Table 4 Data diagram of ablation experiment shows the effect of different functions on the recognition ability of DLVTNet.
From: Lightweight grape leaf disease recognition method based on transformer framework
Method | Accuracy | Precision | Recall | F1 score | Flop (G) | Params (M) |
|---|---|---|---|---|---|---|
Base | 86.76 | 86.23 | 86.45 | 86.23 | 0.329 | 0.719 |
+ GAN | 94.30 | 93.68 | 93.70 | 93.61 | 0.329 | 0.719 |
+ Ghost | 96.31 | 96.02 | 95.97 | 95.35 | 0.278 | 0.603 |
+ CLSHSA | 96.82 | 96.35 | 96.35 | 96.32 | 0.298 | 0.645 |
+ MARI | 97.92 | 97.53 | 97.52 | 97.51 | 0.406 | 0.860 |
+ Dense | 98.48 | 98.48 | 98.47 | 98.46 | 0.493 | 1.054 |
