LEIQ-Assessor: Multi-dimensional Quality Assessment of Low-light Enhanced Images via Multi-task Learning

2026-06-28

AI Summary

This work addresses the lack of reliable no-reference image quality assessment (NR-IQA) methods for low-light enhanced images, which often suffer from noise amplification, color distortion, structural degradation, and overexposure. To overcome these challenges, the study introduces multi-task learning into this domain for the first time, proposing a multidimensional quality assessment model that jointly predicts overall subjective scores along with six perceptual sub-attributes to enhance feature representation. The model employs a pre-trained SigLIP2 Vision Transformer as its backbone and is trained end-to-end using the Pearson Linear Correlation Coefficient (PLCC) loss function. Evaluated on the MLE benchmark, it significantly outperforms existing NR-IQA methods and handcrafted metrics, and achieved second place in the QoMEX 2026 challenge.

Abstract

Low-light image enhancement algorithms (LIEAs) aim to improve the visibility of images captured under poor illumination. However, the enhancement process often introduces artifacts such as noise amplification, color shift, structural damage, and over-exposure, which degrade the perceptual quality of the enhanced images. Therefore, a reliable image quality assessment (IQA) metric for evaluating enhancement effects is of great importance for both the development of LIEAs and their practical applications. In this paper, we present \textbf{LEIQ-Assessor}, a multi-dimensional quality assessment model for low-light image enhancement based on multi-task learning, developed for the QoMEX 2026 Grand Challenge on Low-light Enhanced Image Quality Assessment. Specifically, our method leverages a pre-trained SigLIP2 Vision Transformer as the backbone and simultaneously predicts the overall Mean Opinion Score (MOS) together with six perceptual sub-attributes: lightness, color fidelity, noise level, exposure quality, naturalness, and content recovery. By jointly optimizing these correlated objectives via the PLCC loss, the shared representation captures richer quality-aware features than its single-task counterpart. Experiments on the MLE benchmark demonstrate that LEIQ-Assessor significantly outperforms existing no-reference IQA models and hand-crafted quality descriptors. Our method achieved second place in the QoMEX 2026 Grand Challenge on Low-light Enhanced Image Quality Assessment. The code is available at https://github.com/sunwei925/LEIQ-Assessor.

Problem

• low-light image enhancement
• image quality assessment
• perceptual quality
• artifacts
• no-reference IQA

Innovation

• multi-task learning
• low-light image enhancement
• image quality assessment
• Vision Transformer
• perceptual quality attributes