Research map / Detection and segmentation / CNN features
MESA: Effective Matching Redundancy Reduction by Semantic Area Segmentation
Yesheng Zhang, Shuhan Shen, Xu Zhao
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 |
|---|---|---|---|
| replace SGAM with MESA for area matching | SP+SG · ScanNet1500 area matching | AOR: 60.36 -> 68.44 | Improved |
| replace SGAM with DMESA for area matching | SP+SG · ScanNet1500 area matching | AOR: 60.36 -> 78.13 | Improved |
| relax GAM geometric constraint to phi=3.5 for MESA (ablation) | MESA+SP+SG · ScanNet1500 area matching | AOR: 65.12 -> 68.44 | Improved |
| relax GAM geometric constraint to phi=3.5 for DMESA (ablation) | DMESA+SP+SG · ScanNet1500 area matching | AOR: 71.36 -> 78.13 | Improved |
| use MESA dense area comparison instead of DMESA | SP+SG · ScanNet1500 area matching | ACR: 85.52 -> 94.93 | Improved |
| add MESA area matching to SP+SG | SP+SG · ScanNet1500 point matching 640x640 | MMA@3: 20.50 -> 24.62 | Improved |
| add DMESA area matching to ASpan | ASpan · ScanNet1500 point matching 640x640 | MMA@3: 25.13 -> 28.78 | Improved |
| add MESA area matching to QT | QT · ScanNet1500 point matching 640x640 | MMA@3: 22.85 -> 29.32 | Improved |
| add MESA area matching to ASpan at training size | ASpan · ScanNet1500 pose estimation 640x480 | AUC@5: 28.37 -> 28.23 | Hurt |
| add MESA area matching to SP+SG | SP+SG · ScanNet1500 pose estimation 640x640 | AUC@5: 20.22 -> 23.42 | Improved |
| add SGAM area matching to ASpan at 640x480 | ASpan · ScanNet1500 point matching 640x480 | MMA@3: 27.50 -> 24.51 | Hurt |
| add MESA area matching to DKM | DKM · ScanNet1500 point matching 640x640 | MMA@3: 26.15 -> 30.15 | Improved |
Closest papers by meaning
- MESA: Matching Everything by Segmenting Anything (CVPR 2024 · similarity 0.94)
- Image-to-Image Matching via Foundation Models: A New Perspective for Open-Vocabulary Semantic Segmentation (CVPR 2024 · similarity 0.85)
- MATCHA: Towards Matching Anything (CVPR 2025 · similarity 0.85)
- Matcher: Segment Anything with One Shot Using All-Purpose Feature Matching (ICLR 2024 · similarity 0.85)
- GIM: Learning Generalizable Image Matcher From Internet Videos (ICLR 2024 · similarity 0.84)
- Pixel-level Semantic Correspondence through Layout-aware Representation Learning and Multi-scale Matching Integration (CVPR 2024 · similarity 0.84)
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.
