Table 3 The obtained number of features, the consumed feature extraction time, and the accuracy of the different CNN layers
Layer number | Evaluation metrics | ||
|---|---|---|---|
| Â | No. of features | Feature extraction time | Accuracy (%) |
Layer 2 (conv) | 290400 | 0.24 | 69.16 |
Layer 3 (ReLU) | 290400 | 0.25 | 84.58 |
Layer 4 (norm) | 290400 | 0.33 | 68.28 |
Layer 5 (pool) | 69984 | 0.35 | 66.96 |
Layer 6 (conv) | 186624 | 0.78 | 69.16 |
Layer 7 (ReLU) | 186624 | 0.80 | 48.01 |
Layer 8 (norm) | 186624 | 0.83 | 77.97 |
Layer 9 (pool) | 43264 | 0.85 | 99.55 |
Layer 10 (conv) | 64896 | 1.20 | 49.33 |
Layer 11 (ReLU) | 64896 | 1.21 | 81.93 |
Layer 12 (conv) | 64896 | 1.40 | 65.63 |
Layer 13 (ReLU) | 64896 | 1.41 | 98.23 |
Layer 14 (conv) | 43264 | 1.59 | 88.54 |
Layer 15 (ReLU) | 43264 | 1.60 | 96.47 |