Awesome work! The Git-10M consists of a large amount of images from GoogleEarth. I'm interested in how the images from Google Earth are collected.
- randomly selecting regions worldwide to ensure broad sample distribution.
- manually selecting specific areas to ensure comprehensive coverage of typical geographic features such as urban areas, forests, mountains, and deserts.
Could you please explain the proportion of randomly selected images and the manual selecting strategy? Thank you!
Awesome work! The Git-10M consists of a large amount of images from GoogleEarth. I'm interested in how the images from Google Earth are collected.
Could you please explain the proportion of randomly selected images and the manual selecting strategy? Thank you!