Відео

I have never attempted selection mark within columns. You can click the three vertical dots next to the field and define the sub data type for the field. Tables are tricky and very strict. I would need to see the sample documents and what youre trying to achieve to better direct you

@BrentWesler it's a very narrow table layout. table basically stretches over the whole length and width of the page. 20 or so columns, the rightmost 10 are columns meant to be selectionMark columns. maybe the squeezed layout is a reason the model keeps failing. I had a custom template model trained and in production so far on that same document layout - only without the selection marks. but now the customer wants all the information, also selection mark columns. I tried a minimal working solution with just a table field and one selectionMark column. Trained successfully. Then I added another one right next to it. Trained successfully. But then I completed some unlabeled fields in the second column with that "draw field" method and whoops. training failed again. Document intelligence gives great resukts when it works. When training fails you hardly know why

Kodak Info Input

you can buy it from us at piftech.com, its not an off the shelf product you need to purchase through a dealer like piftech.com

No there is no human in the loop but thats why we use a IDP product like Kodak Info Input which provides a UI to interrogate the data

Classification and separation happens mutually exclusive from extraction. To perform both you need to call both a prebuilt or custom extraction model plus a classification model.

yes, azure has a new Add-on/premuim option called formulas which will extract formulas as extracted. This is available in the general Documents prebuilt parser in azure studio and using the "Formulas"

I had to use a programming language to do it but it does provide a sample code to go off by.

You can then take the document and export it to csv

@@KEVINSURIEL Hi Kevin, I am working on a similar problem to export the extracted data to csv format. Do you mind sharing the steps you performed to export it onto csv?

@@Shivank-op6yy do you have python coding skills by any chance if not download chatgpt and it ask you to make a python script that can pull data from document ai intelligence and export it to csv

@@Shivank-op6yy do you have any experience in python if not you can use chatgpt to help you create a python script.

Kodak Alaris Info Input at alarisworld.com or you can purchase through me at piftech.com

you need to use the same blob storage account and same container name. Happy to consult with you. bwesler@piftech.com

yes I can help. you need to use the Key-Value pair checkbox on the invoice prebuilt model and then find the serial number using its label or key. Happy to consult with you. bwesler@piftech.com

ImageTrust by Image Access or rebranded by Kodak Alaris Info Input. You can buy through piftech.com

You would have to use Power Automate and AI Builder which can monitor an email and call the data model of choice. This way you are not consuming Azure AI Document Intelligence models but rather AI Builde rusing AI credits through your M365 tenant. Its more expensive to use AI builder than the Azure version. You can use a 3rd party IDP software like Kodak Alaris Info Input to read eml or msg emails and rasterize them to PDF

We would have to look at th document and use case. The Azure Invoice prebuilt parser is the only model that will paginate through pages of following line items. We then get the JSON payload back and can use Javascript to find column labels that have repeating controls using Kodak Alaris Info Input

Yeah, that's my question too. Can it tell which training model that file belongs to? So that I do not need to specifically say which training model it should be used to extract the content of the document.

@@rolandyaulingsiang7911 Exactly I am trying to find out answer. Do you have figured it out?

Hi Keith, I am stuck with the same problem. When I create my custom Table with the columns which I need. The Table does not sort the values. I am not sure how will it work.