As you know, to make machine learning work, it needs loads of data. Most picture and videos are taken from a camera at arm's length, not attached to someone's

As you know, to make machine learning work, it needs loads of data. Most picture and videos are taken from a camera at arm's length, not attached to someone's face. so if you want to make AR glasses "see" and "understand" the world from the point of view of a human (ie navigation, where is x, etc etc) then you need to make a dataset with that sensor configuration. https://www.youtube.com/watch?v=6vnZCwf5_QE https://www.youtube.com/watch?v=6vnZCwf5_QE has a simple example from CMU, rather than facebook