From: Identification of citrus diseases based on AMSR and MF-RANet
Method | ResNet50 | MF-RANet |
---|---|---|
Training accuracy | 85.74% | 97.95% |
Testing accuracy | 85.06% | 96.87% |
mAP | 79.49% | 90.27% |
FPS | 49 | 88 |
GFLOPs | 4.12G | 6.1G |
Training time | 5Â h 48Â min 12Â s | 3Â h 47Â min 46Â s |
Model size (MB) | 26Â M | 24Â M |