Texts, tables, and selection marks aren't analysed past the 2nd page of a PDF

Question

Texts, tables, and selection marks aren't analysed past the 2nd page of a PDF

Cyril Carraz 41

I'm trying to train a custom model using Form Recognizer v3.0 but I can't label my data past the second page of a PDF because the text isn't analysed like in the first two pages.

How do I annotate and label the data on the third page?

romungi-MSFT 49,056 Reputation points Microsoft Employee Moderator

2023-02-13T12:40:30.29+00:00

Cyril Carraz Did you try to upgrade the pricing tier and check if it helped? Thanks!!
romungi-MSFT 49,056 Reputation points Microsoft Employee Moderator

2023-02-15T09:55:36.8733333+00:00

Cyril Carraz Did you get a chance to check if the below answer worked for your?

Answer accepted by question author

0 additional answers

Your answer

romungi-MSFT 49,056 Reputation points Microsoft Employee Moderator

2023-02-13T12:40:30.29+00:00

Cyril Carraz Did you try to upgrade the pricing tier and check if it helped? Thanks!!
romungi-MSFT 49,056 Reputation points Microsoft Employee Moderator

2023-02-15T09:55:36.8733333+00:00

Cyril Carraz Did you get a chance to check if the below answer worked for your?

Answer 1

romungi-MSFT 49,056 Microsoft Employee Moderator

Cyril Carraz This is most likely because you are using a resource which is on the free tier. This is documented on the limits and custom model pages of the documentation.

For PDF and TIFF files, up to 2,000 pages can be processed. With a free tier subscription, only the first two pages are processed.

Please update the pricing tier to standard or S0 from Azure portal and then relogin to the studio to check if additional pages are processed. The same applies to analyze operation too. Thanks!!

-Please kindly accept the answer if the answer was helpful to support the community, thanks.

Cyril Carraz 41 Reputation points

2023-02-16T08:32:21.4333333+00:00

Hello,

Yes, this was what was blocking the analysis past the 2nd page.

But I have another related problem: the PDFs I had already analysed before upgrading to S0 are still only analysed up to 2 pages and I don't see a button to reanalyse them. Since I had already labeled them, it would be very time consuming for me to redo them. Is there a way to analyse the rest of the pages and keep the old labels./annotations as is ?
romungi-MSFT 49,056 Reputation points Microsoft Employee Moderator

2023-02-16T10:22:27.89+00:00

For this scenario, you will have to open storage explorer of your storage account and locate the *.pdf.ocr.json file for the document that you have labeled the first 2 pages and rename this file to some other name and then reload the project. This will run OCR analysis again on your document and you will be able to use the existing labels on the first 2 pages and label the rest of the pages.

For ex: If file name is Rohit.pdf the OCR file in storage blob container will be Rohit.pdf.ocr.json, rename this file to 1_Rohit.pdf.ocr.json and reload your project from FR studio and it should run OCR analysis on the entire document and the label file that is already there in the storage container should apply the same labels on the first 2 pages and you can continue labeling for the rest of the pages and train the model.

Share via

Texts, tables, and selection marks aren't analysed past the 2nd page of a PDF

0 additional answers

Your answer