Table 4 CNN classification performance comparison with the IP102 dataset.
From: Deep learning based agricultural pest monitoring and classification
CNN Model | Input Size | Precision | Recall | Training Accuracy | Testing Accuracy |
|---|---|---|---|---|---|
GoogleNet26 | – | 36.80 | 31.70 | – | 40.70 |
VGGNet26 | – | 41.90 | 37.80 | – | 47.10 |
ResNet26 | – | 43.70 | 39.10 | 87.26 | 49.40 |
DenseNet20114 | 128 × 128 × 3 | – | – | – | 61.93 |
GAEnsemble1 | 299 × 299 × 3 | 67.17 | 67.13 | – | 67.13 |
DenseNet with OTAdapt | – | – | – | – | 62.32 |
ResNet50 (Before Augmentation) | 128 × 128 × 3 | 53.20 | 50.10 | 89.70 | 55.60 |
ResNet50 (After Augmentation) | 128 × 128 × 3 | 78.50 | 77.20 | 96.12 | 83.10 |
ResNet152 (Proposed, Before Augmentation) | 224 × 224 × 3 | 54.30 | 52.00 | 90.20 | 57.80 |
ResNet152 (Proposed, After Augmentation) | 224 × 224 × 3 | 80.10 | 79.30 | 91.91 | 84.95 |
