Table 1 Ablation study of preprocessing techniques on soil texture classification.
From: Advanced deep learning framework for soil texture classification
Preprocessing Components | Accuracy | F1-Score | Precision | Recall | AUC |
|---|---|---|---|---|---|
All (Butterworth+ Box-Cox + Augmentation) | 0.981 | 0.896 | 0.902 | 0.877 | 0.981 |
Only Box-Cox + Augmentation | 0.961 | 0.873 | 0.885 | 0.851 | 0.963 |
Only Butterworth + Augmentation | 0.958 | 0.867 | 0.878 | 0.846 | 0.961 |
Only Butterworth + Box-Cox | 0.942 | 0.849 | 0.861 | 0.833 | 0.949 |
Only Augmentation | 0.939 | 0.842 | 0.858 | 0.821 | 0.946 |
Only Butterworth | 0.915 | 0.819 | 0.827 | 0.806 | 0.921 |
Only Box-Cox | 0.911 | 0.811 | 0.823 | 0.798 | 0.918 |
No Preprocessing (Raw Images) | 0.878 | 0.774 | 0.786 | 0.761 | 0.886 |
