Research map / Detection and segmentation / Weakly supervised detection and segmentation
Semantic Foggy Scene Understanding with Synthetic Data
Christos Sakaridis, Dengxin Dai, 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 synthetic Foggy Cityscapes-refined | Dilation10 · Foggy Driving | mean IoU (%): 34.9 -> 37.8 | Improved |
| fine-tune on synthetic fog with DCP dehazing | Dilation10 · Foggy Driving | mean IoU (%): 29.9 -> 37.4 | Improved |
| add guided filtering to fog simulation (poly l.r.) (ablation) | Our fog simulation · Foggy Driving | mean IoU (%): 36.8 -> 37.8 | Improved |
| replace nearest-neighbor depth completion with our pipeline (ablation) | fog simulation · Foggy Driving | mean IoU (%): 36.2 -> 37.8 | Improved |
| use poly learning rate instead of constant (ablation) | fine-tuning on Foggy Cityscapes-refined · Foggy Driving | mean IoU (%): 34.4 -> 37.8 | Improved |
| add dehazing preprocessing (MSCNN) to fine-tuned model, beta 0.01 (ablation) | Dilation10 FT · Foggy Driving | mean IoU (%): 37.8 -> 37.1 | No clear effect |
| add DCP dehazing preprocessing, beta 0.01 (ablation) | Dilation10 FT · Foggy Driving | mean IoU (%): 37.8 -> 37.4 | Hurt |
| add non-local dehazing preprocessing, beta 0.01 (ablation) | Dilation10 FT · Foggy Driving | mean IoU (%): 37.8 -> 36.6 | Hurt |
| add MSCNN dehazing with light fog beta 0.005 (ablation) | Dilation10 FT · Foggy Driving | mean IoU (%): 37.6 -> 38.3 | Improved |
| increase training fog density from beta 0.01 to 0.02 (ablation) | Dilation10 FT, no dehazing · Foggy Driving | mean IoU (%): 37.8 -> 36.1 | Hurt |
| fine-tune on synthetic fog beta 0.005 | Fast R-CNN · Foggy Driving | AP (%) mean frequent: 20.4 -> 22.8 | Improved |
| fine-tune on synthetic fog beta 0.005 | Fast R-CNN · Foggy Driving | AP (%) person: 10.3 -> 10.3 | No clear effect |
Closest papers by meaning
- Model Adaptation with Synthetic and Real Data for Semantic Dense Foggy Scene Understanding (ECCV 2018 · similarity 0.93)
- Curriculum Model Adaptation with Synthetic and Real Data for Semantic Foggy Scene Understanding (IJCV 2019 · similarity 0.93)
- Both Style and Fog Matter: Cumulative Domain Adaptation for Semantic Foggy Scene Understanding (CVPR 2022 · similarity 0.90)
- Learning Generalized Segmentation for Foggy-Scenes by Bi-directional Wavelet Guidance (AAAI 2024 · similarity 0.90)
- SynFog: A Photo-realistic Synthetic Fog Dataset based on End-to-end Imaging Simulation for Advancing Real-World Defogging in Autonomous Driving (CVPR 2024 · similarity 0.87)
- Semantic Segmentation on Raindrop Degraded Images Using Two-Stage Dual Teacher-Student Learning (AAAI 2025 · 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.
