A Multimodal Visible-SAR Dataset for Airport Detection in Remote Sensing Imagery

This dataset is based on Google Earth visible light images and Gaofen-3 (GF-3) synthetic aperture radar (SAR) images, and constructs a visible light SAR remote sensing multimodal target detection dataset focusing on airport targets nationwide. The dataset consists of two parts: image data and annotation data. The image data is divided into visible light modal data and SAR modal data, and the annotation file data includes VOC rotation box annotation file format and YOLO horizontal box annotation file format. In the process of constructing the dataset, follow standardized methods for data collection, processing, and annotation. The dataset has the characteristics of pixel level alignment, large spatial span, large scale span, rich scene diversity, and small targets, which can meet the training needs of mainstream deep learning models. This dataset is the first publicly released visible light SAR spatially aligned remote sensing multimodal object detection dataset, providing important basic data resources for researchers in related fields and having practical application value for promoting research in the field of remote sensing multimodal object detection.

本数据集基于谷歌地球（Google Earth）可见光影像与高分三号（Gaofen-3, GF-3）合成孔径雷达（Synthetic Aperture Radar, SAR）影像，构建了面向全国范围机场目标的可见光-SAR遥感多模态目标检测数据集。数据集包含影像数据与标注数据两大组成部分：其中影像数据划分为可见光模态数据与SAR模态数据；标注文件数据涵盖VOC旋转框标注格式与YOLO水平框标注格式。在数据集构建过程中，严格遵循标准化的数据采集、处理与标注流程。该数据集具备像素级配准、空间跨度大、尺度跨度广、场景多样性丰富且包含小目标等特点，可满足主流深度学习模型的训练需求。本数据集为首个公开发布的像素配准型可见光-SAR遥感多模态目标检测数据集，可为相关领域科研人员提供重要的基础数据资源，对推动遥感多模态目标检测领域的研究具有实际应用价值。

搜集汇总

背景与挑战

背景概述

该数据集是一个多模态可见光-SAR遥感机场检测数据集，包含像素级对齐的可见光和SAR图像数据，以及VOC旋转框和YOLO水平框两种标注格式。数据集具有大空间跨度、多尺度特性和丰富场景多样性，是首个公开的可见光-SAR空间对齐遥感多模态目标检测数据集，适用于深度学习模型训练。