中国90座城市建筑物屋顶矢量数据集

该数据集包含中国90座城市（根据城市行政等级及区域分布综合选取，城市名录详见附件1）建筑物屋顶矢量数据。主要基于深度学习语义分割模型和多源遥感影像进行制作。首先，对原始影像进行预处理，并根据城市等级及其区域分布情况进行分层采样以及目视解译，制作训练和测试数据。然后将训练数据输入深度学习语义分割模型进行训练，使其适用于建筑物屋顶提取任务，并基于测试数据，采用深度学习领域结果评价一般性指标对建筑物屋顶提取模型性能进行评价。最后，将此模型应用于中国90座城市建筑屋顶提取任务中，自动提取建筑物屋顶并进行矢量化。该数据集可以为城市乃至全国尺度以建筑物屋顶为基础的相关研究（如屋顶太阳能潜力评估、城市规划等）提供重要数据支撑。  CnOpenData获得了南师大智慧城市感知与模拟实验室的许可，将该数据收录于公共数据专区，以便各位学者浏览阅读，下方可下载样本数据，完整数据下载请转跳国家青藏高原科学数据中心（http://data.tpdc.ac.cn/zh-hans/news/c1b0ee71-2ceb-4276-8f5c-b53d57caa88d）。

This dataset contains vector data of building rooftops from 90 cities in China, which were selected comprehensively based on their administrative levels and regional distribution (the full list of cities is provided in Appendix 1). It was mainly developed using deep learning semantic segmentation models and multi-source remote sensing imagery. First, preprocessing was performed on the original remote sensing imagery, and stratified sampling and visual interpretation were conducted based on the administrative levels and regional distribution of the cities to create training and test datasets. Next, the training dataset was fed into the deep learning semantic segmentation model for training, adapting the model to the building rooftop extraction task. Subsequently, the performance of the trained model was evaluated using general evaluation metrics in the deep learning field based on the test dataset. Finally, this model was applied to the building rooftop extraction task across the 90 Chinese cities, enabling automatic extraction and vectorization of building rooftops. This dataset can provide critical data support for relevant studies based on building rooftops at the urban or even national scale, such as rooftop solar potential assessment, urban planning, and other related research. CnOpenData has obtained permission from the Smart City Perception and Simulation Laboratory of Nanjing Normal University to include this dataset in the public data section for scholars to browse and access. Sample data can be downloaded below; for full data download, please visit the National Tibetan Plateau Data Center at http://data.tpdc.ac.cn/zh-hans/news/c1b0ee71-2ceb-4276-8f5c-b53d57caa88d.

搜集汇总

背景与挑战

背景概述

该数据集包含中国90座城市建筑物屋顶矢量数据，基于深度学习语义分割模型和多源遥感影像制作，覆盖范围广，数据精度高，适用于屋顶太阳能潜力评估和城市规划等研究。数据格式为Esri Shapefile，包含屋顶面积和几何中心经纬度等关键字段。