Advice for model to recognize a person and his/her expression

Hello :wave:t3:,

I need guidance for a project which I’m building as a way to learn DL.

You see, I would like to build a model that can both:

  • recognize faces
  • and recognize expressions/emotions

Input: photo --> Output: that’s John Doe, he’s happy

My question is:

Should I build 2 classifiers, 1 that identifies faces and 1 which recognizes emotions, then combine them (is transfer learning applicable?)
Should I build a multilabel classifier ?

What do you think is the best approach and why ? Thanks for your advice.