Sample notebook for landcover classification

jdries commented 7 months ago

A project is requesting an update of this usecase: https://docs.openeo.cloud/usecases/landcover/

As this is more of a production project, I suggest to use GFMap for it. Also need some specific updates:

use best available pixel for compositing
feature list to be provided by project
eventually needs to run on cdse

I'm also considering a design where we build a database of extractions, so we can run multiple trainings runs. (The original usecase did extraction and training in one job, which is nice but does not support a large amount of extractions.)

VictorVerhaert commented 6 months ago

Extraction and inference notebooks have been made.

TODO

[ ] add DEM as a feature, waiting for feedback from gisat on how to do this.
[ ] (optional) add BAP
[ ] optimize PG for the extraction (WIP)

HansVRP commented 6 months ago

For now simple masking is enough.

Leave it like this. Spent 1 or 2 days on optimization.

Follow up ticket on including BAP

VictorVerhaert commented 5 months ago

Closing this issue. I will create new issues when they arise

Open-EO / openeo-gfmap

Sample notebook for landcover classification #46