preprocessdata.bash

#!/bin/bash

#download the data
#/snap/bin/gsutil -m cp gs://quickdraw_dataset/full/simplified/*.ndjson .
#for each data set, get n (500?) random samples, that were identified
#grep -c true cat.ndjson | sort -R | head -n 500 > cat_500.ndjson
#cut -d : -f 9 cat_500.ndjson  > cat_500.txt