Research map / Detection and segmentation / Open-set and few-shot detection
Dominant Sets for “Constrained” Image Segmentation
Eyasu Zemene, Leulseged Tesfaye Alemu, Marcello Pelillo
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 |
|---|---|---|---|
| replace Constrained Random Walker with constrained dominant sets (self-tuning sigma) | Constrained Random Walker · GrabCut (scribbles) | error rate (%): 4.1 -> 3.57 | Improved |
| select sigma per database instead of self-tuning | CDS · GrabCut (scribbles) | error rate (%): 3.57 -> 3.80 | Hurt |
| select sigma per image (oracle upper bound) instead of self-tuning (ablation) | CDS · GrabCut (scribbles) | error rate (%): 3.57 -> 2.72 | Improved |
| replace MILCut-Struct bounding-box method with CDS (self-tuning) | MILCut-Struct · Berkeley | Jaccard index: 84 -> 93 | Improved |
| select sigma per image (oracle) instead of self-tuning (ablation) | CDS · Berkeley | Jaccard index: 93 -> 95 | Improved |
| increase bounding-box looseness from L=0 to L=600 for GrabCut | GrabCut · GrabCut loose boxes | error rate (%): 7.4 -> 13.7 | Hurt |
| increase bounding-box looseness from L=0 to L=600 for CDS single sigma | CDS Single Sigma · GrabCut loose boxes | error rate (%): 7.48 -> 6.29 | Improved |
| replace LooseCut with CDS single sigma at L=600 | LooseCut · GrabCut loose boxes | error rate (%): 6.8 -> 6.29 | Improved |
| replace LooseCut with CDS self-tuning at L=0 | LooseCut · GrabCut loose boxes | error rate (%): 7.9 -> 7.54 | Improved |
| replace BNC with constrained dominant sets for interactive co-segmentation | BNC · Image pairs | F-measure: 0.6564 -> 0.7140 | Improved |
| replace BNC with constrained dominant sets for interactive co-segmentation | BNC · Image pairs | recall: 0.8512 -> 0.8208 | No clear effect |
| increase share of erroneous scribbles from 0% to 100% | CDS · GrabCut synthetic scribbles | error rate | No clear effect |
Closest papers by meaning
- Interactive Image Segmentation Using Constrained Dominant Sets (ECCV 2016 · similarity 0.96)
- Convexity Shape Constraints for Image Segmentation (CVPR 2016 · similarity 0.88)
- Convexity Shape Prior for Binary Segmentation (TPAMI 2016 · similarity 0.86)
- Laplacian Coordinates: Theory and Methods for Seeded Image Segmentation (TPAMI 2020 · similarity 0.86)
- Sparse Layered Graphs for Multi-Object Segmentation (CVPR 2020 · similarity 0.86)
- Hierarchy-Agnostic Unsupervised Segmentation: Parsing Semantic Image Structure (NeurIPS 2024 · similarity 0.86)
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.
