IAML: Illumination-Aware Mirror Loss for Progressive Learning in Low-Light Image Enhancement Auto-encoders

2026-03-09

AI Summary

This work addresses the degradation in reconstruction quality caused by illumination variations in low-light image enhancement by proposing a progressive learning approach based on a teacher–student autoencoder architecture. The key innovation lies in the design of an Illumination-Aware Mirror Loss (IAML), which explicitly models illumination changes in the input image during multi-scale feature distillation, enabling layer-wise precise alignment of decoder features from the teacher to the student network. Evaluated on three mainstream low-light image datasets, the proposed method achieves state-of-the-art performance, significantly outperforming existing approaches in terms of SSIM, PSNR, and LPIPS metrics.

Technology Category

• Machine Learning: Deep Generative Models & Autoencoders
• Computer Vision: Learning & Optimization for CV
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

This letter presents a novel training approach and loss function for learning low-light image enhancement auto-encoders. Our approach revolves around the use of a teacher-student auto-encoder setup coupled to a progressive learning approach where multi-scale information from clean image decoder feature maps is distilled into each layer of the student decoder in a mirrored fashion using a newly-proposed loss function termed Illumination-Aware Mirror Loss (IAML). IAML helps aligning the feature maps within the student decoder network with clean feature maps originating from the teacher side while taking into account the effect of lighting variations within the input images. Extensive benchmarking of our proposed approach on three popular low-light image enhancement datasets demonstrate that our model achieves state-of-the-art performance in terms of average SSIM, PSNR and LPIPS reconstruction accuracy metrics. Finally, ablation studies are performed to clearly demonstrate the effect of IAML on the image reconstruction accuracy.

Problem

• low-light image enhancement
• auto-encoders
• feature alignment
• illumination variation
• image reconstruction

Innovation

• Illumination-Aware Mirror Loss
• teacher-student auto-encoder
• progressive learning
• low-light image enhancement
• feature distillation