Table 6.
Pixel-wise accuracy evaluation of the segmentation approach using FCNNs for each microstructure class.
| Network | Data | Balanced | Augmented | Training Strategy | Matrix | Marten. | Temp. Marten. | Bain. | Pear. |
|---|---|---|---|---|---|---|---|---|---|
| FCNN | SEM | — | — | scratch | 86.47 | 36.08 | 0 | 0 | 0 |
| FCNN | SEM | — | — | fine tuning | 92.03 | 77.25 | 71.43 | 22.05 | 63.27 |
| FCNN | SEM | ✓ | — | fine tuning | 92.47 | 74.71 | 57.44 | 33.25 | 69.65 |
| FCNN | SEM | ✓ | ✓ | fine tuning | 94.22 | 79.85 | 72.62 | 37.32 | 70.46 |
| FCNN | LOM | ✓ | ✓ | fine tuning | 94.11 | 50.11 | 58.36 | 5.21 | 19.35 |
The matrix has the highest and bainite the lowest pixel-wise accuracy.