Table 2 Paired t-tests for DualSight improvement over each initial segmentation model across 1,520 particles in the held-out test set.
From: DualSight: multi-stage instance segmentation framework for improved precision
Model | t | p | Mean IoU | Mean DS IoU | Std \(\Delta\)IoU | 95% CI |
|---|---|---|---|---|---|---|
YOLOv8 Nano | 7.72 | < 0.0001 | 0.778 | 0.847 | 0.040 | 0.053–0.087 |
YOLOv8 X-Large | 7.86 | < 0.0001 | 0.749 | 0.818 | 0.039 | 0.053–0.087 |
Mask R-CNN | 11.06 | < 0.0001 | 0.761 | 0.863 | 0.041 | 0.086–0.121 |
