Data fetching issue in Azure OCR

Question

4-f940-chromeprinted.pdf

5 digit pin not getting fetched using form recogniser and custom extraction model in this pdf

4-f940 (1) (1).pdf

while it gets fetched in second pdf i gave

the issues come when we print pdfs as image from WPS office or by chrome.

Answer 1

Hi @Mayank Arora,

Thank you for reaching out to Microsoft Q&A forum!

When tested with the prebuilt model, the 5-digit PIN is successfully fetched (see below image), indicating the issue may be with the custom extraction model.

So, in this case, I recommend you to train the custom extraction model with a larger variety of documents, especially those similar to the problematic PDF. This should improve the model’s ability to accurately recognize the 5-digit PIN in different formats. However, you can also use Prebuilt Layout model.

Here are some possible causes:

• Image Quality: Printing as an image may reduce the quality or alter the text recognition.
• Font and Formatting Changes: Differences in font rendering and layout between the original and printed image versions can affect text extraction.
• OCR Limitations: Optical Character Recognition (OCR) may have difficulty recognizing text in images compared to vector text.

Hope this helps. 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.