Research map / 3D vision / Depth and stereo
Dense Correspondences across Scenes and Scales
Moria Tau, Tal Hassner
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 |
|---|---|---|---|
| use match-aware scale propagation instead of single-scale DSIFT | SIFT-Flow · Scaled-Middlebury Grove2 | angular error: 3.89 -> 0.13 | Improved |
| use match-aware instead of geometric propagation (ablation) | scale propagation · Scaled-Middlebury Grove2 | angular error: 2.30 -> 0.13 | Improved |
| use match-aware instead of image-aware propagation (ablation) | scale propagation · Scaled-Middlebury Dimetrodon | angular error: 2.95 -> 0.15 | Improved |
| use match-aware instead of geometric propagation (ablation) | scale propagation · Scaled-Middlebury Urban3 | endpoint error: 6.10 -> 1.15 | Improved |
| use match-aware instead of image-aware propagation (ablation) | scale propagation · Scaled-Middlebury Grove2 | endpoint error: 9.06 -> 0.71 | Improved |
| image-aware instead of geometric propagation (ablation) | scale propagation · Scaled-Middlebury Dimetrodon | angular error: 0.61 -> 2.95 | Hurt |
| match-aware propagation versus SLS descriptors | SIFT-Flow · Scaled-Middlebury Grove3 | angular error: 0.15 -> 0.18 | Hurt |
| match-aware propagation versus SLS on Urban3 endpoint error | SIFT-Flow · Scaled-Middlebury Urban3 | endpoint error: 9.41 -> 1.15 | Improved |
| match-aware propagation instead of SLS for run-time | SIFT-Flow descriptors · Flow run-time (78x52) | seconds: 13 -> 0.8 | Improved |
| match-aware propagation instead of SID for storage | SIFT-Flow descriptors · Descriptor dimension | dimension: 3,328D -> 128D | Improved |
| add 20% random scale assignment errors to SIFT-Flow scale map | SIFT-Flow · Synthetic self-matching | angular error | Hurt |
| add scale errors on up to half of pixels | SIFT-Flow · Synthetic self-matching | endpoint error | No clear effect |
Closest papers by meaning
- SIFTing Through Scales (TPAMI 2016 · similarity 0.90)
- Progressive Large Scale-Invariant Image Matching in Scale Space (ICCV 2017 · similarity 0.89)
- ScaleNet: A Shallow Architecture for Scale Estimation (CVPR 2022 · similarity 0.88)
- Continuous Surface Embeddings (NeurIPS 2020 · similarity 0.88)
- UFM: A Simple Path towards Unified Dense Correspondence with Flow (NeurIPS 2025 · similarity 0.85)
- SemAlign3D: Semantic Correspondence between RGB-Images through Aligning 3D Object-Class Representations (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.
