search

cc-ai / kdb

AI can help Visualizing the Impacts of Climate Change. This is an open forum to share our work
https://mila.quebec/en/ai-society/visualizing-climate-change/
Apache License 2.0
19 stars 3 forks source link

Reducing randomness among pairs - Synthetic Data #107

Closed gcosne closed 4 years ago

gcosne commented 5 years ago

Difference among pair of non-flooded and flooded images We would like to reduce the randomness for the vegetation and everywhere but the ground. image (1) image image (2)

Vahe987 commented 5 years ago

I can probably stop all movement by just putting timescale=0 (will test) about the rest I don't really know what you call noise, how you get this images? is this on all of the paired images, or just couple of them? thanks

gcosne commented 5 years ago

That would be a good thing to stop the movements: timescale = constant These images are produced by computing the difference between a non-flooded and flooded image of the same scene. This randomness has been observed on several of them (one is already enough to make the optimization harder..)

Vahe987 commented 5 years ago

Ok, on the next set of images I'll consider your comments, thank you however, one thing to try first, is to take the same image and calculated the difference, because I'm surprised that buildings has difference as well ... timescale = 0 will stop the movement, but buildings are not moving anyways ...