Training FaceRecognizer model with improved images

edit

I want to modify plain image samples by adding things like glasses, mustache and other facial and external features( using augmented reality) and train it on a FaceRecognizer model. Then I want to see if that has any effect on the recognition rate of the model when confronted with samples with actual facial features. So I haven't done much on augmented reality and I want some help getting started on that. Thank You.