SAR-Airport_Dataset

The spaceborne SAR airport detection dataset mainly uses data from the European Space Agency's Sentinel-1B satellite to construct a SAR airport target dataset that includes multiple polarization modes, sizes, and covers multiple countries and cities, promoting in-depth research on advanced technologies such as SAR airport target detection. The dataset was constructed by a team from the Remote Sensing Technology Research Institute of Beijing University of Chemical Technology, including Zhang Fan, Ma Fei, Zhou Yongsheng, and others.This dataset uses the first order interferometric broadband GRD product of the European Space Agency Sentinel-1B satellite, with a spatial resolution of 10 × 10 meters, with an incidence angle of about 20 °~45 °, and polarization modes including VH and VV. The data contains a total of 624 SAR airport detection images, covering 104 airport targets and gathering multiple cities from multiple countries in Asia, Europe, Africa, North America, and South America.All image sizes are 2048 × 2048 pixels in single channel grayscale image format, 8-bit deep JPG. The annotation file is in TXT format, with one target labeled per line. It records the detection type, normalized airport center position (column, row number), normalized airport width, and normalized airport length, which comply with the format requirements of mainstream detection networks such as Yolo series, PolarMask, SSD, and Faster RCNN.

本星载合成孔径雷达（Synthetic Aperture Radar，SAR）机场检测数据集主要采用欧洲空间局（European Space Agency，ESA）Sentinel-1B卫星的相关数据，构建了涵盖多极化模式、多尺寸规格，且覆盖多个国家与城市的SAR机场目标数据集，以推动SAR机场目标检测等先进技术的深入研究。本数据集由北京化工大学遥感技术研究所团队构建，成员包括张帆、马飞、周永胜等人。
该数据集采用欧洲空间局Sentinel-1B卫星的一级干涉宽带地面距离向检测（Ground Range Detected，GRD）产品，空间分辨率为10×10米，入射角约20°~45°，极化模式包含VH与VV极化。本数据集共包含624幅SAR机场检测图像，涵盖104个机场目标，覆盖亚洲、欧洲、非洲、北美洲及南美洲多个国家的多座城市。
所有图像均为单通道灰度图，尺寸为2048×2048像素，采用8位深度的JPEG格式。标注文件采用TXT格式，每一行标注一个目标，记录了检测类别、归一化后的机场中心位置（列号、行号）、归一化后的机场宽度与归一化后的机场长度，符合Yolo系列、PolarMask、SSD及Faster RCNN等主流检测网络的格式要求。

搜集汇总

背景与挑战

背景概述

SAR-Airport_Dataset是一个基于Sentinel-1B卫星数据构建的合成孔径雷达机场检测数据集，包含624张图像，覆盖全球多个大洲的104个机场目标，支持VH和VV极化模式。数据集提供2048×2048像素的灰度图像和TXT格式标注，适用于Yolo、Faster RCNN等主流检测网络，旨在推动SAR机场目标检测技术的研究。