PVRF combines zero-shot VLM-based weather perception with perception-adaptive rectified flow refinement to achieve all-in-one adverse weather removal with improved fidelity and cross-dataset generalization.
Bio-inspired image restoration
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
PVRF: All-in-one Adverse Weather Removal via Prior-modulated and Velocity-constrained Rectified Flow
PVRF combines zero-shot VLM-based weather perception with perception-adaptive rectified flow refinement to achieve all-in-one adverse weather removal with improved fidelity and cross-dataset generalization.