Research Article
Attention-Based Convolutional Neural Network for Pavement Crack Detection
Table 5
Public Crack Dataset- Dice loss.
| Model | Precision/(%) | Mean IoU/(%) | Recall/(%) | F1/(%) |
| ENet | 76.18 | 55.45 | 56.68 | 65.00 | ExFuse | 48.92 | 48.88 | 50.00 | 49.45 | FCN | 80.17 | 67.83 | 79.11 | 79.64 | LinkNet | 89.90 | 68.18 | 79.66 | 84.47 | SegNet | 76.46 | 71.08 | 89.38 | 82.42 | UNet | 82.82 | 70.41 | 82.60 | 82.71 |
| CrackResAttentionNet | 90.72 | 71.69 | 81.93 | 86.10 |
|
|