Repo size needlessly large.

[vybhav72954]

I understand that we need the pre-trained Model for detection and mapping of facial coordinates, but a 100 mb file for those who don't want to actually alter or play around with the code doesn't need to download the file while cloning the Repo. A better approach would be to host the file in GDrive and paste the link here.

Update the README, so that contributors can download if needed, this will make Repo significantly smaller and more efficient.