# Mask-ShadowGAN: Learning to Remove Shadows from Unpaired Data

## TL;DR

This paper introduces Mask-ShadowGAN, a novel deep learning framework that effectively removes shadows from images using unpaired data by jointly learning shadow masks and shadow removal, overcoming limitations of traditional cycle-consistency methods.

## Contribution

The paper proposes Mask-ShadowGAN, which automatically learns shadow masks and shadow removal simultaneously from unpaired data, improving shadow removal performance without requiring paired datasets.

## Key findings

- Effective shadow removal demonstrated on unpaired datasets.
- Outperforms existing methods in shadow removal quality.
- Automatically learns shadow masks alongside shadow removal.

## Abstract

This paper presents a new method for shadow removal using unpaired data, enabling us to avoid tedious annotations and obtain more diverse training samples. However, directly employing adversarial learning and cycle-consistency constraints is insufficient to learn the underlying relationship between the shadow and shadow-free domains, since the mapping between shadow and shadow-free images is not simply one-to-one. To address the problem, we formulate Mask-ShadowGAN, a new deep framework that automatically learns to produce a shadow mask from the input shadow image and then takes the mask to guide the shadow generation via re-formulated cycle-consistency constraints. Particularly, the framework simultaneously learns to produce shadow masks and learns to remove shadows, to maximize the overall performance. Also, we prepared an unpaired dataset for shadow removal and demonstrated the effectiveness of Mask-ShadowGAN on various experiments, even it was trained on unpaired data.