OmniLight: One Model to Rule All Lighting Conditions

2026-04-16

AI Summary

This work addresses the significant degradation in image visibility and color fidelity caused by adverse lighting conditions—such as shadows and non-uniform illumination—which severely hinder computer vision system performance. To tackle this challenge, the authors propose two illumination restoration strategies: a specialized model, DINOLight, and a general-purpose model, OmniLight. Central to their approach is the novel integration of a Wavelet Domain Mixture-of-Experts (WD-MoE) module, enabling unified modeling and efficient restoration across diverse illumination degradation scenarios. The method synergistically combines DINO-based illumination normalization, the WD-MoE mechanism, and a multi-dataset joint training strategy. Evaluated in the NTIRE 2026 Challenge, the proposed models achieved top-ranking results across all three illumination-related tracks, demonstrating substantial improvements in perceptual quality and cross-domain generalization capability.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Large Multimodal Models (LMMs)
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

Adverse lighting conditions, such as cast shadows and irregular illumination, pose significant challenges to computer vision systems by degrading visibility and color fidelity. Consequently, effective shadow removal and ALN are critical for restoring underlying image content, improving perceptual quality, and facilitating robust performance in downstream tasks. However, while achieving state-of-the-art results on specific benchmarks is a primary goal in image restoration challenges, real-world applications often demand robust models capable of handling diverse domains. To address this, we present a comprehensive study on lighting-related image restoration by exploring two contrasting strategies. We leverage a robust framework for ALN, DINOLight, as a specialized baseline to exploit the characteristics of each individual dataset, and extend it to OmniLight, a generalized alternative incorporating our proposed Wavelet Domain Mixture-of-Experts (WD-MoE) that is trained across all provided datasets. Through a comparative analysis of these two methods, we discuss the impact of data distribution on the performance of specialized and unified architectures in lighting-related image restoration. Notably, both approaches secured top-tier rankings across all three lighting-related tracks in the NTIRE 2026 Challenge, demonstrating their outstanding perceptual quality and generalization capabilities. Our codes are available at https://github.com/OBAKSA/Lighting-Restoration.

Problem

• adverse lighting conditions
• shadow removal
• illumination normalization
• image restoration
• lighting-related image degradation

Innovation

• Wavelet Domain Mixture-of-Experts
• OmniLight
• lighting-related image restoration
• generalized model
• adverse lighting normalization