Table 6 Test metric results for each semantic segmentation method for the Indian Pines and Salinas datasets while employing patching for different patch sizes. For Deep CNN Ensemble T = 10, for EECNN and TCNN-E-ILS T = 20. An explanation of the metrics can be found in "Experiment configurations" section
From: CEU-Net: ensemble semantic segmentation of hyperspectral images using clustering
Patch | Methods | OA | AA | Kappa | OA | AA | Kappa |
|---|---|---|---|---|---|---|---|
| Â | Â | IP (k=2) | Salinas (k=3) | ||||
|  | HybridSN | 98.55 ± 0.1 | 98.19 ± 0.1 | 98.35 ± 0.1 | 99.80 ± 0.05 | 99.76 ± 0.05 | 99.78 ± 0.05 |
5 x 5 | U-Net | 95.10 ± 0.1 | 94.85 ± 0.1 | 94.82 ± 0.1 | 99.81 ± 0.05 | 99.80 ± 0.05 | 99.87 ± 0.05 |
|  | CEU-Net | 94.50 ± 0.1 | 93.50 ± 0.1 | 93.75 ± 0.1 | 98.78 ± 0.1 | 99.31 ± 0.05 | 98.64 ± 0.05 |
|  | HybridSN | 97.03 ± 0.1 | 95.57 ± 0.1 | 96.62 ± 0.1 | 99.80 ± 0.05 | 99.74 ± 0.05 | 99.73 ± 0.05 |
10 x 10 | U-Net | 97.35 ± 0.1 | 96.64 ± 0.1 | 96.99 ± 0.1 | 99.78 ± 0.05 | 99.76 ± 0.05 | 99.75 ± 0.05 |
|  | CEU-Net | 96.34 ± 0.1 | 95.29 ± 0.1 | 95.37 ± 0.1 | 99.85 ± 0.05 | 99.78 ± 0.05 | 99.78 ± 0.05 |
|  | HybridSN | 97.23 ± 0.1 | 94.08 ± 0.1 | 95.84 ± 0.1 | 99.79 ± 0.05 | 99.75 ± 0.05 | 99.77 ± 0.05 |
15 x 15 | U-Net | 95.70 ± 0.1 | 89.78 ± 0.1 | 95.10 ± 0.1 | 99.80 ± 0.05 | 99.76 ± 0.05 | 99.76 ± 0.05 |
|  | CEU-Net | 97.36 ± 0.1 | 94.68 ± 0.1 | 95.98 ± 0.1 | 99.86 ± 0.05 | 99.77 ± 0.05 | 99.77 ± 0.05 |
|  | EECNN | N/A | N/A | N/A | 98.48 ± 0.03 | 98.37 ± 0.03 | 97.58 ± 0.04 |
25 x 25 | EECNN | 97.57 ± 0.07 | 96.23 ± 0.02 | 97.23 ± 0.08 | N/A | N/A | N/A |
27 x 27 | CNN Ensemble | 92.54 | N/A | 90.94 | 96.05 | N/A | 95.93 |
33 x 33 | TCNN-E-ILS | 91.88 ± 1.13 | 77.37 ± 4.04 | 90.28 ± 1.34 | N/A | N/A | N/A |