In what scenarios is it more suitable to use 2D image to generate point clouds?

When obtaining a 2D mask of an object through deep learning, it is common to use the ‘convert 2D points to 3D points’ step to generate a point cloud and then calculate the corresponding pose using ‘Calc Poses and Dimensions from Planar Point Clouds.’ In what scenarios is this functionality most suitable?

This Step is usually used to match 2D shapes after converting 2D edges to 3D point clouds for matching.
2D cannot calculate dimensions, so 3D is used for dimension calculations.

I hope this answer is helpful, but if it’s not, please feel free to let me know.