Research map / Label-efficient and robust learning / Recognition and long-tail
Detect, Replace, Refine: Deep Structured Prediction for Pixel Wise Labeling
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 |
|---|---|---|---|
| Apply Detect + Replace + Refine for a second iteration (ablation) | Detect + Replace + Refine · Synthetic dataset test | EPE (All): 1.8569 -> 1.6899 | Helped |
| Apply Detect + Replace + Refine for a second iteration (ablation) | Detect + Replace + Refine · Middlebury dataset | EPE (All): 2.368 -> 2.321 | Helped |
| Apply Detect + Replace + Refine for a second iteration (ablation) | Detect + Replace + Refine · KITTI 2015 validation | EPE (All): 0.858 -> 0.835 | Helped |
| Use Detect + Replace + Refine instead of Replace + Refine (ablation) | Dense disparity prediction · Synthetic dataset test | EPE (All): 2.2356 -> 1.8569 | Helped |
| Use Detect + Replace + Refine instead of Replace + Refine (ablation) | Dense disparity prediction · Middlebury dataset | EPE (All): 2.908 -> 2.368 | Helped |
| Use Detect + Replace + Refine instead of Replace + Refine (ablation) | Dense disparity prediction · KITTI 2015 validation | EPE (All): 0.865 -> 0.858 | Helped |
| Replace the sequential Detect + Replace + Refine steps with parallel Replace and Refine (ablation) | Dense disparity prediction · Synthetic dataset test | EPE (All): 1.8569 -> 2.0241 | Hurt |
| Replace the sequential Detect + Replace + Refine steps with parallel Replace and Refine (ablation) | Dense disparity prediction · Middlebury dataset | EPE (All): 2.368 -> 2.552 | Hurt |
| Replace the sequential Detect + Replace + Refine steps with parallel Replace and Refine (ablation) | Dense disparity prediction · KITTI 2015 validation | EPE (All): 0.858 -> 0.962 | Hurt |
| Remove the image input and use initial labels alone (ablation) | Detect + Replace + Refine · Synthetic dataset test | EPE (All): 1.8569 -> 3.8810 | Hurt |
| Remove the image input and use initial labels alone (ablation) | Detect + Replace + Refine · Middlebury dataset | EPE (All): 2.368 -> 3.656 | Hurt |
| Remove the image input and use initial labels alone (ablation) | Detect + Replace + Refine · KITTI 2015 validation | EPE (All): 0.858 -> 0.966 | Hurt |
Closest papers by meaning
- L2B: Learning to Bootstrap Robust Models for Combating Label Noise (CVPR 2024 · similarity 0.86)
- Diffusion Model for Dense Matching (ICLR 2024 · similarity 0.86)
- Directional Label Diffusion Model for Learning from Noisy Labels (CVPR 2025 · similarity 0.86)
- Doomed to Re-Annotate, Forever: The ImageNet Story (NeurIPS 2026 · similarity 0.86)
- Bayesian-guided Label Mapping for Visual Reprogramming (NeurIPS 2024 · similarity 0.85)
- Stereo Risk: A Continuous Modeling Approach to Stereo Matching (ICML 2024 · 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.