Table 3.
Effect of data augmentations of SC-FSL on accuracy of plant disease recognition.
| Data Aug | Accuracy (%) | Data Aug | Accuracy (%) | ||
|---|---|---|---|---|---|
| 5-way 1-shot | 5-way 5-shot | 5-way 1-shot | 5-way 5-shot | ||
| A | 64.16 ± 0.86 | 79.55 ± 0.65 | B+C | 67.53 ± 0.87 | 81.28 ± 0.61 |
| B | 54.19 ± 0.85 | 69.07 ± 0.81 | B+D | 46.89 ± 0.86 | 62.49 ± 0.77 |
| C | 65.43 ± 0.87 | 82.65 ± 0.60 | C+D | 56.54 ± 0.92 | 73.88 ± 0.70 |
| D | 45.35 ± 0.86 | 57.88 ± 0.88 | A+B+C | 78.55 ± 0.81 | 92.90 ± 0.47 |
| A+B | 62.26 ± 0.87 | 78.01 ± 0.64 | A+B+D | 49.47 ± 0.92 | 61.72 ± 0.79 |
| A+C | 75.36 ± 0.80 | 91.55 ± 0.46 | B+C+D | 57.11 ± 0.90 | 74.87 ± 0.75 |
| A+D | 48.46 ± 0.89 | 59.39 ± 0.84 | A+B+C+D | 77.78 ± 0.81 | 91.78 ± 0.45 |
Underline indicates the highest recognition accuracy in a single data augmentation; double underline indicates the highest recognition accuracy in a combination of two kinds of data augmentations; bold indicates the highest recognition accuracy in a combination of three kinds of data augmentations.