Research map / Detection and segmentation / Open-set and few-shot detection
Model Adaptation with Synthetic and Real Data for Semantic Dense Foggy Scene Understanding
Christos Sakaridis, Dengxin Dai, Simon Hecker, Luc Van Gool
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 |
|---|---|---|---|
| fine-tune on lighter synthetic fog (beta=0.005) instead of denser (beta=0.01), Stereo-DBF (ablation) | RefineNet · Foggy Zurich-test | mIoU all classes: 31.2 -> 33.4 | Improved |
| fine-tune on lighter synthetic fog (beta=0.005) instead of denser, Stereo-GF (ablation) | RefineNet · Foggy Zurich-test | mIoU all classes: 30.2 -> 33.9 | Improved |
| fine-tune on light Stereo-GF synthetic fog vs clear-weather model | RefineNet · Foggy Zurich-test | mIoU all classes: 32.0 -> 33.9 | Improved |
| fine-tune on dense Stereo-GF synthetic fog (beta=0.01) vs clear-weather model | RefineNet · Foggy Zurich-test | mIoU all classes: 32.0 -> 30.2 | Hurt |
| use semantic-aware Stereo-DBF instead of Stereo-GF for dense synthetic fog (beta=0.01) (ablation) | RefineNet · Foggy Zurich-test | mIoU all classes: 30.2 -> 31.2 | Improved |
| use Stereo-DBF instead of Stereo-GF at beta=0.005 (ablation) | RefineNet · Foggy Zurich-test | mIoU all classes: 33.9 -> 33.4 | No clear effect |
| use Stereo-DBF instead of Stereo-GF at beta=0.005 (ablation) | RefineNet · Foggy Driving-dense | mIoU all classes: 32.5 -> 32.8 | Improved |
| add second CMAda step (dense synthetic fog plus light real fog) after step 1, Stereo-DBF (ablation) | RefineNet · Foggy Zurich-test | mIoU all classes: 33.4 -> 37.9 | Improved |
| add second CMAda step after step 1, Stereo-GF (ablation) | RefineNet · Foggy Zurich-test | mIoU all classes: 33.9 -> 34.7 | Improved |
| add second CMAda step after step 1, Stereo-DBF (ablation) | RefineNet · Foggy Zurich-test | mIoU frequent classes: 49.0 -> 56.7 | Improved |
| add second CMAda step (with real Foggy Zurich-light) after step 1, Stereo-DBF (ablation) | RefineNet · Foggy Driving-dense | mIoU all classes: 32.8 -> 34.3 | Improved |
| full CMAda pipeline with Stereo-DBF vs clear-weather RefineNet | RefineNet · Foggy Zurich-test | mIoU all classes and frequent classes: +5.9% and +7.9% | Improved |
Closest papers by meaning
- Learning Generalized Segmentation for Foggy-Scenes by Bi-directional Wavelet Guidance (AAAI 2024 · similarity 0.90)
- Semantic Segmentation on Raindrop Degraded Images Using Two-Stage Dual Teacher-Student Learning (AAAI 2025 · similarity 0.87)
- Parsing All Adverse Scenes: Severity-Aware Semantic Segmentation with Mask-Enhanced Cross-Domain Consistency (AAAI 2024 · similarity 0.86)
- Semantic Segmentation in Multiple Adverse Weather Conditions with Domain Knowledge Retention (AAAI 2024 · similarity 0.86)
- UniMix: Towards Domain Adaptive and Generalizable LiDAR Semantic Segmentation in Adverse Weather (CVPR 2024 · similarity 0.86)
- Synthetic-to-Real Self-supervised Robust Depth Estimation via Learning with Motion and Structure Priors (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.
