Table 5. Comparison between fixed learning rate method and cyclical learning rate method with FV-EffResNet.
| Learning method | Database | No of epochs | Accuracy |
|---|---|---|---|
| Fixed LR | FV-USM | 48 | 94.14% |
| SDUMLA | 40 | 92.73% | |
| MMCBNU_600 | 43 | 93.50% | |
| NUPT-FPV | 53 | 93.76% | |
| Cyclical LR | FV-USM | 28 | 95.75% |
| SDUMLA | 24 | 96.11% | |
| MMCBNU_600 | 24 | 98.43% | |
| NUPT-FPV | 32 | 96.84% |