Utilizing Multi-step Loss for Single Image Reflection Removal

TL;DR

This paper introduces a novel multi-step loss training technique for single image reflection removal, utilizing a synthetic dataset and depth features to improve performance over existing methods.

Contribution

It proposes a new training method with multi-step loss, a synthetic dataset RefGAN, and the use of depth maps to enhance reflection removal from a single image.

Findings

Outperforms state-of-the-art models on SIR^2 benchmark

Effective use of synthetic dataset RefGAN improves learning

Multi-step loss enhances reflection removal accuracy

Abstract

Image reflection removal is crucial for restoring image quality. Distorted images can negatively impact tasks like object detection and image segmentation. In this paper, we present a novel approach for image reflection removal using a single image. Instead of focusing on model architecture, we introduce a new training technique that can be generalized to image-to-image problems, with input and output being similar in nature. This technique is embodied in our multi-step loss mechanism, which has proven effective in the reflection removal task. Additionally, we address the scarcity of reflection removal training data by synthesizing a high-quality, non-linear synthetic dataset called RefGAN using Pix2Pix GAN. This dataset significantly enhances the model's ability to learn better patterns for reflection removal. We also utilize a ranged depth map, extracted from the depth estimation of the ambient image, as an auxiliary feature, leveraging its property of lacking depth estimations for reflections. Our approach demonstrates superior performance on the SIR^2 benchmark and other real-world datasets, proving its effectiveness by outperforming other state-of-the-art models.