RevealLayer: Disentangling Hidden and Visible Layers via Occlusion-Aware Image Decomposition

TL;DR

RevealLayer is a diffusion-based framework that improves natural image layer decomposition by disentangling hidden and visible layers, accurately recovering occluded content, and providing a new high-quality dataset and benchmark.

Contribution

It introduces a novel diffusion-based method with region-aware attention and occlusion-guided components, along with a new dataset and benchmark for multi-layer natural images.

Findings

RevealLayer outperforms existing methods in layer decomposition accuracy.

The RevealLayer-100K dataset enables better training and evaluation.

Extensive experiments validate the effectiveness of the proposed approach.

Abstract

Recent diffusion-based approaches have made substantial progress in image layer decomposition. However, accurately decomposing complex natural images remains challenging due to difficulties in occlusion completion, robust layer disentanglement, and precise foreground boundaries. Moreover, the scarcity of high-quality multi-layer natural image datasets limits advancement. To address these challenges, we propose RevealLayer, a diffusion-based framework that decomposes an RGB image into multiple RGBA layers, enabling precise layer separation and reliable recovery of occluded content in natural images. RevealLayer incorporates three key components: (1) a Region-Aware Attention module to disentangle hidden and visible layers; (2) an Occlusion-Guided Adapter to leverage contextual information to enhance overlapping regions; and (3) a composite loss to enforce sharp alpha boundaries and suppress residual artifacts. To support training and evaluation, we introduce RevealLayer-100K, a high-quality multi-layer natural image constructed through a collaboration between automated algorithms and human annotation, and further establish RevealLayerBench for benchmarking layer decomposition in general natural scenes. Extensive experiments demonstrate that RevealLayer consistently outperforms existing approaches in layer decomposition.