Table 5 Quantitative comparison between the proposed method and seven alternative approaches

From: Multi-objective band selection algorithm based on NSGA-II for pattern segmentation of textile hyperspectral images

Method

PA (%)

IoU (%)

Precision (%)

Recall (%)

F1 (%)

Time (s)

Method in45 on HSI

82.46

55.44

63.53

81.31

71.33

33.03

Unet

85.96

64.52

72.47

86.89

74.36

0.05

Pspnet

87.87

67.45

69.39

76.57

76.27

0.06

DeepLabv3+

88.56

68.23

75.13

89.84

79.95

0.07

TEED

88.63

69.08

75.17

89.49

81.71

0.19

PiDiNet

84.43

64.37

71.79

79.14

70.44

0.37

DexiNed

84.01

60.75

74.49

71.44

67.34

3.19

Our method

93.74

73.19

75.78

95.54

84.52

5.73

  1. Bold values denote the optimal performance metrics among the compared methods in the respective evaluation tasks.
Back to article page