Document Intelligence Studio OCR Fails to Recognize Dash/Hyphen in PDF Tables

Question

Document Intelligence Studio OCR Fails to Recognize Dash/Hyphen in PDF Tables

P, Priyadharshini 0

Hi,

When working with Document Intelligence Studio (Layout), we're encountering an issue where the OCR functionality fails to recognize "-" (dash/hyphen) characters within tables in PDF statements. Instead of identifying these characters, it marks the content as empty or blank. This problem wasn't present in the API version from 2022, but it has been occurring with versions from 2023 onwards.

Attached the test data for reference test_pdf.pdf

Anonymous

2024-07-18T13:08:59.4733333+00:00

Hi @P, Priyadharshini,

Did you got any chance to check the below response?
Anonymous

2024-07-19T11:45:20.56+00:00

Hi @P, Priyadharshini,Following up to see if the below answer was helpful. If this answers your query, do click Accept Answer and Yes for was this answer helpful. And, if you have any further query do let us know.

1 answer

Your answer

Anonymous

2024-07-18T13:08:59.4733333+00:00

Hi @P, Priyadharshini,

Did you got any chance to check the below response?
Anonymous

2024-07-19T11:45:20.56+00:00

Hi @P, Priyadharshini,Following up to see if the below answer was helpful. If this answers your query, do click Accept Answer and Yes for was this answer helpful. And, if you have any further query do let us know.

Answer 1

Hi @P, Priyadharshini,

Thank you for reaching out to Microsoft Q&A forum and reporting the issue you're experiencing with Azure Document Intelligence Studio (Layout) failing to recognize dash/hyphen characters in tables within PDF statements is known to occur in versions from 2023 onwards, whereas it wasn't present in the 2022 API version. This can be due to changes or updates in the OCR engine. As a workaround, you can preprocess the PDF to enhance the visibility of these characters or consider reverting to an earlier API version if feasible.

Additionally, raising a support ticket with Microsoft might help to address the issue or get a patch fix. Here is the link: Azure support request.

Also, you can share your feedback through the Azure Feedback Forum: Post idea · Community (azure.com)

I hope you understand. Do let us know if you any further queries.

If this answers your query, do click Accept Answer and Yes for was this answer helpful.

Daniel 0 Reputation points

2024-09-29T10:45:52.0566667+00:00

This is not a visibility issue in PDF, clearly some OCR engine update broke this. We are using Document intelligence to OCR books, and now many dashes in dialogues are just omitted whereas previously they were recognized correctly. For our use case, document intelligence became totally unusable now.