Table 3 Performance comparison with state-of-the-art methods.
From: A hybrid YOLO-UNet3D framework for automated protein particle annotation in Cryo-ET images
Method | Recall | F4-score | IoU | Localization Error (voxels) |
|---|---|---|---|---|
DeepFinder | 0.4387 | 0.4125 | 0.2341 | 25.678 |
UNet3D | 0.5672 | 0.5143 | 0.1897 | 23.456 |
YOLOv5-3D | 0.7150 | 0.6821 | 0.0780 | 19.234 |
UNet3D + DBSCAN | 0.7823 | 0.7057 | 0.2134 | 21.345 |
YOLOv5-3D + DBSCAN | 0.7685 | 0.7325 | 0.1256 | 18.765 |
3D ResNet | 0.5122 | 0.4876 | 0.1678 | 24.127 |
Ours | 0.8848 | 0.7969 | 0.3211 | 15.678 |
