Can you please clarify whether facial features are added for training?

[Dharbaneshwer]

In the paper, it has been mentioned that the facial features are added for training the model, but the architecture mentioned in the repo code is not asserting the claim in the paper..

Please clarify ASAP.