LIGHTS Specularity Dataset

LIGHTS Specularity Dataset是由意大利技术研究院的Visual Geometry and Modelling (VGM) Lab和Pattern Analysis and Computer Vision (PAVIS)创建的高质量建筑场景数据集，包含18个场景，总计2603个多视角渲染图像。该数据集通过物理基础渲染技术生成，旨在解决镜面高光检测的挑战，特别是在真实世界环境中难以控制的光照条件下。数据集不仅用于镜面高光分析，还可应用于阴影检测、法线估计和深度估计等领域。

LIGHTS Specularity Dataset is a high-quality architectural scene dataset developed by the Visual Geometry and Modelling (VGM) Lab and Pattern Analysis and Computer Vision (PAVIS) at the Italian Institute of Technology. It comprises 18 scenes with a total of 2603 multi-view rendered images. Generated via physically based rendering (PBR) techniques, this dataset is designed to address the challenges of specular highlight detection, especially under uncontrollable lighting conditions in real-world environments. Beyond specular highlight analysis, the dataset can also be applied to tasks including shadow detection, normal estimation and depth estimation.