Table 5 Comparison of the performance of different models on the CrackTree206.
From: An efficient semantic segmentation method for road crack based on EGA-UNet
Method | Dice(%) | IoUb(%) | IoUf(%) | mIoU(%) | Params(M) | FLOPs(G) | FPS |
|---|---|---|---|---|---|---|---|
FCN | - | - | - | - | 134.27 | 370.64 | 26 |
U-Net | 70.1 | 99.8 | 55.5 | 77.6 | 4.32 | 74.16 | 61 |
SegNet | 61.2 | 99.7 | 45.0 | 74.2 | 29.48 | 311.67 | 28 |
PSPNet | 28.2 | 99.2 | 16.4 | 57.8 | 46.59 | 325.16 | 27 |
DeepLabV3 | 20.8 | 98.8 | 12.2 | 55.5 | 73.22 | 300.54 | 30 |
EGA-UNet | 73.1 | 99.8 | 58.1 | 78.9 | 4.73 | 71.46 | 64 |
