Table 2 Image data used for Mask R-CNN training and validation
From: A generalised approach for high-throughput instance segmentation of stomata in microscope images
Dataset | Training images | As a % of dataset | Val. images | Avg. stomata/image |
|---|---|---|---|---|
Gymnosperm 40× | 43 | 15.14% | 14 | 6 |
Gymnosperm 10× | 25 | 4.99% | 16 | 27 |
Cuticle | 28 | 4.20% | 19 | 32 |
Ginkgo | 15 | 7.61% | 10 | 10 |
USNM/USBG | 31 | 4.44% | 30 | 26 |
Poplar | 15 | 8.50% | 10 | 33 |
Total | 157 | − | 99 | - |