Rain to Rain: Learning Real Rain Removal without Ground Truth

Image deraining is a low-level restoration task that has become quite popular during the past decades. Although recent data-driven models exhibit promising results, most of these are trained on synthetic rain data sets which do not generalize well when applied to real images. While real-rain have achieved favorable generalization performance, generating rain-free ground-truths can be tedious and time-consuming. To address this problem, in work, we present training, an unsupervised training method for single image deraining. Our experiments show it possible train by using only This simply map pairs We also introduce idea least overlapping pairs, selecting adequate enables achieve equivalent performance compared supervised training.