MLS-Dataset

MLS-Dataset是由俄罗斯科学院信息传输问题研究所创建的一个实验室数据集，主要用于评估计算色彩恒常性算法。该数据集包含24个场景，每个场景记录了18种不同光照条件下的图像，这些光照条件在主光谱颜色、强度和距离上有所变化。数据集不仅提供了图像，还包括相机的光谱特性、光源和物体的光谱特性，以及像素级的均匀着色物体表面真实标注。创建过程中，精确测量了场景中的所有特性，包括光谱和物体位置。该数据集适用于计算机视觉中涉及色彩或基于材料的图像分割、光度不变量提取等多个领域，旨在解决多光源下的色彩估计问题。

The MLS-Dataset is a laboratory dataset created by the Institute of Information Transmission Problems of the Russian Academy of Sciences, primarily used for evaluating computational color constancy algorithms. This dataset consists of 24 scenes, each of which has images captured under 18 different lighting conditions that vary in dominant spectral color, intensity, and distance. In addition to the images, the dataset also provides the spectral characteristics of the camera, the spectral properties of the light sources and objects, as well as pixel-level ground truth annotations for uniformly colored object surfaces. During its creation, all characteristics of the scenes were precisely measured, including the spectral properties and object positions. This dataset is applicable to multiple fields in computer vision such as color or material-based image segmentation, photometric invariant extraction, and aims to address the problem of color estimation under multiple light sources.