Rethinking Real-world Image Deraining via An Unpaired Degradation-Conditioned Diffusion Model

arXiv:2301.09430v4 Announce Type: replace
Abstract: Recent diffusion models have exhibited great potential in generative modeling tasks. Part of their success can be attributed to the ability of training stable on huge sets of paired synthetic data. However, adapting these models to real-world image deraining remains difficult for two aspects. First, collecting a large-scale paired real-world clean/rainy dataset is unavailable while regular conditional diffusion models heavily rely on paired data for training. Second, real-world rain usually reflects real-world scenarios with a …

abstract arxiv cs.cv data diffusion diffusion model diffusion models generative generative modeling however image modeling part success synthetic synthetic data tasks training type via world