Verifying a satellite image with temporal evidence
We ran a simulation to extract dates of undated satellite images using dated, low-resolution image time-series. The results are fairly accurate. The emem plugin is available in chatgpt, vscode, claude. Give it a try yourelf.
How does it handle cases where the image is 100%, or close to 100%, cloud cover? This occurs frequently with satellite imagery.
100% cloud cover is kind of useless until you are exclusively tracking cloud.
It may be useful to know the location of a satellite image, even if there was cloud cover when it was taken.
URL?
https://github.com/Vortx-AI/emem [ https://emem.dev ]
Also a short video on the flow - https://www.youtube.com/shorts/sxcG2q5q5bI