Table 4 Test results of the improved model with freezing 5-layers consequently without data augmentation
From: Houseplant leaf classification system based on deep learning algorithms
No. | No. of freeze layers | Accuracy (%) | Precision (%) | Recall (%) | F1-score (%) |
|---|---|---|---|---|---|
1 | ResNet-(1–5) | 98.00 | 98.10 | 98.00 | 98.01 |
2 | ResNet-(6–10) | 97.60 | 97.69 | 97.60 | 97.61 |
3 | ResNet-(11–15) | 98.00 | 98.11 | 98.00 | 98.01 |
4 | ResNet-(15–20) | 97.80 | 97.91 | 97.80 | 97.81 |
5 | ResNet-(21–25) | 98.60 | 98.67 | 98.60 | 98.61 |
6 | ResNet-(26–30) | 98.20 | 98.32 | 98.20 | 98.22 |
7 | ResNet-(31–35) | 98.40 | 98.49 | 98.40 | 98.42 |
8 | ResNet-(36–40) | 98.20 | 98.26 | 98.20 | 98.21 |
9 | ResNet-(41–45) | 98.20 | 98.26 | 98.20 | 98.21 |
10 | ResNet-(46–50) | 97.60 | 97.77 | 97.60 | 97.63 |