SAR建筑样本数据集（2019年）

本数据集是面向深度学习建筑区提取的中高分辨率SAR建筑区数据集SARBuD1.0 (SAR Building Dataset)。该数据集采集于SAR图像，结合光学图像与专家解译，制作了与样本数据对应的标签图像。SARBuD1.0数据集包含了不同地形场景类型、不同分布类型、不同区域的建筑区。该数据集可支持研究者对建筑区进行图像特征分析、辅助图像理解，并可对当前热点深度学习方法提供训练、测试数据支持。研究者可以基于SARBuD1.0数据集开展建筑区图像特征分析与语义分割提取等方面的研究。

This is a medium-to-high resolution SAR building dataset designated as SARBuD1.0 (SAR Building Dataset) for deep learning-based built-up area extraction. It is acquired from SAR images, and matching label images corresponding to the sample data are created by integrating optical images and expert interpretation. The SARBuD1.0 dataset encompasses built-up areas across diverse terrain scene types, distribution patterns and regional settings. This dataset enables researchers to conduct image feature analysis and auxiliary image understanding for built-up areas, and provides training and test data support for state-of-the-art deep learning methods. Researchers can carry out research on tasks such as built-up area image feature analysis and semantic segmentation extraction based on the SARBuD1.0 dataset.

搜集汇总

背景与挑战

背景概述

SAR建筑样本数据集（2019）是一个面向深度学习的中高分辨率合成孔径雷达（SAR）建筑数据集，包含SAR图像斑块和手动标注的标签图像，覆盖中国不同地形和区域的建筑区域。该数据集主要用于支持建筑SAR图像特征分析，并为深度学习的语义分割和提取研究提供训练和测试数据，数据分辨率为10米，来源于高分三号SAR数据。