Can you train the Dlib facial recognition with more than 1 photo per person?

Is it possible to use more than one photo per person to train the Dlib Facial Recognition, to improve the accuracy?

The example in the component docs (https://www.home-assistant.io/components/image_processing.dlib_face_identify/) only show one photo per person, but this seems a little limited?

I’m wondering the same. I doubt the accuracy is there with just a single photo.