Table 2 Performance in key evaluation criteria of six methods and ensemble model on the test set.
From: A deep learning model for detection of leukocytes under various interference factors
Model | mAP | mAP @IoU0.50 | mAP @IoU0.75 | mAR | AP-BG | AP-EG | AP-NG | AP-M | AP-L |
|---|---|---|---|---|---|---|---|---|---|
FSAF | 0.742 | 0.874 | 0.860 | 0.866 | 0.618 | 0.684 | 0.870 | 0.756 | 0.785 |
FCOS | 0.795 | 0.896 | 0.880 | 0.913 | 0.642 | 0.826 | 0.894 | 0.795 | 0.819 |
Faster R-CNN | 0.815 | 0.940 | 0.919 | 0.879 | 0.738 | 0.847 | 0.893 | 0.785 | 0.814 |
Grid R-CNN | 0.822 | 0.926 | 0.920 | 0.903 | 0.709 | 0.836 | 0.847 | 0.890 | 0.832 |
DH Faster R-CNN | 0.848 | 0.951 | 0.939 | 0.910 | 0.753 | 0.892 | 0.908 | 0.842 | 0.843 |
Ensemble | 0.853 | 0.940 | 0.933 | 0.922 | 0.752 | 0.898 | 0.925 | 0.843 | 0.848 |
Cascade R-CNN | 0.856 | 0.948 | 0.938 | 0.909 | 0.781 | 0.905 | 0.916 | 0.838 | 0.841 |
