CityLens

该数据集是一个全面性的基准测试，旨在评估大型语言视觉模型通过分析全球17个不同城市卫星和街景图像来预测社会经济指标的能力。该数据集包含11项预测任务，并使用R²和nRMSE等指标进行评估。其规模覆盖了6个关键领域：经济、教育、犯罪、交通、健康和环境，任务目标是利用视觉数据预测社会经济指标。

This dataset is a comprehensive benchmark designed to evaluate the capability of Large Vision-Language Models (LVLMs) in predicting socioeconomic indicators by analyzing satellite and street-view images from 17 distinct cities across the globe. It includes 11 prediction tasks, with model performance evaluated using metrics such as R² and normalized Root Mean Squared Error (nRMSE). The dataset spans six core domains: economy, education, crime, transportation, health, and environment, with the overarching objective of leveraging visual data to forecast socioeconomic indicators.