Table 6.
Macro average(overall) results obtained on ensembling models.
| Ensemble Model | Macro avg. Precision (%) | Macro avg. Recall (%) | Macro avg. F1-score (%) | Macro avg. Accuracy (%) | Zero-one Loss (Out of 900 test samples) | AUC-ROC-Score (%) | Error Rate (%) | Test Time |
|---|---|---|---|---|---|---|---|---|
| VGG16 + VGG19 + DenseNet201 | 86.45 | 85.44 | 85.11 | 96.76 | 131 | 91.81 | 14.55 | 4.130 s |
| VGG16 + VGG19 | 74.87 | 74.78 | 74.39 | 94.39 | 227 | 85.81 | 25.22 | 0.569 s |
| VGG16 + DenseNet201 | 89.41 | 88.11 | 87.69 | 97.35 | 107 | 93.31 | 11.88 | 3.518 s |
| VGG19 + DenseNet201 | 87.62 | 86.22 | 85.70 | 96.93 | 124 | 92.24 | 13.77 | 3.458 s |
| VGG16 + VGG19 + DenseNet201 + ResNet50 | 85.43 | 84.22 | 83.85 | 96.49 | 142 | 91.12 | 15.77 | 4.560 s |
| VGG16 + VGG19 + DenseNet201 + ResNet50 + MobileNetV2 | 91.60 | 91.00 | 90.81 | 98 | 81 | 94.93 | 9 | 11.203 s |
| DenseNet201 + MobileNetV2 | 95.17 | 94.89 | 94.83 | 98.86 | 46 | 97.12 | 5.11 | 3.918 s |
| VGG16 + VGG19 + DenseNet201 + ResNet50 + MobileNetV2 + NASNetMobile | 90.53 | 89.56 | 89.24 | 97.67 | 94 | 94.12 | 10.44 | 9.923 s |
| VGG16 + VGG19 + DenseNet201 + ResNet50 + MobileNetV2 + NASNetMobile + ResNet152V2 + DenseNet169 | 95.66 | 95.44 | 95.41 | 98.98 | 41 | 97.43 | 4.55 | 18.254 s |
| ResNet152V2 + DenseNet169 | 97.21 | 97.11 | 97.11 | 99.35 | 26 | 98.37 | 2.88 | 5.400 s |
| DenseNet201 + MobileNetV2 + ResNet152V2 + DenseNet169 | 98.57 | 98.56 | 98.56 | 99.67 | 13 | 99.18 | 1.44 | 11.087 s |
| MobileNetV2 + ResNet152V2 + DenseNet169 | 98.68 | 98.67 | 98.67 | 99.70 | 12 | 99.24 | 1.33 | 8.590 s |
| MobileNetV2 + ResNet152V2 | 97.94 | 97.89 | 97.89 | 99.53 | 19 | 98.81 | 2.11 | 3.515 s |
| TF: Bold values are observed as optimal results. | ||||||||