Computer Vision API - OCR to Json V1 and V3 action never returns valid results

Question

Computer Vision API - OCR to Json V1 and V3 action never returns valid results

Omokhoa Agbagbara 95

I have been working with some OCR integrations and came upon the Computer Vision Api connector with the OCR to Json V1 and V3 actions.

However, the actions do not return any valid result.

output

With this input or any other input which I have tried.

11104-035-Cropped - 20

However, if a function app is used as a proxy to call the imageAnalysisClient. The results are different, and I have summarized the response below.

output-01

Is this a known bug which will be fixed or is there an extra parameter to be sent in the request.

Below is the setup for the action.

config

Best Regards

Sandhya Kommineni 2,895 Reputation points Microsoft External Staff Moderator

2025-07-25T10:49:17.93+00:00

Hello Omokhoa Agbagbara ,

Thank you for posting in Microsoft Q&A forum.

I am checking with the internal team on the below and will update you here once I hear back from them.

I understand that the OCR to JSON V3 action leverages the Image Analysis REST API: Image Analysis - Analyze Image - REST API (Azure Azure AI Services) | Microsoft Learn

If that’s the case, you can still invoke the REST API directly using the HTTP action in Logic Apps.

If you have any queries, please feel free to post them here.

1 answer

Your answer

Sandhya Kommineni 2,895 Reputation points Microsoft External Staff Moderator

2025-07-25T10:49:17.93+00:00

Hello Omokhoa Agbagbara ,

Thank you for posting in Microsoft Q&A forum.

I am checking with the internal team on the below and will update you here once I hear back from them.

I understand that the OCR to JSON V3 action leverages the Image Analysis REST API: Image Analysis - Analyze Image - REST API (Azure Azure AI Services) | Microsoft Learn

If that’s the case, you can still invoke the REST API directly using the HTTP action in Logic Apps.

If you have any queries, please feel free to post them here.

Answer 1

It really does seem like the issue lies in how the built-in OCR to JSON actions are implemented within Logic Apps. I’ve run into the same behavior — calling the Image Analysis API directly via an HTTP request works just fine and returns valid OCR results. But when using the official V1 or V3 OCR to JSON actions, the output is either empty or completely invalid.
At this point, the more reliable workaround might be to bypass the built-in actions altogether and call the REST API directly using an HTTP action or route the request through an Azure Function as a proxy. It’s a bit more setup, but at least you get consistent results.
Would definitely be helpful if Microsoft could confirm whether this is a known issue or if there’s something missing in how the actions are configured under the hood.

Share via

Computer Vision API - OCR to Json V1 and V3 action never returns valid results

1 answer

Your answer