mmorellini/specs

SPECS数据集是一个用于研究城市视觉感知与社会经济因素关系的数据集，包含多个城市街道景观的上下文信息和感知评分。数据集内容结构包括不同城市的CSV文件、标签文件和图像数据，具体目录结构如下：SPECS/abuja/包含该城市图像的上下文信息CSV文件；global-streetscapes/包含原始Global Streetscapes数据集的上下文数据；labels/包含最终成对比较的XLSX文件、感知模型推断的CSV文件以及计算的感知Q分数的CSV文件；svi/包含图像路径、元数据和视觉复杂度的CSV文件。该数据集由新加坡-ETH中心的Future Cities Lab Global与新加坡国立大学城市分析实验室(UAL)合作开发。

The SPECS dataset is designed for studying the relationship between urban visual perception and socioeconomic factors, containing contextual information and perception scores for street scenes in multiple cities. The dataset structure includes CSV files for different cities, label files, and image data, with the following directory layout: SPECS/abuja/ contains CSV files with contextual information for images of this city; global-streetscapes/ contains contextual data from the original Global Streetscapes dataset; labels/ includes XLSX files with final pairwise comparisons, CSV files with inferred perception scores using the perception model from Global Streetscapes, and CSV files with computed perception Q scores; svi/ contains CSV files for image paths, metadata, and visual complexity. This dataset was developed at the Future Cities Lab Global in the Singapore-ETH Centre in collaboration with the Urban Analytics Lab (UAL) at the National University of Singapore (NUS).