Research map / Label-efficient and robust learning / Domain adaptation and generalization
Zoom and Learn: Generalizing Deep Stereo Matching to Novel Domains
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 graph Laplacian regularization to iterative learning from zoomed-in predictions (ablation) | ZOLE · Smartphone test | PSNR: 22.84 -> 23.12 | Helped |
| Add graph Laplacian regularization to iterative learning from zoomed-in predictions (ablation) | ZOLE · Smartphone test | SSIM: 0.851 -> 0.855 | Helped |
| Add graph Laplacian regularization to iterative learning from zoomed-in predictions (ablation) | ZOLE · KITTI stereo 2015 test | EPE: 1.34 -> 1.25 | Helped |
| Add graph Laplacian regularization to iterative learning from zoomed-in predictions (ablation) | ZOLE · KITTI stereo 2015 test | 3ER: 7.56% -> 6.76% | Helped |
| Adapt the synthetic-pretrained model using target-domain stereo pairs without disparity labels | DispNetC · Smartphone test | PSNR: 21.99 -> 23.12 | Helped |
| Adapt the synthetic-pretrained model using target-domain stereo pairs without disparity labels | DispNetC · KITTI stereo 2015 test | EPE: 1.64 -> 1.25 | Helped |
| Adapt the synthetic-pretrained model using target-domain stereo pairs without disparity labels | DispNetC · KITTI stereo 2015 test | 3ER: 11.41% -> 6.76% | Helped |
| Use ZOLE adaptation rather than the synthetic-only DispNetC-80 model | DispNetC-80 · FlyingThings3D-80 test | EPE: 0.93 -> 1.11 | Hurt |
| Use ZOLE adaptation rather than the synthetic-only DispNetC-80 model | DispNetC-80 · FlyingThings3D-80 test | 3ER: 5.11% -> 6.54% | Hurt |
| Increase input resolution from 896 through 1664, then increase it further | DispNetC · KITTI stereo 2015 training set | 3ER: 14.26% -> 8.81% -> 10.53% | Mixed |
| Increase input resolution from 896 through 2048, then increase it further | DispNetS · KITTI stereo 2015 training set | 3ER: 18.95% -> 8.64% -> 9.08% | Mixed |
| Vary the zoom ratio used to create finer-grain pseudo-targets from 1.2 to 2 (ablation) | ZOLE · Smartphone adaptation | performance | No clear effect |
Closest papers by meaning
- Domain Generalized Stereo Matching via Hierarchical Visual Transformation (CVPR 2023 · similarity 0.91)
- Guided Stereo Matching (CVPR 2019 · similarity 0.91)
- Efficient Deep Learning for Stereo Matching (CVPR 2016 · similarity 0.91)
- Stereo Risk: A Continuous Modeling Approach to Stereo Matching (ICML 2024 · similarity 0.90)
- Robust Synthetic-to-Real Transfer for Stereo Matching (CVPR 2024 · similarity 0.90)
- Hierarchical Object-Aware Dual-Level Contrastive Learning for Domain Generalized Stereo Matching (NeurIPS 2024 · similarity 0.89)
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.