目标检测-无人机、航拍数据集总结
今天导师让我整理一下无人机类，航拍类的数据集，大概搜索了一下，找到以下几种数据集，在这边总结分享，方便后人
1、DOTA Dataset（A Large-scale Dataset for Object DeTection in Aerial Images）是用于航拍图像中目标检测的图像数据集，其被用于发现和评估航拍图像中的物体，DOTA – V1.0 包含来自不同传感器和平台共计 2806 幅航拍图，每张图像的像素尺寸在 800800 到 40004000 的范围内，其中包含不同尺度、方向和形状的物体。之后这些 DOTA 图像经由专家使用 15 个常见目标类别进行注释，完全注释的 DOTA 图像包含 188,282 个实例，每个实例均由任意四边形进行标记。
DOTA 数据集由武汉大学于 2017 年 11 月 28 日发布在 arXiv 上，后于 2018 年 6 月在 IEEE 计算机视觉和模式识别会议（CVPR）上发布，DOTA – V1.0 的图像来自于 Google Earth，其中一些由 JL – 1 卫星拍摄，其余由中国资源卫星数据和应用中心的 GF – 2 卫星拍摄。
下载链接magnet:?xt=urn:btih:AD823417DC70658B89CD803584F6B6809EA9D3CB
2、航空遥感数据集、无人机航拍数据集
类别包括有：农田、飞机、海滩、房屋、森林、高速公路、港口、路口、立交桥、停车场、河、跑道、储油罐、网球场等
尺寸256*256
网址：http://weegee.vision.ucmerced.edu/datasets/landuse.html
迅雷下载链接：http://weegee.vision.ucmerced.edu/datasets/UCMerced_LandUse.zip
3、UCAS-AOD: Dataset of Object Detection in Aerial Images，中国科学院大学模式识别与智能系统开发实验室标注的，只包含两类目标：汽车，飞机，以及背景负样本。样本数量如下：
链接：https://pan.baidu.com/s/11Z4fFO0MsFArhRmZhxtBlQ
提取码：u58c
4、NWPU VHR-10：西北工业大学标注的航天遥感目标检测数据集，共有800张图像，其中包含目标的650张，背景图像150张，目标包括：飞机、舰船、油罐、棒球场、网球场、篮球场、田径场、港口、桥梁、车辆10个类别。开放下载，大概73M.
链接：https://pan.baidu.com/s/1VxooSn4TL3TldCgNx6YK4Q
提取码：d4bs
5、RSOD-Dataset
It is an open dataset for object detection in remote sensing images. The dataset includes aircraft, oiltank, playground and overpass.
The format of this dataset is PASCAL VOC.
The dataset includes 4 files, and each file is for one kind of object. Please download the dataset files from BaiduYun.
aircraft dataset, 4993 aircrafts in 446 images.
playground, 191 playgrounds in 189 images.
overpass, 180 overpasses in 176 images.
oiltank, 1586 oiltanks in 165 images.
Github 下载地址：https://github.com/RSIA-LIESMARS-WHU/RSOD-Dataset-
6、
The Inria Aerial Image Labeling addresses a core topic in remote sensing: the automatic pixelwise labeling of aerial imagery (link to paper).
Dataset features:
Coverage of 810 km² (405 km² for training and 405 km² for testing)
Aerial orthorectified color imagery with a spatial resolution of 0.3 m
Ground truth data for two semantic classes: building and not building (publicly disclosed only for the training subset)
The images cover dissimilar urban settlements, ranging from densely populated areas (e.g., San Francisco’s financial district) to alpine towns (e.g,. Lienz in Austrian Tyrol).
Instead of splitting adjacent portions of the same images into the training and test subsets, different cities are included in each of the subsets. For example, images over Chicago are included in the training set (and not on the test set) and images over San Francisco are included on the test set (and not on the training set). The ultimate goal of this dataset is to assess the generalization power of the techniques: while Chicago imagery may be used for training, the system should label aerial images over other regions, with varying illumination conditions, urban landscape and time of the year.
The dataset was constructed by combining public domain imagery and public domain official building footprints.
Linux
to download the dataset you can use this command
curl -k https://files.inria.fr/aerialimagelabeling/getAerial.sh | bash
or you can download all files manually and use 7z and unzip to extract files
https://files.inria.fr/aerialimagelabeling/aerialimagelabeling.7z.001
https://files.inria.fr/aerialimagelabeling/aerialimagelabeling.7z.002
https://files.inria.fr/aerialimagelabeling/aerialimagelabeling.7z.003
https://files.inria.fr/aerialimagelabeling/aerialimagelabeling.7z.004
https://files.inria.fr/aerialimagelabeling/aerialimagelabeling.7z.005
