HDR-VLM：面向 HDR 视频调色的视觉语言模型 HDR 域适配与偏好对齐质量评估
HDR-VLM: HDR-Domain Adaptation of VLMs and Preference-Aligned Quality Assessment for HDR Video Color Grading
把预训练 VLM 从 SDR 域迁移到 HDR 域的两阶段范式（HLG 统一编码 + 渐进适配 + 课程式奖励强化学习对齐人类偏好）可直接复用到其他「预训练模型域不匹配」的感知质量评估任务，其归因理由输出也为质量评估提供了可解释性接口。
原文依据 it first bridges the domain gap using a unified HLG-based encoding and progressive adaptation
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论调色是高动态范围（HDR）视频制作的核心环节，它决定了内容在各类显示设备上的感知色调、对比度与亮度。然而，由于 HDR 调色质量评估具有语义性、依赖内容的特点，且缺乏大规模标注数据，其评估尤为困难。尽管预训练视觉语言模型（VLM）具备强大的语义先验与泛化能力，但其接触的数据仅限于标准动态范围（SDR），因而难以处理 HDR 的光度特性与感知细节。我们提出 HDR-VLM，这是首个将 VLM 适配到 HDR 域以进行感知质量评估的方法。具体而言，HDR-VLM 采用两阶段设计：首先通过统一的基于 HLG 的编码与渐进式适配来弥合域间差距；随后借助课程启发的奖励，通过强化学习使模型评估与带噪的多尺度人类偏好对齐。在来自真实生产环境的 HDR 数据集上的实验表明，HDR-VLM 不仅优于现有质量评估方法，还能给出可解释的归因理由。这些理由为内容创作者提供了可操作的指导，提升了自动化 HDR 质量评估的可靠性与透明度。
英文原文摘要 1289 字符
Color grading is central to High Dynamic Range (HDR) video production, shaping the perceptual tone, contrast, and luminance of content across diverse displays. However, evaluating HDR color grading quality is particularly difficult due to its semantic, content-dependent nature and the lack of large-scale annotated data. While pre-trained Vision-Language Models (VLMs) offer strong semantic priors and generalization ability, their exposure is limited to Standard Dynamic Range (SDR) data, making them poorly equipped to handle HDR photometry and perceptual nuances. We propose HDR-VLM, the first method to adapt a VLM to the HDR domain for perceptual quality assessment. Specifically, HDR-VLM employs a two-stage design: it first bridges the domain gap using a unified HLG-based encoding and progressive adaptation; then it aligns model assessments with noisy, multi-scale human preferences via reinforcement learning with curriculum-inspired rewards. Experiments on a real-world, production-sourced HDR dataset show that HDR-VLM not only outperforms existing quality assessment methods but also produces interpretable attribution rationales. These rationales offer actionable guidance for content creators, enhancing the reliability and transparency of automated HDR quality evaluation.