SpaceNet 数据集

SpaceNet 数据集

SpaceNet是DigitalGlobe商业卫星公司提供的遥感图像集合,包含一些标记信息可用作机器学习研究.

SpaceNet Challenge主页: https://spacenetchallenge.github.io/

数据集下载命令: https://github.com/SpaceNetChallenge/utilities/tree/master/content/download_instructions

关于SpaceNet数据集

Source	DigitalGlobe, Inc.
Category	Computer Vision, Geospatial
Format	GeoTIFF, GeoJSON
License	Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International
Storage Service	Amazon S3
Location	s3://spacenet-dataset in us-east-1
Update Frequency	New imagery and features are added quarterly

下载

安装aws-cli客户端

pip install awscli

如果安装时提示找不到yaml.h,安装libyaml-dev (apt-get install libyaml-dev)

配置aws账号

$# aws configure
AWS Access Key ID [None]: 你的账号
AWS Secret Access Key [None]: 你的账号
Default region name [None]: us-west-2
Default output format [None]: json

查看SpaceNet Data中的所有文件列表,访问该数据集需要添加--request-payer选项来访问收取流量费用的bucket存储:

$# aws s3 ls spacenet-dataset --request-payer requester
                           PRE /
                           PRE AOI_1_Rio/
                           PRE AOI_2_Vegas/
                           PRE AOI_3_Paris/
                           PRE AOI_3_Paris2/
                           PRE AOI_4_Shanghai/
                           PRE AOI_5_Khartoum/
                           PRE SpaceNet_Roads_Competition/
                           PRE competition1/
                           PRE mvs_dataset/
                           PRE test/
2016-08-19 02:52:48       2742 .
2016-09-02 03:10:03       2910 AOI_1_Rio_manifest.txt
2017-01-05 07:22:45   17953095 Rio_BuildingLabels.tar.gz
2016-11-16 11:03:40        338 competition_manifest.txt
2016-11-16 11:10:23       3404 manifest.txt
2017-02-16 02:24:49          0 manifest2.txt
2017-03-10 00:29:14  213227956 spacenet_sample.tar.gz

下载上海数据AOI 4 (23 GB)

aws s3api get-object --bucket spacenet-dataset --key AOI_4_Shanghai/AOI_4_Shanghai_Train.tar.gz --request-payer requester AOI_4_Shanghai_Train.tar.gz

---

彩蛋

镜像地址

数据处理

To be continued...

参考链接

• http://blog.csdn.net/dynastyrumble/article/details/76649120
• https://medium.com/the-downlinq/creating-training-datasets-for-the-spacenet-road-detection-and-routing-challenge-6f970d413e2f