Table 5 Quantitative comparison of different neural networks on validation sets (%).
Model | Acc | Precision | Recall | F1 | IoU | BER |
|---|---|---|---|---|---|---|
UNet | 95.57 | 88.28 | 90.15 | 88.89 | 82.87 | 5.16 |
SegNet | 95.72 | 88.18 | 90.24 | 88.89 | 82.08 | 5.07 |
PSPNet | 92.69 | 82.85 | 79.80 | 79.55 | 69.41 | 10.98 |
Improved DeepLabV3+ | 96.21 | 88.56 | 90.70 | 89.01 | 82.95 | 4.93 |
