Table 3 Classification results using different combinations of Relu and LeakyRelu on four datasets.
From: Adaptive pixel attention network for hyperspectral image classification
Data Set | Metric | LR+LR | LR+R | R+LR | R+R |
|---|---|---|---|---|---|
IP | OA(%) | 74.78±2.19 | 74.69±1.69 | 74.17±1.66 | 74.4±1.60 |
AA(%) | 70.11±2.38 | 69.87±2.87 | 69.56±3.59 | 69.14±2.9 | |
\(Kap \times 100\) | 71.12±2.52 | 70.01±1.95 | 70.4±1.9 | 70.78±1.84 | |
KSC | OA(%) | 86.82±3.44 | 85.53±3.19 | 85.42±3.23 | 85.26±2.33 |
AA(%) | 82.31±4.09 | 81.9±3.78 | 82.28±3.77 | 81.76±3.44 | |
\(Kap \times 100\) | 85.32±3.83 | 83.88±3.56 | 83.75±3.6 | 83.57±2.59 | |
UP | OA(%) | 97.14±0.66 | 97.08±0.48 | 96.95±0.53 | 96.44±2.02 |
AA(%) | 96.53±0.61 | 96.48±0.71 | 96.2±0.72 | 95.92±2.21 | |
Kap100 | 96.21±0.87 | 96.12±0.65 | 95.96±0.71 | 95.26±2.71 | |
HU | OA(%) | 82.42±2.60 | 83.23±3.38 | 81.59±4.01 | 81.05±4.18 |
AA(%) | 84.60±2.65 | 85.56±3.32 | 83.91±3.29 | 83.6±3.26 | |
\(Kap \times 100\) | 80.99±2.80 | 81.87±3.65 | 80.09±4.33 | 79.5±4.53 |
