How to get a MS Read like OCR result from Document Intelligence Custom Extraction Model

Question

I am looking to see if it is possible to when calling a Document Intelligence get not only the response from the custom model that we are looking for, but also get a MS Read type response to be able to create an entire "Text Layer" on a document that is passed to be able to do "Full Text" type searches later. When using the Document Intelligence response, it appears as if we can only get the response from the model which will give us back the labels that we trained as well as the data associated and the confidence, which is perfect. However, we would also like to get more of a "pure OCR" layer back as well that Read provides. Is that possible without calling them both?

Answer 1

Hi Spisak, Josh,
Currently, there is no direct way to get both outputs (custom extraction and full OCR) in a single API call. However, we can achieve this by:

Calling Both Models Separately:

• Use the Custom Extraction Model to get the structured data for your specific fields.
• Use the Read Model to extract the full text layer for the document.

Implementation:

• You can integrate both calls into your workflow. For example:
• First, call the Read model to extract the full text layer.
• Then, call the Custom Extraction model to extract the specific fields you need.

Efficiency:

While this approach requires two API calls, you can optimize it by running them in parallel if your application allows.

Hope this helps. Do let us know if you have any further queries.

---

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