DERAINIZER for Unpaired Single Image De-raining

Abstract

Rain streaks are known to cause many performance or accuracy reductions in many computer vision systems. The problem of obstructing rain streaks is addressed within the domain of “De-raining”. De-raining is the process of removing rain streaks. It has two main sub-categories which are video de-raining and single image de-raining. Many works on single image de-raining have taken place using both traditional approaches as well as deep learning approaches. This paper does a critical analysis of credible, novel, and best performing single image de-raining systems. Also, the paper addresses an unpaired training gap which exists within the domain with the novel de-raining system ”DERAINIZER”, enabling all future researchers to employ unpaired training data to train their models. It is followed by results, research limitations and possible future works.