Skip to main content

Table 5 Classification accuracies for the SAM classifications

From: Classification of hyperspectral imagery with neural networks: comparison to conventional tools

  194-band 30-band 13-band
  User Producer User Producer User Producer
Class Accuracy Accuracy Accuracy
A 98.34 95.48 98.70 73.55 98.77 77.42
B 96.05 36.50 54.05 20.00 55.26 21.00
C 54.63 56.00 35.42 42.50 38.14 45.00
D 91.44 96.67 91.44 96.67 91.86 96.67
E 88.20 67.62 89.16 70.48 89.40 64.29
F 41.94 65.00 40.26 77.50 42.17 87.50
G 100.00 8.09 100.00 18.50 100.00 24.86
H 98.14 95.91 97.56 90.91 98.04 90.91
I 33.49 70.00 24.80 63.00 26.52 61.00
J 100.00 100.00 98.41 99.20 98.80 99.20
K 97.22 84.00 99.04 82.40 97.51 78.40
L 93.69 97.20 94.84 94.39 94.50 96.26
M 100.00 27.00 94.44 17.00 92.98 26.50
N 97.66 83.33 5.51 2.33 12.16 6.00
O 88.86 95.16 49.24 62.90 49.24 62.90
P 80.00 96.00 72.84 95.67 70.05 94.33
Q 99.07 96.82 97.71 96.82 96.77 95.45
R 100.00 94.00 79.63 86.00 78.43 80.00
S 81.94 98.33 63.01 76.67 61.04 78.33
T 45.61 80.00 53.54 81.54 48.21 83.08
U 43.21 78.00 39.21 64.50 36.67 60.50
V 71.43 70.00 41.94 78.00 40.79 62.00
W 95.40 83.00 93.46 71.50 94.70 71.50
  Accuracy (%) = 79.18 Accuracy (%) = 66.37 Accuracy (%) = 66.81
  κ = 0.7808 κ = 0.6466 κ = 0.6510