AID (Aerial Image Dataset)

AID是一个新的大规模航空图像数据集，通过从谷歌地球图像中收集样本图像。请注意，尽管Google Earth图像是使用原始光学航空图像的RGB效果图进行后处理的，但事实证明，即使在像素级土地使用/覆盖映射中，Google Earth图像与真实光学航空图像之间也没有显着差异。因此，Google Earth图像也可以用作航空图像来评估场景分类算法。新数据集由以下30种航空场景类型组成: 机场，裸地，棒球场，海滩，桥梁，中心，教堂，商业，密集住宅，沙漠，农田，森林，工业，草甸，中等住宅，山地，公园，停车场，游乐场，池塘，港口，火车站，度假村，河流，学校，稀疏住宅，广场，体育场，储罐和高架桥。所有的图像都由遥感图像解释领域的专家标记，每个类别的一些样本如图1所示。总之，AID数据集在30个类中具有多个10000图像。AID is a new large-scale aerial image dataset constructed by collecting sample images from Google Earth imagery. Please note that although Google Earth images are post-processed RGB renderings of raw optical aerial images, studies have shown that there is no significant difference between Google Earth images and real optical aerial images, even in pixel-level land use/cover mapping. Therefore, Google Earth images can also be used as aerial imagery to evaluate scene classification algorithms. The new dataset comprises the following 30 aerial scene categories: airport, bare land, baseball field, beach, bridge, city center, church, commercial area, dense residential area, desert, farmland, forest, industrial area, meadow, medium-density residential area, mountain, park, parking lot, playground, pond, port, railway station, resort, river, school, sparse residential area, square, stadium, storage tank, and viaduct. All images are annotated by experts in the field of remote sensing image interpretation, and some samples of each category are shown in Figure 1. In summary, the AID dataset contains more than 10,000 images across the 30 categories.

搜集汇总

背景与挑战

背景概述

AID是一个大规模航空图像数据集，包含超过10,000张图像，涵盖30种航空场景类型，如机场、森林和农田等，通过从谷歌地球图像收集并由专家标注。该数据集由武汉大学和华中科技大学于2017年发布，主要用于评估场景分类算法，适用于计算机视觉任务如语义分割和目标检测。