SAVOIAS

SAVOIAS数据集由波士顿大学计算机科学系创建，包含1420张图像，分为七个类别，涵盖场景、广告、可视化、物体、室内设计、艺术和至上主义。数据集通过众包方式获取超过37,000对图像的成对比较，使用Bradley-Terry方法和矩阵完成转换为绝对视觉复杂性分数。该数据集旨在支持视觉复杂性分析，特别是在计算机视觉领域，帮助开发新的算法和方法，以理解和量化图像的视觉复杂性。

The SAVOIAS dataset was created by the Department of Computer Science at Boston University. It contains 1,420 images categorized into seven classes, covering scenes, advertisements, visualizations, objects, interior design, art, and Suprematism. The dataset collected over 37,000 pairwise image comparisons via crowdsourcing, and converted these comparisons into absolute visual complexity scores using the Bradley-Terry method and matrix completion. This dataset aims to support visual complexity analysis, particularly in the field of computer vision, to assist in developing novel algorithms and methods for understanding and quantifying the visual complexity of images.