@sheikhakbar2067

Interesting observation; you talk about data for vision what about textual data; NLP ....For example I am interested in Arabic, it is very widely used language (1400 years old language) there are lots of data for it (but unlabeled data! except for may be the Quran (religious text), where we have POS tags etc)... I've been thinking about this problem for quite some long, even considered labeling some data by myself.. but I want to do the minimum amount of data, and I want to do it efficiently!