Research map / Label-efficient and robust learning / Domain adaptation and generalization
Revisiting Domain Generalized Stereo Matching Networks From a Feature Consistency Perspective
Check 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 stereo contrastive feature loss, momentum encoder, and stereo whitening (ablation) | PSMNet · KITTI2015 | threshold error rate (%): 27.9 -> 7.5 | Helped |
| Add contrastive loss without momentum encoder or stereo whitening (ablation) | PSMNet · KITTI2015 | threshold error rate (%): 27.9 -> 19.0 | Helped |
| Add momentum encoder to the contrastive-loss variant (ablation) | PSMNet · KITTI2015 | threshold error rate (%): 19.0 -> 12.7 | Helped |
| Add stereo whitening without contrastive loss or momentum encoder (ablation) | PSMNet · KITTI2015 | threshold error rate (%): 27.9 -> 15.5 | Helped |
| Add stereo contrastive feature loss and momentum encoder (ablation) | PSMNet · Middlebury half | threshold error rate (%): 26.9 -> 22.2 | Helped |
| Add stereo whitening (ablation) | PSMNet · ETH3D | threshold error rate (%): 23.8 -> 14.1 | Helped |
| Replace batch normalization with instance normalization (ablation) | PSMNet · KITTI2015 | threshold error rate (%): 12.7 -> 8.5 | Helped |
| Use domain normalization instead of the baseline normalization (ablation) | PSMNet · KITTI2015 | threshold error rate (%): 12.7 -> 8.1 | Helped |
| Use instance whitening instead of the baseline normalization (ablation) | PSMNet · KITTI2015 | threshold error rate (%): 12.7 -> 8.0 | Helped |
| Use stereo selective whitening instead of the baseline normalization (ablation) | PSMNet · KITTI2015 | threshold error rate (%): 12.7 -> 7.5 | Helped |
| Increase momentum from 0.9 to 0.9999 (ablation) | PSMNet contrastive feature learning · SceneFlow | feature cosine similarity: 0.88 -> 0.97 | Helped |
| Increase momentum from 0.9 to 0.9999 (ablation) | PSMNet contrastive feature learning · KITTI | feature cosine similarity: 0.82 -> 0.92 | Helped |
Closest papers by meaning
- Hierarchical Object-Aware Dual-Level Contrastive Learning for Domain Generalized Stereo Matching (NeurIPS 2024 · similarity 0.93)
- GraftNet: Towards Domain Generalized Stereo Matching With a Broad-Spectrum and Task-Oriented Feature (CVPR 2022 · similarity 0.91)
- Robust Synthetic-to-Real Transfer for Stereo Matching (CVPR 2024 · similarity 0.91)
- StereoGAN: Bridging Synthetic-to-Real Domain Gap by Joint Optimization of Domain Translation and Stereo Matching (CVPR 2020 · similarity 0.91)
- Stereo Risk: A Continuous Modeling Approach to Stereo Matching (ICML 2024 · similarity 0.88)
- Stereo Anywhere: Robust Zero-Shot Deep Stereo Matching Even Where Either Stereo or Mono Fail (CVPR 2025 · similarity 0.87)
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.