Separate In Latent Space: Unsupervised Single Image Layer Separation

TL;DR

This paper introduces an unsupervised approach for single image layer separation that leverages latent space assumptions and GANs, eliminating the need for ground truth triplet data and outperforming existing methods.

Contribution

It proposes a novel unsupervised layer separation pipeline based on latent space assumptions and GANs, enabling effective separation without ground truth triplet data.

Findings

Outperforms existing unsupervised methods on synthetic and real data

Successfully handles multi-layer separation tasks

Demonstrates effectiveness without ground truth triplet data

Abstract

Many real world vision tasks, such as reflection removal from a transparent surface and intrinsic image decomposition, can be modeled as single image layer separation. However, this problem is highly ill-posed, requiring accurately aligned and hard to collect triplet data to train the CNN models. To address this problem, this paper proposes an unsupervised method that requires no ground truth data triplet in training. At the core of the method are two assumptions about data distributions in the latent spaces of different layers, based on which a novel unsupervised layer separation pipeline can be derived. Then the method can be constructed based on the GANs framework with self-supervision and cycle consistency constraints, etc. Experimental results demonstrate its successfulness in outperforming existing unsupervised methods in both synthetic and real world tasks. The method also shows its ability to solve a more challenging multi-layer separation task.