ASF-Net: Robust video deraining via temporal alignment and online adaptive learning

Xinwei Xue, Jia He, Long Ma, Xiangyu Meng, Wenlin Li, Risheng Liu

Abstract: Highlights•We introduce a novel video deraining paradigm, alignment-shift-fusion.•Constructing a large-scale rainy video dataset, LArge-scale RAiny Video Dataset.•Establishing a correspondence between the synthetic and real-world domains.•Our method outperforms others, demonstrating superior performance and adaptability.