Research Article
Disease Identification of Lentinus Edodes Sticks Based on Deep Learning Model
Table 2
Classification results of the ResNeXt-50(32 × 4d) model under different fine-tuning strategies.
| Fine-tuning method | Accuracy (%) | Precision | Recall | F1-score |
| FC0 | 91.39 | 0.8611 | 0.8891 | 0.8749 | FC1 | 92.39 | 0.9074 | 0.8938 | 0.9005 | FC2 | 90.58 | 0.9318 | 0.9041 | 0.9177 | FC3(the best) | 94.27 | 0.9432 | 0.9412 | 0.9422 | FC4 | 91.78 | 0.8879 | 0.9130 | 0.9003 | FC5 | 91.99 | 0.8959 | 0.9180 | 0.9068 | FC6 | 92.59 | 0.9357 | 0.9032 | 0.9192 |
|
|