Research map / 3D vision / Depth and stereo
Self-Supervised Monocular Depth Hints
Jamie Watson, Michael Firman, Gabriel J. Brostow, Daniyar Turmukhambetov
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use hints only where hint reprojection beats prediction, instead of supervised-only Fused SGM (ablation) | Fused SGM proxy supervision · KITTI Eigen split (stereo, 192x640, PP) | Sq Rel: 0.850 -> 0.780 | Improved |
| use selective Depth Hints loss instead of summing hint loss with reprojection loss (ablation) | l_sum · KITTI Eigen split (stereo, 192x640, PP) | Abs Rel: 0.108 -> 0.106 | Improved |
| use selective Depth Hints instead of hint pretraining then reprojection finetuning (ablation) | l_ps Fused SGM -> l_r · KITTI Eigen split (stereo, 192x640, PP) | Sq Rel: 0.916 -> 0.780 | Improved |
| use selective Depth Hints instead of uncertainty-modelled hints (ablation) | Klodt uncertainty · KITTI Eigen split (stereo, 192x640, PP) | Sq Rel: 0.905 -> 0.780 | Improved |
| add left-right consistency check to Random SGM hints (ablation) | Random SGM proxy supervision · KITTI Eigen split (stereo, 192x640, PP) | Sq Rel: 0.901 -> 0.877 | Improved |
| use Fused SGM hints instead of Random SGM hints (ablation) | proxy supervision · KITTI Eigen split (stereo, 192x640, PP) | Sq Rel: 0.901 -> 0.850 | Improved |
| add Depth Hints | Monodepth2 (stereo) · KITTI Eigen (stereo, 192x640) | Abs Rel: 0.110 -> 0.109 | Improved |
| add Depth Hints | Monodepth2 (stereo video, 320x1024) · KITTI Eigen (stereo video, 320x1024) | Abs Rel: 0.106 -> 0.100 | Improved |
| add Depth Hints | Monodepth2 (no pretraining) · KITTI Eigen (stereo, no pretraining) | Abs Rel: 0.129 -> 0.127 | Improved |
| add Depth Hints | Monodepth2 on FlyingThings3D · FlyingThings3D Sceneflow | Abs Rel: 0.340 -> 0.219 | Improved |
| add Depth Hints | 3Net (ResNet18) · KITTI Eigen (stereo, 192x640) | Abs Rel: 0.112 -> 0.112 | No clear effect |
| add Depth Hints | Kuznietsov (LiDAR semi-supervised) · KITTI Eigen | Abs Rel: 0.109 -> 0.108 | No clear effect |
Closest papers by meaning
- Digging Into Self-Supervised Monocular Depth Estimation (ICCV 2019 · similarity 0.93)
- On the Uncertainty of Self-Supervised Monocular Depth Estimation (CVPR 2020 · similarity 0.93)
- Reversing the cycle: self-supervised deep stereo through enhanced monocular distillation (ECCV 2020 · similarity 0.92)
- The Temporal Opportunist: Self-Supervised Multi-Frame Monocular Depth (CVPR 2021 · similarity 0.91)
- Self-Supervised Deep Monocular Depth Estimation With Ambiguity Boosting (TPAMI 2021 · similarity 0.91)
- Adaptive Confidence Thresholding for Monocular Depth Estimation (ICCV 2021 · similarity 0.91)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
