More Landmarks

I've got a script working fairly well. It uses the buffalo_l model. It takes a reference image, and compares it to a folder of images, aligning them and then giving a similarity score for each. However, given that it is only using 5 landmarks, I believe it could be more accurate. I imagine it's just taking these few distances into account, but not any of the other information that it could glean from comparing more landmarks.

As a result I looked into the insightface/alignment/coordinate_reg/[image_infer.py example](https://github.com/deepinsight/insightface/blob/master/alignment/coordinate_reg/image_infer.py). It uses the landmark_2d_106 model. And after some trial and error, I got it working correctly and drawing 106 landmarks on my images.

But now I don't know how to utilize these landmarks. face_align.norm_crop() seems to only work with 5 landmarks. Ultimately I'm trying to align and score on similarity as accurately as possible taking as many facial features as possible into account. How can I do this as accurately as possible given the fact that I'm not very adept at coding and have struggled to even get this far.

Thank you!