It does not look like there are currently anybody using this service from mendix.
I might be wrong though.
Well I developed an OCR application as a project.
In Mendix case. You first have to train your model (note: The more you train, the more accurate your model becomes). Please make sure that all the images that you upload for training are of same type and on same template, in order for the OCR to work correctly. Please make sure that you have already configured the binding keys in your app.
after the model is trained. you need to publish the model and get the JSON Object so that you can import data in your app.
It works pretty fine.
you can follow this doc for the refference.
hope it helps!
I think also, that it does not look like there are currently anybody using this service from mendix.