Research map / Label-efficient and robust learning / Domain adaptation and generalization
PHATNet: A Physics-guided Haze Transfer Network for Domain-adaptive Real-world Image Dehazing
Fu-Jen Tsai, Yan-Tsung Peng, Yen-Yu Lin, Chia-Wen Lin
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 |
|---|---|---|---|
| add PHATNet-generated fine-tuning data | FocalNet · Setting1 average | PSNR (dB): 16.25 -> 17.07 | Improved |
| add PHATNet-generated fine-tuning data | SGDN · Setting1 average | PSNR (dB): 14.84 -> 16.77 | Improved |
| add PHATNet-generated fine-tuning data | MITNet · Setting2 average | PSNR (dB): 12.44 -> 15.79 | Improved |
| add PHATNet-generated fine-tuning data | SGDN · Setting2 average | PSNR (dB): 12.99 -> 16.17 | Improved |
| add PHATNet-generated fine-tuning data | Dehamer · Setting2 NH-Haze21 | SSIM: 0.748 -> 0.747 | No clear effect |
| add PHATNet-generated fine-tuning data | FocalNet · Setting1 DenseHaze | SSIM: 0.434 -> 0.423 | Hurt |
| add PHATNet-generated fine-tuning data | SGDN · RTTS (source NH-Haze20) | BRISQUE: 34.27 -> 22.21 | Improved |
| add PHATNet-generated fine-tuning data | FocalNet · RTTS (source NH-Haze20) | NIQE: 4.19 -> 3.92 | Improved |
| add both ALE and TME encoders instead of ALE only (ablation) | PHATNet · Setting1 average, FocalNet | PSNR (dB): 16.46 -> 17.07 | Improved |
| add both ALE and TME encoders instead of TME only (ablation) | PHATNet · Setting1 average, FocalNet | PSNR (dB): 16.80 -> 17.07 | Improved |
| use ASM-guided ALE+TME fusion instead of feature concatenation (ablation) | PHATNet · Setting1 average, FocalNet | PSNR (dB): 16.62 -> 17.07 | Improved |
| increase PHDT stages from 1 to 3 (ablation) | PHATNet · Setting1 average, FocalNet | PSNR (dB): 16.68 -> 17.07 | Improved |
Closest papers by meaning
- Learning Hazing to Dehazing: Towards Realistic Haze Generation for Real-World Image Dehazing (CVPR 2025 · similarity 0.91)
- Adaptive Dynamic Dehazing via Instruction-Driven and Task-Feedback Closed-Loop Optimization for Diverse Downstream Task Adaptation (AAAI 2026 · similarity 0.88)
- PromptHaze: Prompting Real-world Dehazing via Depth Anything Model (AAAI 2025 · similarity 0.88)
- CoA: Towards Real Image Dehazing via Compression-and-Adaptation (CVPR 2025 · similarity 0.88)
- PRISM: Rethinking Atmospheric Scattering Reconstruction as a Unified Understanding and Restoration Model for Real-world Dehazing (NeurIPS 2026 · similarity 0.88)
- Real-world Image Dehazing with Coherence-based Pseudo Labeling and Cooperative Unfolding Network (NeurIPS 2024 · similarity 0.87)
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.
