PVRF: All-in-one Adverse Weather Removal via Prior-modulated and Velocity-constrained Rectified Flow
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2026
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866910218581966848 |
|---|---|
| author | Dong, Wei Zhou, Han Ji, Terry Zhao, Guanhua Asoodeh, Shahab Zhang, Yulun Zhai, Guangtao Chen, Jun Liu, Xiaohong |
| author_facet | Dong, Wei Zhou, Han Ji, Terry Zhao, Guanhua Asoodeh, Shahab Zhang, Yulun Zhai, Guangtao Chen, Jun Liu, Xiaohong |
| contents | Adverse weather removal (AWR) in real-world images remains challenging due to heterogeneous and unseen degradations, while distortion-driven training often yields overly smooth results. We propose PVRF, a unified framework that integrates zero-shot soft weather perceptions with velocity-constrained rectified-flow refinement. PVRF introduces an AWR-specific question answering module (AWR-QA) that uses frozen vision--language models (VLMs) to estimate soft probabilities of weather types and low-level attribute scores. These perceptions condition restoration networks via attribute-modulated normalization (AMN) and weather-weighted adapters (WWA), producing an anchor estimate for refinement. We then learn a terminal-consistent residual rectified flow with perception-adaptive source perturbation and a terminal-consistent velocity parameterization to stabilize learning near the terminal regime. Extensive experiments show that PVRF improves both fidelity and perceptual quality over state-of-the-art baselines, with strong cross-dataset generalization on single and combined degradations. Code will be released at https://github.com/dongw22/PVRF. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_14045 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | PVRF: All-in-one Adverse Weather Removal via Prior-modulated and Velocity-constrained Rectified Flow Dong, Wei Zhou, Han Ji, Terry Zhao, Guanhua Asoodeh, Shahab Zhang, Yulun Zhai, Guangtao Chen, Jun Liu, Xiaohong Computer Vision and Pattern Recognition Adverse weather removal (AWR) in real-world images remains challenging due to heterogeneous and unseen degradations, while distortion-driven training often yields overly smooth results. We propose PVRF, a unified framework that integrates zero-shot soft weather perceptions with velocity-constrained rectified-flow refinement. PVRF introduces an AWR-specific question answering module (AWR-QA) that uses frozen vision--language models (VLMs) to estimate soft probabilities of weather types and low-level attribute scores. These perceptions condition restoration networks via attribute-modulated normalization (AMN) and weather-weighted adapters (WWA), producing an anchor estimate for refinement. We then learn a terminal-consistent residual rectified flow with perception-adaptive source perturbation and a terminal-consistent velocity parameterization to stabilize learning near the terminal regime. Extensive experiments show that PVRF improves both fidelity and perceptual quality over state-of-the-art baselines, with strong cross-dataset generalization on single and combined degradations. Code will be released at https://github.com/dongw22/PVRF. |
| title | PVRF: All-in-one Adverse Weather Removal via Prior-modulated and Velocity-constrained Rectified Flow |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2605.14045 |