UDP is a nice way of encapsulating / encoding ML feature engineering recipes, to use during model training in EOTDL and then for inference in a different environment.
EDC and EOxHub, SentinelHub as well as CDSE support openEO API and process graphs (yet not all openEO processes…).
We should evaluate if we can use this to store, exchange and reuse feature engineering workflows in EOTDL.
earthpulse
commented
5 months ago
Feature stores
Here is the link to the UDP example: https://github.com/Open-EO/openeo-community-examples/blob/main/python/Sentinel1_Stats/Sentinel1_Stats.ipynb
UDP is a nice way of encapsulating / encoding ML feature engineering recipes, to use during model training in EOTDL and then for inference in a different environment.
EDC and EOxHub, SentinelHub as well as CDSE support openEO API and process graphs (yet not all openEO processes…).
We should evaluate if we can use this to store, exchange and reuse feature engineering workflows in EOTDL.