Unsupervised Monocular Depth and Ego-motion Learning with Structure and Semantics
About
We present an approach which takes advantage of both structure and semantics for unsupervised monocular learning of depth and ego-motion. More specifically, we model the motion of individual objects and learn their 3D motion vector jointly with depth and ego-motion. We obtain more accurate results, especially for challenging dynamic scenes not addressed by previous approaches. This is an extended version of Casser et al. [AAAI'19]. Code and models have been open sourced at https://sites.google.com/corp/view/struct2depth.
Vincent Casser, Soeren Pirk, Reza Mahjourian, Anelia Angelova• 2019
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Depth Estimation | KITTI (Eigen split) | RMSE4.7503 | 276 | |
| Monocular Depth Estimation | KITTI (test) | Abs Rel Error0.1529 | 103 | |
| Monocular Depth Estimation | Cityscapes | Accuracy (delta < 1.25)82.6 | 62 | |
| Depth Prediction | Cityscapes (test) | RMSE7.0237 | 52 | |
| Depth Estimation | Cityscapes (test) | -- | 40 | |
| Depth Estimation | KITTI 50m cap (test) | Abs Rel0.103 | 24 | |
| Monocular Depth Estimation | Cityscapes (test) | Abs Rel Error0.145 | 11 | |
| Monocular Depth Estimation | Cityscapes 12 (test) | Abs Rel0.145 | 11 | |
| Depth Estimation | KITTI 80m cap revised evaluation code (test) | Abs Rel0.1108 | 9 |
Showing 9 of 9 rows