Table 5.
Comparison of our proposal against related works.
| Reference | Techniques | Classification problem | # Classes | Results |
|---|---|---|---|---|
| This paper | ResNet101v2 | Multi-class | 5 | Accuracy = 98.00% |
| [19] | InceptionV3 | Two-class | 2 | Accuracy = 100% |
| [20] | SRCNet | Multi-class | 3 | Accuracy = 98.70% |
| [21] | ResNet50 and YOLOv2 | Two-class | 2 | Average Precision = 81% |
| [22] | YOLOv3 and Faster R-CNN (ResNet101-FPN) | Two-class | 2 | Average Precision = 62% |
| [23] | ResNet50 and SVM | Two-class | 2 | Accuracy = 100% |
| [24] | Faster R-CNN and InceptionV2 | Two-class | 2 | F1-score = 94.19% |
| [25] | VGG16 | Multi-class | 7 | Accuracy = 99.29% |
| [27] | SE-YOLOv3 | Multi-class | 3 | Average Precision = 73.7% |
| [28] | CSPDarkNet53, PANet and YOLOv4 | Multi-class | 3 | Average Precision = 98.30%, F1-score = 96.7% |
| [29] | ResNet50 | Two-class | 2 | Accuracy = 98.20% |
| [30] | YOLOv3 | Two-class | 2 | Average confidence = 97.00% |
| [31] | MobileNetv2 | Two-class | 2 | Accuracy = 96.85% |
| [32] | MobileNetv2 and Single Shot Detector | Two-class | 2 | Accuracy = 91.70% |
| [33] | MobileNetv2 | Two-class | 2 | Accuracy= 98.00% |
| [34] | MobileNetv2 | Two-class | 2 | Accuracy = 98.00% |
| [35] | InceptionV3 | Two-class | 2 | Accuracy = 98.00% |
| [36] | Single Shot Multibox Detector and MobileNetv2 | Two-class | 2 | Accuracy = 92.64%, F1-score = 93.00% |
| [37] | YOLOv4 | Two-class | 2 | Average Precision = 88%, F1-score = 99.54% |
| [38] | MobileNetv2 | Two-class | 2 | Accuracy = 81.74% |
| [39] | Multigraph CN-VGG16 | Two-class | 2 | Accuracy = 97.9% |
| [40] | MobileNetv2, DenseNet121, NASNet | Two-class | 2 | F1-score = 99.40% |
| [41] | MobileNetv2-SVM | Two-class | 2 | Accuracy = 97.11% |
| [42] | VGG-16 | Multi-class | 3 | Accuracy = 99.81% |