ReLo-IRR: Reflection-Guided LoRA Framework for Image Reflection Removal

2026-07-03

AI Summary

Existing single-image reflection removal methods typically apply a uniform processing strategy regardless of reflection intensity, often resulting in either residual strong reflections or loss of fine details in weak reflections. To address this limitation, this work proposes a rectified flow-based, reflection-guided LoRA framework that adaptively modulates the denoising process. Specifically, a lightweight estimator predicts an image-dependent reflection intensity descriptor, which—combined with time-conditioned embeddings—enables dynamic, intensity-aware LoRA modulation during denoising. This approach represents the first effort to explicitly integrate reflection intensity priors with temporal dynamics in the denoising trajectory, significantly enhancing both reflection removal performance and generalization capability. Extensive experiments demonstrate its effectiveness in suppressing diverse reflection artifacts across multiple challenging benchmarks.

Abstract

Single-image reflection removal (SIRR) aims to recover the clean transmission layer from a reflection-contaminated image. Although recent methods achieve promising results with large diffusion models, they rely on image-agnostic adaptation strategies, e.g., fine-tuning or ControlNet, that enforce uniform suppression regardless of reflection severity. As a result, heavy reflections often leave residuals, while weak ones suffer from detail loss. To this end, we propose ReLo-IRR, a reflection-guided LoRA framework built upon the rectified flow model. First, a lightweight estimator is designed to predict the reflection strength descriptor, providing an explicit prior of reflection dominance for each image and enabling image-dependent LoRA modulation. Second, we introduce a time-conditioned mechanism that fuses this reflection descriptor with timestep embeddings, enabling LoRA modulation to evolve consistently with the coarse-to-fine denoising process. By jointly modeling reflection strength and denoising dynamics, our ReLo-IRR achieves robust suppression of diverse reflection conditions. Extensive experiments on challenging benchmarks validate the effectiveness of ReLo-IRR, demonstrating superior dereflection performance and robust generalization. The code is released at https://github.com/KONGBAI-8080/ReLo-IRR.

Problem

• Single-image reflection removal
• Reflection suppression
• Transmission layer recovery
• Reflection severity
• Image restoration

Innovation

• Reflection Removal
• LoRA
• Rectified Flow
• Image-Adaptive Modulation
• Time-Conditioned Denoising