Dense regression problem of reconstructing a pixel-wise depth map from a single image.
Main evaluation metric used here is Root Mean Square Error (RMSE).
Max depth is restricted to 50 meters.
| Model | Score | Weights | Paper | Code |
|---|---|---|---|---|
| DORN (ResNet) | 2.271 | Caffe | paper | Caffe |
| DORN (VGG) | 2.517 | N/A | paper | N/A |
| Kuznietsov et al. | 3.518 | TensorFlow | paper | TensorFlow |
Max depth is restricted to 80 meters.
| Model | Score | Weights | Paper | Code |
|---|---|---|---|---|
| DORN (ResNet) | 2.727 | Caffe | paper | Caffe |
| DORN (VGG) | 3.056 | N/A | paper | N/A |
| Kuznietsov et al. | 4.621 | TensorFlow | paper | TensorFlow |
| Model | Score | Weights | Paper | Code |
|---|---|---|---|---|
| DORN (ResNet) | 0.509 | Caffe | paper | Caffe |
| MS-CRF | 0.586 | Caffe | paper | Caffe |
| Li et al. | 0.635 | N/A | paper | N/A |