LiWi: Layering in the Wild

TL;DR

This paper introduces LiWi, a framework for high-fidelity natural image decomposition that leverages a novel data synthesis pipeline and joint modeling techniques, achieving state-of-the-art results in layered in-the-wild image separation.

Contribution

The authors propose a new data synthesis pipeline and a joint modeling framework that significantly improve layered image decomposition in natural scenes.

Findings

Achieved state-of-the-art performance in RGB L1 and Alpha IoU metrics.

Constructed a large-scale layered in-the-wild image dataset, LiWi-100k.

Demonstrated effective modeling of illumination effects and boundary accuracy.

Abstract

Recent advances in generative models have empowered impressive layered image generation, yet their success is largely confined to graphic design domains. The layering of in-the-wild images remains an underexplored problem, limiting fine-grained editing and applications of images in real-world scenarios. Specifically, challenges remain in scalable layered data and the modeling of object interaction in natural images, such as illumination effects and structural boundary. To address these bottlenecks, we propose a novel framework for high-fidelity natural image decomposition. First, we introduce an Agent-driven Data Decomposition (ADD) pipeline that orchestrates agents and tools to synthesize layered data without manual intervention. Utilizing this pipeline, we construct a large-scale dataset, named LiWi-100k, with over 100,000 high-quality layered in-the-wild images. Second, we present a novel framework that jointly improves photometric fidelity and alpha boundary accuracy. Specifically, shadow-guided learning explicitly models the illumination effects, and degradation-restoration objective provides boundary-correction supervision by recovering clean foreground image from degraded one. Extensive experiments demonstrate that our framework achieves state-of-the-art (SoTA) performance in natural image decomposition, outperforming existing models in RGB L1 and Alpha IoU metrics. We will soon release our code and dataset.