Research Article
Hyperspectral Image Classification Model Using Squeeze and Excitation Network with Deep Learning
Table 6
Evaluation of accuracy on each class of Pavia University dataset with overall accuracy.
| Class | VGG-16 | Inception-v3 | ResNet-50 | Proposed |
| Asphalt | 93.85 | 92.61 | 96.20 | 96.89 | Meadows | 96.04 | 96.35 | 97.52 | 98.74 | Gravel | 78.33 | 81.26 | 80.09 | 81.63 | Trees | 90.12 | 96.79 | 96.62 | 96.08 | Painted metal sheets | 99.90 | 100 | 99.85 | 100 | Bare soil | 89.44 | 91.73 | 94.26 | 93.38 | Bitumen | 85.60 | 90.83 | 86.90 | 89.16 | Self-blocking bricks | 86.75 | 89.48 | 92.07 | 92.94 | Shadows | 100 | 99.73 | 99.97 | 100 | Overall accuracy | 94.85 | 95.14 | 95.57 | 96.05 |
|
|