55(7.66)

94.25(3.89)

53.84(6.22)

60.04(7.21)

77.15(9.81)

99.82(0.14)

Painted metal sheets

99.97(0.07)

100.00(0.00)

99.50(1.08)

99.66(0.78)

99.90(0.29)

99.93(0.20)

96.95(0.02)

97.70(2.71)

99.53(1.09)

100.00(0.00)

Bare Soil

63.55(15.61)

85.81(6.56)

55.88(28.03)

80.88(11.61)

97.50(3.12)

93.28(9.54)

88.15(9.01)

86.51(10.23)

96.33(3.09)

98.85(1.21)

Bitumen

66.99(24.17)

99.07(0.87)

88.48(22.49)

98.78(1.02)

99.80(0.25)

99.32(2.02)

92.48(2.51)

100.00(0.00)

94.67(6.44)

100.00(0.00)

Self-Blocking Bricks

52.47(16.11)

53.43(19.12)

38.93(25.68)

52.38(10.33)

92.65(4.35)

88.03(13.55)

81.68(7.21)

81.19(14.35)

96.56(1.74)

98.28(1.95)

Shadows

96.39(5.04)

99.88(0.14)

38.03(16.27)

90.70(6.25)

91.41(8.97)

96.94(3.11)

78.25(7.25)

94.96(5.70)

87.34(10.53)

100.00(0.00)

OA

70.54(4.95)

72.09(4.42)

58.44(8.47)

75.60(5.38)

89.17(3.73)

86.48(4.39)

72.42(6.73)

78.28(5.28)

84.61(2.06)

98.23(0.88)

AA

74.69(3.25)

80.72(2.85)

61.93(7.30)

81.09(2.56)

92.50(3.06)

91.84(1.98)

77.75(2.69)

83.58(2.23)

88.31(2.26)

98.79(0.57)

\(\kappa\)

62.76(5.58)

65.41(4.94)

48.61(9.36)

69.23(6.10)

86.13(4.61)

82.81(5.24)

65.68(7.55)

72.64(6.04)

80.27(2.58)

97.65(1.16)

  1. The best results are highlighted in bold