NightIQA: A Comprehensive Dataset and Benchmark for MLLM-Based Nighttime Image Quality Assessment
作者: Yiding Tian, Jiaojiao Yi, Jiayuan Lu, Pufan Xu, Jinlin Wu, Changyuan Wen, Hongyuan Yu
分类: cs.CV
发布日期: 2026-11-03
备注: Accepted to WACV 2027. 25 pages including supplementary material
💡 一句话要点
提出NightIQA以解决夜间图像质量评估问题
🎯 匹配领域: 支柱九:具身大模型 (Embodied Foundation Models)
关键词: 夜间图像 图像质量评估 多模态大语言模型 数据集构建 监督学习 合成图像 性能提升
📋 核心要点
- 现有的图像质量评估方法主要集中在白天图像,缺乏针对夜间图像的多维度评估,导致评估结果不准确。
- 本文提出NightIQA数据集,包含真实和合成的夜间图像,结合多模态大语言模型进行质量评估,填补了现有数据集的空白。
- 通过对Qwen3-VL-8B的微调,实验结果显示其在夜间图像质量评估中表现优于现有开源方法,提升了整体准确性。
📝 摘要(中文)
夜间图像由于复杂的降质因素,质量评估面临巨大挑战。现有的基于分数的图像质量评估方法无法满足夜间图像的多维度评估需求。为此,本文提出了NightIQA数据集和基准,包含1,769张高分辨率真实图像、1,078个多项选择题以及80,000张合成图像,旨在为多模态大语言模型提供支持。通过对Qwen3-VL-8B的微调,实验结果表明,该模型在夜间图像质量评估中超越了现有的开源方法,展现了领域特定数据的重要性。
🔬 方法详解
问题定义:本文旨在解决夜间图像质量评估中的数据不足和评估方法单一的问题。现有方法主要集中在白天图像,缺乏针对夜间图像的多维度评估,导致评估结果的准确性和可靠性不足。
核心思路:提出NightIQA数据集和基准,结合多模态大语言模型(MLLM)进行夜间图像质量评估。通过构建真实和合成的夜间图像数据集,提供丰富的多维度标注,提升评估的全面性和准确性。
技术框架:NightIQA包含三个核心组件:NightIQA-Real(1,769张真实图像)、NightIQA-Bench(1,078个多项选择题)和NightIQA-Syn(80,000张合成图像)。这些组件共同构成了一个完整的评估体系,支持多模态学习。
关键创新:最重要的创新在于构建了一个专门针对夜间图像的多模态数据集,填补了现有研究中的数据空白,并通过结构化的监督学习提升了模型的评估能力。
关键设计:在模型微调过程中,采用了监督微调(SFT)和组相对策略优化(GRPO),并针对夜间图像的特性设计了特定的损失函数和网络结构,以提高评估的准确性和鲁棒性。
🖼️ 关键图片
📊 实验亮点
实验结果表明,经过微调的Qwen3-VL-8B在夜间图像质量评估中超越了现有的开源方法,整体准确性显著提高,具体性能数据未明确给出,但相较于领先的闭源模型也表现出更高的评估能力,验证了领域特定数据的重要性。
🎯 应用场景
该研究的潜在应用领域包括夜间监控、交通监测和夜间摄影等场景。通过提供准确的夜间图像质量评估工具,能够有效提升相关领域的图像处理和分析能力,具有重要的实际价值和广泛的应用前景。
📄 摘要(原文)
Nighttime images are inherently susceptible to complex, compound degradations, making accurate quality assessment exceptionally challenging. Score-based IQA methods condense quality into a single number, lacking the multi-dimensional, language-grounded reasoning required for nighttime evaluation. Multimodal Large Language Models (MLLMs) offer a more comprehensive alternative, yet remain constrained by a critical data gap: existing real-world IQA datasets are overwhelmingly daytime-centric and often built from outdated captures, while synthetic alternatives lack nighttime-specific degradation modeling, yielding artifacts far removed from authentic nighttime conditions. To bridge this gap, we present NightIQA, a dataset and benchmark suite purpose-built for MLLM-based nighttime IQA. It comprises three core components: NightIQA-Real, a real-world dataset of 1,769 high-resolution images that we captured ourselves with modern equipment, encompassing six canonical distortion types and structured multi-dimensional annotations; NightIQA-Bench, a benchmark of 1,078 visually grounded multiple-choice questions tailored to MLLM-based nighttime IQA, organized around a four-level evaluation hierarchy spanning Perception, Technical Assessment, Localization, and Reasoning; and NightIQA-Syn, an inherently scalable, scene-aware synthesis pipeline that jointly models global-level and instance-level degradations, with an initial release of 80k high-fidelity images alongside the full generation code. Fine-tuned on NightIQA with Supervised Fine-Tuning (SFT) and Group Relative Policy Optimization (GRPO), Qwen3-VL-8B outperforms open-source IQA methods and achieves higher overall accuracy than leading closed-source models, highlighting the value of domain-specific data and structured supervision.