Extracting embeeded text out of images

[EmPi44]

Hi Team!
Hope things are going great! Just trying out llamaparse and having following challenge:
I am extracting information from a property floorplan. In some document the name and dimension of the rooms is text in the PDF documents which works fine.
For some documents the information is embeeded in the image (see attachment). I tried to use "withInstructionParsing" by also using the premium mode as well as vision capabilities of vendor for multimodular processing...but it is actually not extracting the values...
Any hints how I can solve the challenge?