Road Damage Classification on Damaged Roads Dataset
94.12AccuracyResNet-18 (TL)
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| ResNet-18 (TL)Training Mode=Transfer Learning, Training Time (s)=378.8, Model Size (MB)=44.72026.01 | 94.12 | 94.12 | 94.12 | 94.01 | |
| Custom CNNTraining Mode=From scratch, Training Time (s)=426.4, Model Size (MB)=0.852026.01 | 89.71 | 90 | 89.71 | 89.18 | |
| VGG-16 (TL)Training Mode=Transfer Learning, Training Time (s)=1232.8, Model Size (MB)=5282026.01 | 88.24 | 91.72 | 88.24 | 88.74 | |
| ResNet-18Training Mode=From scratch, Training Time (s)=552.65, Model Size (MB)=44.72026.01 | 83.82 | 84.84 | 83.82 | 84.16 | |
| VGG-16Training Mode=From scratch, Training Time (s)=2716.26, Model Size (MB)=5282026.01 | 72.06 | 51.92 | 72.06 | 60.36 |