Research map / Label-efficient and robust learning / Domain adaptation and generalization
Digging Into Uncertainty-Based Pseudo-Label for Robust Stereo Matching
Zhelun Shen, Xibin Song, Yuchao Dai, Dingfu Zhou, Zhibo Rao, Liangjun Zhang
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 attention-based disparity refinement module (ablation) | UCFNet · KITTI2015 (train on SceneFlow only) | D1_all (%): 5.8% -> 5.2% | Improved |
| feed left image or pixel uncertainty into the refinement module in addition to disparity (ablation) | UCFNet refinement · KITTI2015 (train on SceneFlow only) | D1_all (%): 5.2% -> 5.4% | No clear effect |
| feed left image feature into the refinement module (ablation) | UCFNet refinement · KITTI2015 (train on SceneFlow only) | D1_all (%): 5.2% -> 5.2% | No clear effect |
| domain adaptation with pixel-level pseudo-label (ablation) | CFNet · KITTI2015 (adaptation) | D1_all: 5.8 -> 4.5 | Improved |
| use area-level instead of pixel-level pseudo-label (ablation) | CFNet · KITTI2015 (adaptation) | D1_all: 4.5 -> 3.88 | Improved |
| domain adaptation with pixel-level pseudo-label (ablation) | UCFNet · KITTI2015 (adaptation) | D1_all: 5.2 -> 3.89 | Improved |
| use area-level instead of pixel-level pseudo-label (ablation) | UCFNet · KITTI2015 (adaptation) | D1_all: 3.89 -> 3.64 | Improved |
| raise pixel-uncertainty threshold from 0.9 to 1.0 (ablation) | UCFNet U_pixel pseudo-label · KITTI2015 (adaptation) | D1_all (%): 3.89 -> 4.17 | Hurt |
| lower area-uncertainty threshold from 0.2 to 0.1 (sparser, more accurate labels) (ablation) | UCFNet U_area pseudo-label · KITTI2015 (adaptation) | D1_all (%): 3.64 -> 3.76 | Hurt |
| run a second adaptation iteration (ablation) | UCFNet · KITTI2015 (adaptation) | D1_all: 3.64 -> 3.14 | Improved |
| run a third adaptation iteration (ablation) | UCFNet · KITTI2015 (adaptation) | D1_all: 3.14 -> 3.16 | No clear effect |
| add uncertainty map input to area-level estimator (ablation) | area-level uncertainty network · KITTI2015 | D1_all of pseudo-label (%): 0.81% -> 0.46% | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- CFNet — CFNet: Cascade and Fused Cost Volume for Robust Stereo Matching (CVPR 2021) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Learning the Distribution of Errors in Stereo Matching for Joint Disparity and Uncertainty Estimation (CVPR 2023 · similarity 0.91)
- On the confidence of stereo matching in a deep-learning era: a quantitative evaluation (TPAMI 2021 · similarity 0.89)
- Robust Synthetic-to-Real Transfer for Stereo Matching (CVPR 2024 · similarity 0.89)
- Stereo Risk: A Continuous Modeling Approach to Stereo Matching (ICML 2024 · similarity 0.88)
- Consistency-aware Self-Training for Iterative-based Stereo Matching (CVPR 2025 · similarity 0.88)
- Modeling Stereo-Confidence out of the End-to-End Stereo-Matching Network via Disparity Plane Sweep (AAAI 2024 · similarity 0.88)
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.
