Table 3 Performance evaluation and comparison of different models on the test dataset.
From: Machine learning approach for wheat variety identification using single-seed imaging
Model | Wheat varieties | Precision (%) | Recall (%) | F1_score (%) | 95% CI for accuracy (%) |
|---|---|---|---|---|---|
Proposed model | Baran | 0.929 | 0.943 | 0.936 | 0.922 ± 0.044 |
Hashtrood | 0.933 | 0.931 | 0.932 | ||
Heydari | 0.911 | 0.924 | 0.917 | ||
Heyran | 0.920 | 0.906 | 0.913 | ||
Pishgham | 0.913 | 0.882 | 0.897 | ||
Zarrineh | 0.925 | 0.945 | 0.935 | ||
Inception-Resnet-v2 | Baran | 0.833 | 0.835 | 0.834 | 0.830 ± 0.081 |
Hashtrood | 0.825 | 0.827 | 0.826 | ||
Heydari | 0.831 | 0.832 | 0.832 | ||
Heyran | 0.823 | 0.824 | 0.824 | ||
Pishgham | 0.815 | 0.815 | 0.815 | ||
Zarrineh | 0.826 | 0.824 | 0.825 | ||
EfficientNet-B4 | Baran | 0.856 | 0.857 | 0.857 | 0.852 ± 0.076 |
Hashtrood | 0.852 | 0.853 | 0.853 | ||
Heydari | 0.855 | 0.856 | 0.856 | ||
Heyran | 0.849 | 0.849 | 0.849 | ||
Pishgham | 0.844 | 0.843 | 0.844 | ||
Zarrineh | 0.850 | 0.851 | 0.851 | ||
MLP-PCA | Baran | 0.860 | 0.880 | 0.870 | 0.86 ± 0.059 |
Hashtrood | 0.865 | 0.843 | 0.854 | ||
Heydari | 0.843 | 0.851 | 0.847 | ||
Heyran | 0.855 | 0.869 | 0.862 | ||
Pishgham | 0.859 | 0.839 | 0.849 | ||
Zarrineh | 0.882 | 0.872 | 0.882 |
