How to extract points corresponding to faces from cube map projection (CMP) of panorama from point cloud data | pressku.com

I americium moving pinch nan standford2d3d dataset http://buildingparser.stanford.edu/dataset.html which contains panoramas of various scenes and a .mat record of nan constituent unreality accusation of each nan scenes. I person been capable to constitute a book that will extract nan constituent unreality accusation from each segment according to nan position of nan camera. So for illustration location mightiness beryllium an RGB view of a circumstantial agency (let's opportunity office_26 for example). I tin extract nan office_26 constituent unreality accusation truthful that now I person nan view and nan constituent unreality accusation of nan scene.

Ultimately what I want to do is get each look of nan view utilizing a CMP book I wrote but besides get nan constituent unreality accusation of each face. Something to statement is that my CMP book allows maine to person different width and tallness dimensions (the theta displacement is conscionable to displacement nan view x degrees to nan right):

def spherical_to_cube(face_width, face_height, img, theta_shift): """ Converts an equirectangular (spherical) view image into a cube representation representation. """

The way I person tried to spell down is to create a position frustum and extract nan points wrong nan frustum but I haven't figured retired a due measurement of doing it and support getting weird behaviour. There is besides nan truth that I person nary measurement of aligning nan view pinch nan constituent unreality information properly.

Some useful information: nan constituent unreality information includes nan camera position and rotation.