Research map / 3D vision / Depth and stereo
Reversing the cycle: self-supervised deep stereo through enhanced monocular distillation
Filippo Aleotti, Fabio Tosi, Li Zhang, Matteo Poggi, Stefano Mattoccia
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 |
|---|---|---|---|
| add random input point sampling (R) (ablation) | MCN-SGM/L · KITTI 2015 proxy quality (Noc) | D1 (%): 6.36 -> 5.28 | Improved |
| add consensus mechanism on top of random sampling (ablation) | MCN-SGM/L-R · KITTI 2015 proxy quality (Noc) | D1 (%): 5.28 -> 2.95 | Improved |
| add augmentation to random sampling and consensus (ablation) | MCN-SGM/L-RC · KITTI 2015 proxy quality (Noc) | D1 (%): 2.95 -> 2.78 | Improved |
| add random sampling to MCN with BM/W points (ablation) | MCN-BM/W · KITTI 2015 proxy quality (Noc) | D1 (%): 11.86 -> 6.79 | Improved |
| add consensus to random sampling for BM/W (ablation) | MCN-BM/W-R · KITTI 2015 proxy quality (Noc) | D1 (%): 6.79 -> 3.79 | Improved |
| train PSMNet on MCN-BM/W-ARC proxies instead of photometric loss | PSMNet · KITTI 2015 | D1 Noc (%): 6.62 -> 3.68 | Improved |
| use MCN-BM/W-ARC proxies instead of BM/W proxies | PSMNet · KITTI 2015 | D1 Noc (%): 6.89 -> 3.68 | Improved |
| use MCN-BM/W-ARC proxies instead of BM/W proxies | Stereodepth · KITTI 2015 | D1 Noc (%): 5.19 -> 4.21 | Improved |
| add hole-filling to SGM/L proxies (ablation) | PSMNet · KITTI 2015 | D1 Noc (%): 5.46 -> 6.06 | Hurt |
| add full consensus and augmentation (ARC) to MCN-BM/W-R (ablation) | PSMNet · KITTI 2015 | D1 Noc (%): 4.30 -> 3.68 | Improved |
| use MCN-BM/W-ARC proxies instead of LiDAR/SGM | PSMNet · KITTI 2015 | D1 Noc (%): 3.93 -> 3.68 | Improved |
| use MCN proxies instead of BM/W on DrivingStereo | PSMNet · DrivingStereo (DS->DS) | D1 (%): 8.81 -> 1.87 | Improved |
Closest papers by meaning
- Revealing the Reciprocal Relations Between Self-Supervised Stereo and Monocular Depth Estimation (ICCV 2021 · similarity 0.92)
- Self-Supervised Monocular Depth Hints (ICCV 2019 · similarity 0.92)
- Self-Supervised Learning of Depth Inference for Multi-View Stereo (CVPR 2021 · similarity 0.92)
- The Temporal Opportunist: Self-Supervised Multi-Frame Monocular Depth (CVPR 2021 · similarity 0.92)
- Self-Supervised Deep Monocular Depth Estimation With Ambiguity Boosting (TPAMI 2021 · similarity 0.91)
- Self-Supervised Monocular Trained Depth Estimation Using Self-Attention and Discrete Disparity Volume (CVPR 2020 · 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.
