harshinde/spacenet-rio

该数据集包含高分辨率卫星图像和相应的建筑足迹注释，用于训练深度学习模型进行语义分割和建筑足迹提取。数据集总共有6,940个图像块，包括3波段（RGB）全色锐化GeoTIFFs（高空间分辨率）和8波段多光谱GeoTIFFs。注释以GeoJSON格式（矢量多边形）提供，可轻松转换为二进制栅格掩码（0：背景，1：建筑）。数据集适用于语义分割（二进制或实例分割）、建筑足迹检测以及测试新的地球观测模型架构（如U-Nets）。数据来源于SpaceNet Challenge系列，旨在加速开发用于从商业卫星图像自动制图的开源算法。

This dataset contains high-resolution satellite imagery and corresponding building footprint annotations for (Rio de Janeiro) from the SpaceNet Building Detection Challenge. It is designed for training deep learning models for semantic segmentation and building footprint extraction. The dataset includes 6,940 image tiles, with 3-band (RGB) Pan-sharpened GeoTIFFs (high spatial resolution) and 8-band Multispectral GeoTIFFs. Annotations are provided in GeoJSON format (vector polygons), easily convertible to binary raster masks (0: background, 1: building). The dataset is directly applicable for semantic segmentation (binary or instance segmentation), building footprint detection, and testing new earth-observation model architectures (like U-Nets). The data originates from the SpaceNet Challenge series, which aims to accelerate the development of open source algorithms for automating mapping from commercial satellite imagery.