Research map / Point clouds / Registration and matching
Best-Buddies Similarity—Robust Template Matching Using Mutual Nearest Neighbors
Shaul Oron, Tali Dekel, Tianfan Xue, William T. Freeman, Shai Avidan
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 |
|---|---|---|---|
| consider top 3 modes with NMS instead of only the best mode (ablation) | BBS with color features · Web video template matching (df=25, 270 pairs) | mAP: 0.589 -> 0.648 | Improved |
| use Deep (VGG) features instead of color features (ablation) | BBS · Web video template matching (df=25, 270 pairs), top 3 modes | mAP: 0.648 -> 0.684 | Improved |
| use BBS instead of BDS (runner up) with color features | template matching · Web video template matching (df=25, 270 pairs) color, top mode | accuracy margin: 4.6% | Improved |
| use BBS instead of BDS with color features, top 3 modes | template matching · Web video template matching (df=25, 270 pairs) color, top 3 modes | accuracy margin: 5.9% | Improved |
| use BBS instead of second best with Deep features, top mode | template matching · Web video template matching (df=25, 270 pairs) Deep, top mode | accuracy margin: 2.4% | Improved |
| use BBS instead of SSD with Deep features, top 3 modes | template matching · Web video template matching (df=25, 270 pairs) Deep, top 3 modes | accuracy margin: 5.2% | Improved |
| use BBS instead of SSD with Deep features at large baseline df=100 | template matching · Web video template matching (df=100) Deep features, top 3 modes | mAP: 0.6 vs 0.6 | No clear effect |
| use BBS instead of histogram matching (chi-square) with color features | template matching · Web video template matching (df=25, 270 pairs) color features | mAP | Improved |
Closest papers by meaning
- OATM: Occlusion Aware Template Matching by Consensus Set Maximization (CVPR 2018 · similarity 0.88)
- Template Matching With Deformable Diversity Similarity (CVPR 2017 · similarity 0.87)
- Robust Point Matching with Distance Profiles (JMLR 2025 · similarity 0.87)
- Joint Probabilistic Matching Using m-Best Solutions (CVPR 2016 · similarity 0.86)
- NCMNet: Neighbor Consistency Mining Network for Two-View Correspondence Pruning (TPAMI 2024 · similarity 0.86)
- Locality Preserving Refinement for Shape Matching with Functional Maps (AAAI 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.
