Table 21.
The comparative study between different well-known CNN and the proposed models
| Model | Accuracy % | Time (Sec.) | Number of model parameters | |
|---|---|---|---|---|
| VGG16 | Mean | 92.07 | 975.15 | 165,730,115 |
| SD | ± 0.61 | ± 134.70 | 151,015,427 | |
| VGG19 | Mean | 93.05 | 1189.35 | 171,039,811 |
| SD | ± 0.94 | ± 129.52 | 151,015,427 | |
| InceptionV3 | Mean | 80.35 | 1637.19 | 23,904,035 |
| SD | ± 2.42 | ± 204.06 | 2,101,251 | |
| ResNet50 | Mean | 74.48 | 428.53 | 23,593,859 |
| SD | ± 2.15 | ± 9.17 | 6,147 | |
| MobileNet | Mean | 89.16 | 555.74 | 4,256,867 |
| SD | ± 0.98 | ± 17.88 | 1,028,003 | |
| The initial proposed model in case 1 | Mean | 93.08 | 41.43 | 583,613 |
| SD | ± 0.44 | ± 1.61 | 583,613 | |
| The initial proposed model in case 2 | Mean | 98.63 | 28.77 | 583,613 |
| SD | ± 0.44 | ± 0.23 | 583,613 | |
| The final proposed BTC-fCNN model in case 3 | Mean | 98.86 | 26.92 | 583,613 |
| SD | ± 0.45 | ± 0.23 | 583,613 | |