question

amitlal-azureguy avatar image
0 Votes"
amitlal-azureguy asked amitlal-azureguy commented

Building Form Recognizer with labels and output required in pdf

Hello Members,

Started working Form Recognizer by building a training model for our custom pdfs sources.
And trying to extract the selective tables/images from those 100 paged PDF. To train this custom model, we uploaded 10 different PDF versions and now successfully receiving all outputs in JSON very well with a 90% + score on the training model.

Our baseline question => How to convert JSON output received from those PDFs to a similar PDF format?
Any inputs are highly appreciated.

Regards,
Amit Lal

azure-form-recognizerazure-ink-recognizer
· 1
5 |1600 characters needed characters left characters exceeded

Up to 10 attachments (including images) can be used with a maximum of 3.0 MiB each and 30.0 MiB total.

Please let me know if you have any question according to it. Thanks.

0 Votes 0 ·
YutongTie-MSFT avatar image
0 Votes"
YutongTie-MSFT answered

Hello Amit,

Thanks for reaching out to us. There is no way to output pdf from form recognizer, but you can use logic apps to do it (Form recognizer as a part). There is a sample solution for you please feel free to refer to it.

https://powerusers.microsoft.com/t5/Building-Flows/Extracting-PDF-data-with-Form-Recognizer-and-saving-it-to/td-p/429459

And the document for Logic apps
https://azure.microsoft.com/en-us/services/logic-apps/

Regards,
Yutong

5 |1600 characters needed characters left characters exceeded

Up to 10 attachments (including images) can be used with a maximum of 3.0 MiB each and 30.0 MiB total.

amitlal-azureguy avatar image
0 Votes"
amitlal-azureguy answered amitlal-azureguy commented

Hi Yutong,
Thanks for your input. I understand Logic apps required here.
The bigger question can form recognizer layout API able to fetch selective images and tables from pdf report? If yes, please share some insight/GitHub etc.

Thank you,
Amit Lal

· 4
5 |1600 characters needed characters left characters exceeded

Up to 10 attachments (including images) can be used with a maximum of 3.0 MiB each and 30.0 MiB total.

Hello Amit,

Thanks for the response. For tables, yes, please refer to the document: https://docs.microsoft.com/en-us/azure/cognitive-services/form-recognizer/concept-layout?

For image, I am checking internal to see if there any raodmap here.


Regards,
Yutong

0 Votes 0 ·

Hello Yutong, Thanks for your inputs. Any Github ref. for Tables fetching, that should help.
Perhaps, I'll wait for your inputs on the image fetching. Thank you,
Amit

0 Votes 0 ·

Hello,

What's kind of GitHub Reference you want? I think below two are enough. The first one is the introduce and the second one is the API reference.


https://docs.microsoft.com/en-us/azure/cognitive-services/form-recognizer/concept-layout#tables

https://westcentralus.dev.cognitive.microsoft.com/docs/services/form-recognizer-api-v2-1-preview-3/operations/AnalyzeLayoutAsync

Regards,
Yutong

0 Votes 0 ·
Show more comments