Table 3.
The results of the classification models based on deep spectral features.
| Spectra | Models | Cal | Val | Pre |
|---|---|---|---|---|
| Vis-NIR | SVM | 99.612 ± 0.208 a | 99.186 ± 0.373 a | 99.496 ± 0.294 a |
| LR | 99.683 ± 0.136 a | 99.109 ± 0.591 a | 99.496 ± 0.402 a | |
| CNN | 99.729 ± 0.132 a | 99.070 ± 0.373 a | 99.535 ± 0.221 a | |
| NIR | SVM | 100.000 ± 0.000 a | 99.845 ± 0.087 a | 99.884 ± 0.106 a |
| LR | 99.994 ± 0.014 a | 99.806 ± 0.137 a | 99.767 ± 0.213 a | |
| CNN | 99.974 ± 0.027 a | 99.767 ± 0.212 a | 99.806 ± 0.237 a |
The letters in each column indicate the significance in the difference in the accuracy of different models at the confidence level of 5%. Data followed by the same letter are not significantly different.