Research map / Graph learning / GNN architectures
Efficient Graph Cut Optimization for Full CRFs with Quantized Edges
Olga Veksler
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 |
|---|---|---|---|
| use expansion graph cut instead of mean field | Quantized-edge Full CRF · Multilabel Full CRF energy | relative energy increase | Improved |
| use graph cut instead of mean field at large smoothness | Binary Full CRF · lambda=0.1 vs 2 | relative energy increase: 5.3% -> 40% (mean field) | Improved |
| add Full CRF optimization to unary terms | CNN unary · PASCAL VOC 2012 test | IOU: 67.143 -> 67.7484 | Improved |
| add Full CRF optimization on top of superpixel labeling (ablation) | superpixel labeling · PASCAL VOC 2012 test | IOU: 65.8899 -> 67.7484 | Improved |
| use our optimization instead of mean field | Quantized-edge Full CRF · PASCAL VOC 2012 test | IOU: 67.3 -> 67.7484 | Improved |
| use our method instead of mean field | Quantized-edge Full CRF · Binary Full CRF running time | running time (s): 0.012 -> 0.31 | Hurt |
| use our method instead of exact graph cut | Binary Full CRF · 70x70 images | running time (s): 7.1 -> 0.31 | Improved |
| use our method instead of superpixel ICM at large lambda | Multilabel Full CRF · larger lambda | relative energy increase | Improved |
Closest papers by meaning
- Efficient Linear Programming for Dense CRFs (CVPR 2017 · similarity 0.89)
- Efficient Continuous Relaxations for Dense CRF (ECCV 2016 · similarity 0.89)
- Regularized Frank-Wolfe for Dense CRFs: Generalizing Mean Field and Beyond (NeurIPS 2021 · similarity 0.87)
- Fast, Exact and Multi-scale Inference for Semantic Image Segmentation with Deep Gaussian CRFs (ECCV 2016 · similarity 0.87)
- Dense and Low-Rank Gaussian CRFs Using Deep Embeddings (ICCV 2017 · similarity 0.87)
- Soft Self-labeling and Potts Relaxations for Weakly-supervised Segmentation (CVPR 2025 · 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.
