Large Scale Inference & Cloud Scaling Session about sat-ml-training HOT 3 CLOSED

The Sentinel 2 for Africa that @vincentsarago mentioned is Requester Pays, so wasn't easy to work into an example.

I don't think it's in requester-pays (maybe it changed recently)

from sat-ml-training.

@vincentsarago provided a notebook on using STAC and COGs, put that in as a draft to start the material.

• You can see it online, will be added to homepage when it's ready https://devseed.com/sat-ml-training/Scaling
• Code
• TODO: testing in Colab - possibly simplify if it's too complicated.

@drewbo Add what you want to teach on Inference scaling.

from sat-ml-training.

Added https://developmentseed.org/sat-ml-training/Earth-Engine-Inference

The workflow itself is functional but it contains one bug: the images generated from GEE are indices scaled from -1 to 1 and the training data is per-image, per-band scaled indices from 0-1. That latter normalization needs to be added to the doPrediction function for this to work properly. cc: @lillythomas

from sat-ml-training.