Reputation: 161
I have been using the form recognizer service and form labeller tool, using the version 2 of the api, to train my models to read a set of forms. But i have the need to use more than one layout of the forms, not knowing which form (pdf) layout is being uploaded.
Is it as simple as labelling the different layouts within the same model. Or is there another way to identify which model is to be used with which form.?
any help greatly appreciated
Upvotes: 1
Views: 644
Reputation: 312
This is a common request. for now, if the 2 forms styles are not that different, you could try to train one model and see if that model could correctly extract key/value. Another option is to train two different forms, you could write a simple classification program to decide which model to use.
Form Recognizer team is working on a feature to allow user just submit the document and it would pick the most appropriate model to analyze the document. Please stay tuned for our update.
thanks
Upvotes: 1