Skip to main content

Table 6 Classification accuracies for the MLH 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    96.81 97.74 96.87 99.68
B    96.97 16.00 87.23 61.50
C    58.08 84.50 80.44 90.50
D    62.86 94.29 87.34 95.24
E    35.52 98.10 41.50 100.00
F    54.00 67.50 10.93 100.00
G    14.87 83.82 48.89 50.87
H    100.00 5.91 77.51 73.64
I    82.08 87.00 36.72 94.00
J    100.00 1.20 0.00 0.00
K    80.00 8.00 98.55 54.40
L    32.47 99.53 66.88 97.20
M    54.55 6.00 100.00 2.50
N    92.73 17.00 100.00 6.33
O    56.20 65.81 49.31 69.03
P    95.42 76.33 71.32 90.33
Q    100.00 12.27 100.00 17.73
R    100.00 68.00 100.00 32.00
S    96.88 51.67 93.33 46.67
T    100.00 30.77 85.71 92.31
U    93.24 34.50 80.39 82.00
V    51.47 70.00 100.00 26.00
W    89.66 13.00 65.70 79.50
  Accuracy (%) = N/A Accuracy (%) = 49.72 Accuracy (%) = 63.23
  κ = N/A κ = 0.4713 κ = 0.6136