metadata
license: bsd-3-clause
task_categories:
- feature-extraction
tags:
- street view image
- satellite image
- cross-view geo-localization
- visual place recognition
size_categories:
- 100K<n<1M
github: https://github.com/GaoShuang98/CVCities
paper: https://doi.org/10.1109/JSTARS.2024.3502160
We collected 223,736 ground images and 223,736 satellite images with high-precision GPS coordinates of 16 typical cities in five continents.
Notice:
- In order to reduce the file size and improve the download efficiency of this dataset, we have reduced the resolution of the ground level street view image in this dataset from 4096x2048pix of the original image to 1024x512pix.
- Since most of the CVGL models nowadays resize the street view images when they are trained and used, this will not affect the CVGL research.
- If you need the original size of the ground level street view image from the CVCities dataset, please contact me at huanggaoshuang123#at#163.com. Thanks for understanding☺️