When creating dimensions of a sketch or model, you need to be able to understand the geometry. Then pick the graphical points (geometry intent) of that geometry and feed it to a create dimension command (of various sorts).
I've used sketch lines with hidden attributes added by my own code in advance, or user work points (added by code or user), or label surfaces (added by code or user). You could also simply query the shown geometry and create dimensions from anything that fits a logic pattern (longest line that is horizontal, or maximum and minimum points, or whatever). Inventor 2018 and up can label edges which makes an attribute you can look for.
The logic pattern can be very complex if you want to dimension every possible scenario of geometry you can run across (holes, curves, chamfers, lines, arcs, splines, etc.) so most of us shy away from the logic pattern method, and focus on pre-labeling the objects of interest.
Another 'pre-label' method could be actual dimensions in a sketch created on the flat pattern using projected geometry of the flattened shape. These dimension objects will give you all the data you need, and may even be able to retrieve them directly into the drawing space.
jvj
