Azure form-recogniser fails with umlaut and Danish Ø

Question

I've been using Azure form recogniser for about a month now, and I've tried with with a variety of forms and European languages. However recently I've run into what I believe is a bug. That's when I use their UI through the dockerized label tool mcr.microsoft.com/azure-cognitive-services/custom-form/labeltool which exposes a web app for labelling and training.

I am working on a German document, and for some reason the UI will ignore umlauts and danish O, even before training and most definitely after. Is this something anyone has run into before, is there a solution or a compromise to be made? I know this is a long shot, but I thought it's worth asking.

Answer 1

The current version of service doesn't fully support those European languages yet.