Research map / 3D vision / Face and shape
Best Segmentation Buddies for Image-Shape Correspondence
Itai Lang, Dongwei Lyu, Dale Decatur, Rana Hanocka
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 |
|---|---|---|---|
| select the candidate vertex with highest segment IoU instead of random among k=100 (ablation) | BSB · PartNet image-shape | success rate: 0.48 -> 0.74 | Improved |
| use BSB segment-level matching instead of nearest-neighbor vertex | Feature-space matching · PartNet image-shape | success rate: 0.73 -> 0.74 | Improved |
| use BSB instead of nearest-neighbor vertex | Feature-space matching · Perceptual study (1-5) | effectiveness: 3.26 -> 4.63 | Improved |
| increase number of vertex candidates k (ablation) | BSB · PartNet image-shape | success rate: saturates at 0.74 near k = 100 | Improved |
| swap DINOv2 features for DIFT diffusion features (ablation) | BSB · qualitative | correspondence | No clear effect |
| swap SAM for SAM2 as segmentation backbone (ablation) | BSB · qualitative | correspondence | No clear effect |
| use textured shapes instead of untextured (ablation) | BSB · qualitative | correspondence | No clear effect |
| compare BSB with NBB and DIFT (different views) | BSB vs baselines · PartNet image-shape | success rate: 0.64 -> 0.74; 0.39 -> 0.74 | Improved |
Closest papers by meaning
- 3x2: 3D Object Part Segmentation by 2D Semantic Correspondences (ECCV 2024 · similarity 0.87)
- Partial-to-Partial Shape Matching with Geometric Consistency (CVPR 2024 · similarity 0.87)
- EchoMatch: Partial-to-Partial Shape Matching via Correspondence Reflection (CVPR 2025 · similarity 0.86)
- SegMASt3R: Geometry Grounded Segment Matching (NeurIPS 2025 · similarity 0.86)
- Stable-SCore: A Stable Registration-based Framework for 3D Shape Correspondence (CVPR 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.
