Depth Prediction on Make3D C2 (test)
0.198Relative ErrorResnet50 10K
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Resnet50 10KBackbone=ResNet-50, Architecture=Multi-scale, Training data size=~10K image pairs2017.04 | 0.198 | 4.53 | 8.56 | |
| Resnet50-Multi-scaleBackbone=ResNet-50, Architecture=Multi-scale2017.04 | 0.212 | 4.71 | 8.73 | |
| ResNet50-CascadeBackbone=ResNet-50, Architecture=Cascade2017.04 | 0.221 | 4.79 | 8.81 | |
| Li et al.2017.04 | 0.279 | 0.102 | 10.27 | |
| Liu et al. [20]2017.04 | 0.307 | 0.125 | 12.89 | |
| Liu et al. [21]2017.04 | 0.338 | 0.134 | 12.6 | |
| Karsch et al.2017.04 | 0.361 | 0.148 | 15.1 |