Research map / 3D vision / Depth and stereo
UltraStereo: Efficient Learning-Based Matching for Active Stereo Systems
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 |
|---|---|---|---|
| learn a data-driven binary patch representation instead of using LSH | stereo matching on the synthetic test set · Synthetic test set (exhaustive disparity search) | precision: 51% -> 72% | Helped |
| learn a data-driven binary patch representation instead of using Census | stereo matching on the synthetic test set · Synthetic test set (exhaustive disparity search) | precision: 61% -> 72% | Helped |
| use 8 non-zero elements in the hyperplanes (ablation) | UltraStereo · Synthetic test set (exhaustive disparity search) | precision: 72% -> 74% | Helped |
| increase the number of non-zero hyperplane elements to k = 32 (ablation) | UltraStereo · Synthetic test set (exhaustive disparity search) | precision: 72% -> 76% | Helped |
| use dense hyperplanes (ablation) | UltraStereo · Synthetic test set (exhaustive disparity search) | precision: 72% -> 78% | Helped |
| use sparse learned hyperplanes with k = 4 rather than dense hyperplanes (ablation) | UltraStereo · Synthetic test set (exhaustive disparity search) | precision: 78% -> 72% | Hurt |
| use the learned sparse representation rather than PatchMatch Stereo's conventional matching cost | stereo matching on synthetic data · Synthetic data | depth bias | Helped |
| use the learned representation and matching framework | active stereo depth estimation · Real flat-wall sequences at multiple distances | depth bias | Mixed |
| transfer learned hyperplanes to a different camera without per-camera training | UltraStereo · Different camera sensor | depth estimation quality | No clear effect |
| use UltraStereo rather than HyperDepth when multiple active illuminators overlap | active stereo · Interference evaluation | depth-map quality | Helped |
Closest papers by meaning
- Connecting the Dots: Learning Representations for Active Monocular Depth Estimation (CVPR 2019 · similarity 0.91)
- Active Event-based Stereo Vision (CVPR 2025 · similarity 0.87)
- A Simple yet Universal Framework for Depth Completion (NeurIPS 2024 · similarity 0.87)
- Stereo Anywhere: Robust Zero-Shot Deep Stereo Matching Even Where Either Stereo or Mono Fail (CVPR 2025 · similarity 0.86)
- FoundationStereo: Zero-Shot Stereo Matching (CVPR 2025 · similarity 0.86)
- DEFOM-Stereo: Depth Foundation Model Based Stereo Matching (CVPR 2025 · similarity 0.85)
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.