search

nealjean / predicting-poverty

Combining satellite imagery and machine learning to predict poverty
http://sustain.stanford.edu/predicting-poverty
MIT License
453 stars 232 forks source link

the 2nd step #8

Closed mtz1402 closed 7 years ago

mtz1402 commented 7 years ago

Hey so for the 2nd step training step, night lights are used. Is that in image form or values ranging from 0 to 62? Also the second training step predicts the nightlight intensities from the daytime images. So does that mean you also derive a new data set of predicted light intensity values from your training along with the third row of images from figure 2?

nealjean commented 7 years ago

We bin the nightlights into 3 classes, so the labels should be "low", "medium", and "high".

You could produce predicted nightlight intensities from daytime imagery, but we only use the nightlights as a way to learn satellite image features.