Real-World Adverse Weather Image Restoration via Dual-Level Reinforcement Learning with High-Quality Cold Start
Fuyang Liu, Jiaqi Xu, Xiaowei Hu · 2025
Read Real-World Adverse Weather Image Restoration via Dual-Level Reinforcement Learning with High-Quality Cold Start by Fuyang Liu, Jiaqi Xu, Xiaowei Hu (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-05095.txt.
Description
Fuyang Liu and co-authors present a two-level learning method to restore photos spoiled by bad weather. Rain, fog, and snow blur real scenes, and models trained on fake damage often fail on the mixed harm found in the wild.
First they built HFLS-Weather, a physics-based data set that mimics many weather harms with high fidelity, to give training a strong cold start. Then local models for each weather type are tuned by reward, while a global controller picks and blends them. The aim is clear images with no need for paired clean shots.
Catalog record
| CREATOR | Fuyang Liu, Jiaqi Xu, Xiaowei Hu |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 34 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2511-05095