Contains Rufa dataset and a cleaned version of the CNN Arabic wordlist.
RuFa contents:
/rufa (40,516 images)
/real (516 images)
/ruqaa (260 images)
/farsi (256 images)
/synth (40,000 images)
/ruqaa (20,000 images)
/farsi (20,000 images)
More details in RuFa-metadata.md.