FlowIID: Single-Step Intrinsic Image Decomposition via Latent Flow Matching

2026-01-18

AI Summary

This work addresses the challenge of high parameter counts in existing intrinsic image decomposition methods, which hinder deployment in resource-constrained or real-time settings. To this end, we introduce flow matching—a technique previously unexplored in this domain—and propose a novel single-step decomposition framework operating in a variational autoencoder (VAE)-guided latent space. By jointly optimizing the latent representation and the flow matching module, our method efficiently and stably disentangles albedo and shading in a single inference step. Extensive experiments on multiple benchmark datasets demonstrate that the proposed model achieves decomposition performance comparable to or better than state-of-the-art approaches while using significantly fewer parameters. This favorable balance of efficiency and accuracy makes our approach particularly suitable for real-time and lightweight applications.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Deep Generative Models & Autoencoders
• Intelligent Robots: State Estimation

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for incomplete, noisy, or partially observed Web-related graphs
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web

Abstract

Intrinsic Image Decomposition (IID) separates an image into albedo and shading components. It is a core step in many real-world applications, such as relighting and material editing. Existing IID models achieve good results, but often use a large number of parameters. This makes them costly to combine with other models in real-world settings. To address this problem, we propose a flow matching-based solution. For this, we design a novel architecture, FlowIID, based on latent flow matching. FlowIID combines a VAE-guided latent space with a flow matching module, enabling a stable decomposition of albedo and shading. FlowIID is not only parameter-efficient, but also produces results in a single inference step. Despite its compact design, FlowIID delivers competitive and superior results compared to existing models across various benchmarks. This makes it well-suited for deployment in resource-constrained and real-time vision applications.

Problem

• Intrinsic Image Decomposition
• parameter efficiency
• real-time vision
• model deployment
• albedo and shading

Innovation

• Flow Matching
• Intrinsic Image Decomposition
• Latent Space
• Parameter-Efficient
• Single-Step Inference