飞机遥感影像分类数据集

飞机在遥感影像中是一种典型的目标。作者基于公开数据集DIOR、UCAS_AOD、NWPU VHR-10、DOTA以及Google earth中的高分辨率遥感影像，确定了全球多处机场，筛选有效飞机遥感数据3594张。然后，结合人体视觉注意力机制，对有效飞机数据进行划分，先根据飞机机翼和螺旋桨，粗略划分为7类；再根据飞机的颜色特征、发动机位置等，进行二次精细划分为14类。整理得到飞机遥感影像分类数据集。该数据集包括7大类飞机：（1）后掠翼飞机；（2）前缘后掠翼飞机；（3）后缘前掠翼飞机；（4）三角翼飞机；（5）平直翼飞机；（6）螺旋桨飞机；（7）直升飞机。该数据集存储为.png格式，由3594个数据文件组成，数据量为69.3MB（压缩为1个文件，69.3MB）。

Aircraft constitutes a typical type of target in remote sensing imagery. The authors collected high-resolution remote sensing images from public datasets including DIOR, UCAS_AOD, NWPU VHR-10, DOTA, and Google Earth, identified multiple airports worldwide, and screened out 3594 valid aircraft remote sensing images. Subsequently, leveraging the human visual attention mechanism, the valid dataset was partitioned: it was first roughly categorized into 7 classes based on aircraft wings and propellers, then further refined into 14 classes based on features such as aircraft color and engine position. Finally, a classification dataset of aircraft remote sensing images was compiled. This dataset includes 7 major aircraft categories: (1) Swept-wing aircraft; (2) Leading-edge swept-wing aircraft; (3) Trailing-edge forward-swept-wing aircraft; (4) Delta-wing aircraft; (5) Straight-wing aircraft; (6) Propeller-driven aircraft; (7) Helicopter. The dataset is stored in .png format, consisting of 3594 individual data files with a total size of 69.3 MB (compressed into a single archive file of 69.3 MB).

搜集汇总

背景与挑战

背景概述

该数据集是从多个公开遥感数据源精选的3594张飞机影像，采用两级分类体系（先按机翼特征分7大类，再细分为14小类），专门用于飞机目标识别与分类研究。