Can I turn off recognition for handwritten words using Azure document intelligence Service?

Elmongy, Samy 0 Reputation points
2023-12-11T20:31:57.9266667+00:00

We are using Azure document intelligence service to read receipts data.

what we need is to reject the receipts that include handwritten words, so the question is do we have that option through Azure portal, or it needs customization/development efforts.

Also, can we increase the confidence level of accepting the receipts, so we need to reject what below 90% confidence level.

Thanks.

Azure AI Document Intelligence
Azure AI Document Intelligence
An Azure service that turns documents into usable data. Previously known as Azure Form Recognizer.
1,843 questions
{count} votes

1 answer

Sort by: Most helpful
  1. navba-MSFT 27,185 Reputation points Microsoft Employee
    2023-12-12T05:31:53.65+00:00

    @Elmongy, Samy Welcome to Microsoft Q&A Forum, Thank you for posting your query here!

    .

    Azure Document Intelligence does have the capability to extract both printed and handwritten text from documents. However, it doesn’t inherently provide an option to reject documents with handwritten words. You can try to customize it from your application side and check if it is feasible.

    .

    As for the confidence level, Document Intelligence provides confidence scores for extracted fields. These scores represent an estimated probability that the prediction is correct. For scenarios where accuracy is critical, these confidence scores can be used to determine whether to automatically accept the prediction or flag it for human review. Therefore, you could potentially set a threshold (like 90% as you mentioned) and reject those below this confidence level. This might also require some customization from your application side.
    .

    For improving the accuracy and confidence scores of your model, Azure AI Document Intelligence suggests using the Human In The Loop (HITL) feature in the Studio. This feature allows you to selectively add documents that the model produces low confidence or erroneous results, thereby improving the model’s accuracy over time.
    .
    Hope this helps. If you have any follow-up questions, please let me know. I would be happy to help.

    0 comments No comments

Your answer

Answers can be marked as Accepted Answers by the question author, which helps users to know the answer solved the author's problem.