Research map / Learning theory and algorithms / Clustering
MultiLink: Multi-Class Structure Recovery via Agglomerative Clustering and Model Selection
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 |
|---|---|---|---|
| Use MultiLink instead of Prog-X for plane segmentation | Adelaide RMF plane segmentation · Adelaide RMF | ME (%): 6.86 -> 6.46 | Helped |
| Use MultiLink instead of MCT with per-sequence ε tuning | Plane segmentation · Adelaide RMF | ME (%): 6.13 -> 4.10 | Helped |
| Use MultiLink instead of the strongest listed baseline | Two-view segmentation with fundamental matrices · Adelaide RMF | ME (%): 10.73 -> 8.59 | Helped |
| Use MultiLink instead of Multi-X | Two-view segmentation with affine fundamental matrices · Adelaide RMF | ME (%): 10.5 -> 9.84 | Helped |
| Mix fundamental, affine fundamental, and homography models (ablation) | MultiLink two-view segmentation · Adelaide RMF | ME (%): 8.59 -> 7.75 | Helped |
| Fit mixed 2D and 3D subspaces instead of only 3D subspaces (ablation) | MultiLink video motion segmentation · Hopkins 155 | ME (%): 8.34 -> 9.83 | Hurt |
| Use MultiLink instead of Prog-X | Video motion segmentation with 3D subspaces · Hopkins 155 | ME (%): 8.41 -> 8.34 | Helped |
| Use MultiLink instead of MCT | Conic fitting · Cases (c), (d), (e), (f), and (g) | ME (%): 2.33 -> 1.55; 5.23 -> 1.83; 7.12 -> 0.87; 5.38 -> 2.46; 6.23 -> 4.28 | Helped |
| Use MultiLink instead of MCT | Conic fitting · Cases (a) and (b) | ME (%): 0.67 -> 2.17; 2.00 -> 2.13 | Hurt |
| Use MultiLink instead of T-linkage | Line fitting clustering · Line fitting | clustering runtime (s): 0.76 s -> 0.26 s | Helped |
| Fit new models on-the-fly during clustering rather than relying only on sampled hypotheses (ablation) | MultiLink line fitting · Line fitting across inlier thresholds and sampling runs | ME | Helped |
| Use mixed model classes rather than only 3D subspaces (ablation) | MultiLink video motion segmentation · Traffic 3 | ME (%): 7.51% -> 1.37% | Helped |
Closest papers by meaning
- Fitting Multiple Heterogeneous Models by Multi-Class Cascaded T-Linkage (CVPR 2019 · similarity 0.89)
- Geometric Multi-Model Fitting With a Convex Relaxation Algorithm (CVPR 2018 · similarity 0.88)
- Multiple Model Fitting as a Set Coverage Problem (CVPR 2016 · similarity 0.85)
- New Bounds on the Cohesion of Complete-link and Other Linkage Methods for Agglomerative Clustering (ICML 2024 · similarity 0.84)
- On the cohesion and separability of average-link for hierarchical agglomerative clustering (NeurIPS 2024 · similarity 0.83)
- PARSAC: Accelerating Robust Multi-Model Fitting with Parallel Sample Consensus (AAAI 2024 · similarity 0.83)
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.