I recall the 'trick' from the vision training where we had to identify labels, but needed to teach just the label first (with no box).

I have a similar issue with a relatively tall object, and I can use this trick to teach just the top of the object to create the model.  But then when I run the program the scale of the object becomes an issue.  How does the camera software handle scale?  Is it just a percentage match like the model?

@JeanPhilippe_Jobin can you jump in?

@Grady_Turner
yes it is a percentage of the model and this percentage is quite high. So you don't have a lot of margin on the difference of height. What I suggest is that you place you planar model (label or whatever) at the height of the final part for the photo steps. Then use the final object for the scan step.