Table 4 The number of seedlings classified correctly in each class and metrics for external classification using Ilastik descriptors of soybean seeds and seedlings according to their physiological quality.
From: Interactive machine learning for soybean seed and seedling quality classification
Method | Class | Training set | Cross-validation | Validation set |
|---|---|---|---|---|
(n = 422) | (n = 178) | |||
Hits (Total) | ||||
LDA | Vigorous seedlings | 159 (162) | – | 64 (69) |
Weak seedlings | 122 (131) | – | 44 (55) | |
Non-germinated seeds | 129 (129) | – | 52 (54) | |
Overall accuracy | 0.97 | 0.92 ± 0.04 | 0.90 | |
Kappa | 0.96 | 0.89 ± 0.06 | 0.85 | |
Precision | 0.97 | 0.92 ± 0.04 | 0.90 | |
Sensitivity | 0.97 | 0.93 ± 0.04 | 0.90 | |
Specificity | 0.99 | 0.96 ± 0.02 | 0.95 | |
RF | Vigorous seedlings | 160 (162) | – | 67 (69) |
Weak seedlings | 120 (131) | – | 49 (55) | |
Non-germinated seeds | 128 (129) | – | 52 (54) | |
Overall accuracy | 0.97 | 0.93 ± 0.03 | 0.94 | |
Kappa | 0.95 | 0.89 ± 0.04 | 0.92 | |
Precision | 0.97 | 0.93 ± 0.03 | 0.94 | |
Sensitivity | 0.97 | 0.93 ± 0.03 | 0.94 | |
Specificity | 0.98 | 0.96 ± 0.02 | 0.97 | |
SVM | Vigorous seedlings | 158 (162) | – | 60 (69) |
Weak seedlings | 110 (131) | – | 42 (55) | |
Non-germinated seeds | 127 (129) | – | 48 (54) | |
Overall accuracy | 0.94 | 0.89 ± 0.03 | 0.84 | |
Kappa | 0.90 | 0.83 ± 0.04 | 0.76 | |
Precision | 0.94 | 0.89 ± 0.02 | 0.84 | |
Sensitivity | 0.93 | 0.89 ± 0.03 | 0.84 | |
Specificity | 0.97 | 0.94 ± 0.02 | 0.92 | |
