In this project, I have experimented the effect of Multi-task Learning in Bilingual Image Captioning.
I used COCO dataset and translated captions to Persian using Google Translate. It not a relient translation for sure, specially when we're facing captions which are slightly different than normal English text in most cases. But all in all, it for sure is a great starting point since we lack such datasets in Persian. \
Ledengary/bilingual-image-captioning
Bilingual Image Captioning (Persian - English) with Multi-task Learning
Jupyter Notebook