Table 6 Performance improvement after fine tuning the stomata model to a specific dataset
From: A generalised approach for high-throughput instance segmentation of stomata in microscope images
Dataset | Quality | Train images | New precision (%) | New recall (%) | New F-score (%) |
|---|---|---|---|---|---|
Ferns:Â low | Low | 12 | 87.98 (+9.07) | 81.13 (+29.89) | 84.42 (+22.10) |
Grass | Low–Med | 15 | 90.91 (+5.71) | 82.68 (+27.02) | 86.60 (+19.28) |