Skip to main content

Table 3 Classification accuracies for the ANN 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 98.36 96.45 97.69 95.48 98.23 89.35
B 95.33 71.50 67.95 53.00 88.29 49.00
C 95.05 86.50 79.65 90.00 77.78 87.50
D 92.49 93.81 81.63 95.24 81.67 93.33
E 93.88 87.62 92.68 90.48 94.82 87.14
F 40.48 85.00 41.98 85.00 39.00 97.50
G 100.00 64.74 87.72 57.80 100.00 64.16
H 96.92 100.00 97.44 69.09 95.24 90.91
I 50.81 63.00 79.05 83.00 48.08 75.00
J 100.00 96.00 99.04 82.40 100.00 94.40
K 93.25 94.00 97.11 67.20 95.24 64.00
L 97.66 97.66 85.46 90.65 94.81 93.93
M 91.67 82.50 84.21 56.00 93.20 48.00
N 88.32 98.33 70.37 19.00 58.33 30.33
O 94.90 96.13 62.68 86.13 80.86 66.77
P 97.25 82.67 77.52 66.67 70.03 92.67
Q 98.60 96.36 95.67 90.45 97.20 94.55
R 88.46 92.00 95.45 84.00 100.00 78.00
S 90.57 80.00 89.09 81.67 78.67 98.33
T 90.91 92.31 92.65 96.92 91.80 86.15
U 77.78 73.50 73.50 86.00 75.86 66.00
V 100.00 62.00 43.21 70.00 23.27 74.00
W 94.36 92.00 85.29 72.50 79.49 77.50
  Accuracy (%) = 88.71 Accuracy (%) = 75.02 Accuracy (%) = 76.39
  κ = 0.8811 κ = 0.7380 κ = 0.7522