Research map / Label-efficient and robust learning / Test-time adaptation
Prompt-Based Test-Time Real Image Dehazing: A Novel Pipeline
Zixuan Chen, Zewei He, Ziqian Lu, Xuecheng Sun, Zheming Lu
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 |
|---|---|---|---|
| apply PTTD test-time prompt and feature adaptation to pretrained dehazing model | AECRNet · O-HAZE | PSNR: 17.24 -> 20.23 | Improved |
| apply PTTD | AECRNet · Dense-Haze | PSNR: 8.59 -> 13.67 | Improved |
| apply PTTD | Dehazeformer · I-HAZE | SSIM: 0.8184 -> 0.8186 | No clear effect |
| apply PTTD | AECRNet · RTTS | FADE: 1.285 -> 0.7120 | Improved |
| apply PTTD | NAFNet · Fattal's dataset | FADE: 0.4183 -> 0.4289 | Hurt |
| add partition strategy to ILN prompt generation (ablation) | PTTD PGM · O-HAZE (AECRNet) | PSNR: 16.90 -> 18.28 | Improved |
| replace ILN with color balanced ILN (CBILN) (ablation) | PTTD PGM · O-HAZE (AECRNet) | PSNR: 18.28 -> 19.46 | Improved |
| use ILN prompt or raw y_s without partition (ablation) | PTTD PGM · O-HAZE (AECRNet) | PSNR: 17.24 -> 16.90 / 16.70 | Hurt |
| use FLN instead of adaIN in feature adaptation (ablation) | PTTD FAM · O-HAZE (AECRNet) | PSNR: 19.46 -> 20.23 | Improved |
| use a random haze-free y_s instead of an elaborately selected one (ablation) | AECRNet-PTTD · O-HAZE | PSNR: 20.23 -> 19.51 | Hurt |
| apply PTTD to AECRNet at inference | AECRNet · Inference time 512x512 | runtime (ms): 23.48 -> 38.03 | Hurt |
Closest papers by meaning
- PromptHaze: Prompting Real-world Dehazing via Depth Anything Model (AAAI 2025 · similarity 0.92)
- Learning Hazing to Dehazing: Towards Realistic Haze Generation for Real-World Image Dehazing (CVPR 2025 · similarity 0.90)
- Adaptive Dynamic Dehazing via Instruction-Driven and Task-Feedback Closed-Loop Optimization for Diverse Downstream Task Adaptation (AAAI 2026 · similarity 0.88)
- SelfPromer: Self-Prompt Dehazing Transformers with Depth-Consistency (AAAI 2024 · similarity 0.88)
- Dehaze-RetinexGAN: Real-World Image Dehazing via Retinex-based Generative Adversarial Network (AAAI 2025 · similarity 0.88)
- Exploiting Diffusion Prior for Real-World Image Dehazing with Unpaired Training (AAAI 2025 · similarity 0.88)
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.
