Conceptually, if you had an internal 3d model of all objects in CV you could do a scan matching algorithm.
This algorithm would be ridiculously computationally intensive, but it would have a high degree of accuracy (I believe).
I'm wondering if ANN can be used to approximate this internal scan matching in a way that greatly reduces computational load?
For example let's say you have a 3d scan of a car. This car can take a variety of orientations.
One way to do scan matching is pick a bunch of orientations, project to 2d and do a sliding scan match.
Could this entire operation be approximated by a NN which ultimately would have a reduced computational load?