Table 1 The IoU (blue area), mIoU, Accuracy, Dice (blue area) and mDice of Image A1 using other methods and our CES method trained on group B.
From: Center-environment feature models for materials image segmentation based on machine learning
Methods | IoU | mIoU | Accuracy | Dice | mDice |
|---|---|---|---|---|---|
A1 | |||||
MRF | 0.315 | 0.502 | 0.727 | 0.479 | 0.647 |
Watershed | 0.763 | 0.862 | 0.966 | 0.866 | 0.923 |
Han | 0.529 | 0.721 | 0.921 | 0.692 | 0.823 |
Meanshift | 0.424 | 0.621 | 0.840 | 0.595 | 0.748 |
DT | 0.708 | 0.831 | 0.958 | 0.829 | 0.903 |
KNN | 0.688 | 0.821 | 0.958 | 0.815 | 0.896 |
K-means | 0.352 | 0.631 | 0.914 | 0.521 | 0.737 |
Navie Bayes | 0.664 | 0.807 | 0.955 | 0.798 | 0.886 |
SVM(RBF) | 0.686 | 0.820 | 0.958 | 0.813 | 0.895 |
SVM(Sigmoid) | 0.798 | 0.883 | 0.972 | 0.887 | 0.936 |
RF | 0.723 | 0.841 | 0.963 | 0.840 | 0.909 |
AdaBoost | 0.733 | 0.847 | 0.964 | 0.846 | 0.913 |
XGBoost | 0.781 | 0.874 | 0.971 | 0.877 | 0.930 |
DART | 0.769 | 0.867 | 0.969 | 0.870 | 0.926 |
GBDT | 0.783 | 0.876 | 0.971 | 0.879 | 0.931 |
CES | 0.788 | 0.878 | 0.972 | 0.881 | 0.933 |
