Azure Form Recognizer Read Container Image

azureuser123 1

Azure documentation states that the Form Recognizer Read Model is available as a container:
https://learn.microsoft.com/en-us/azure/applied-ai-services/form-recognizer/containers/form-recognizer-container-install-run?view=form-recog-3.0.0&tabs=layout

The pricing page indicates that as well:
https://azure.microsoft.com/en-us/pricing/details/form-recognizer/

However, the documentation does not list the Docker image for the "read" model anywhere (but only for the other 7 models):
https://learn.microsoft.com/en-us/azure/applied-ai-services/form-recognizer/containers/form-recognizer-container-image-tags?view=form-recog-3.0.0&tabs=current#feature-containers

Can anyone help me find the Docker image for the Read Model?

Gautam Kasturi 1 Reputation point

2022-11-11T14:13:16.467+00:00

I have a similar issue. I am trying to procure the Read Document Type that this user is talking about but I am not even trying to use it as a Container. Just regular Web. When I try to provision or price this in the Azure Portal this pricing tier does not even show up.

Can anyone help me procure this tier? i.e. Document Type = "Read". Refer screenshot attached.

gk
romungi-MSFT 42,286 Reputation points Microsoft Employee

2022-11-14T04:08:33.577+00:00

Are you not able to select the S0 pricing tier while creating the FR resource? This should be available as seen in the dropdown.

If you are unable to view it in the portal, then it could be an issue with the page not loading correctly.

1 answer

romungi-MSFT 42,286 Reputation points Microsoft Employee

2022-11-02T07:36:07.43+00:00
@azureuser123 The Read container is available at the following location:

mcr.microsoft.com/azure-cognitive-services/vision/read:3.2-model-2022-04-30

To setup and use the Read container please check the computer vision documentation on installing and configuring the container.
As mentioned in the setup of Receipt and OCR Read containers the image location is set to image: mcr.microsoft.com/azure-cognitive-services/vision/read:3.2 where you need to configure the computer vision endpoint and key and the pricing for this container is same as CV container i.e for 1M+ transactions - $0.60 per 1,000 transactions

If an answer is helpful, please click on or upvote which might help other community members reading this thread.
Please sign in to rate this answer.

1 person found this answer helpful.
azureuser123 1 Reputation point

2022-11-02T08:26:15.75+00:00

Thanks.
This is not the model I was referring to, though.

It is confusing, but there seem to be three Azure OCR-like models:

"Computer Vision v3.2 GA Read". The one you link to, right?

"Computer Vision v4.0 preview". New version of 1.

"Form Recognizer Read", aka "prebuilt-read". This is the one I am looking for.

romungi-MSFT 42,286 Reputation points Microsoft Employee

2022-11-02T11:39:03.067+00:00

@azureuser123 The first and the third should be the same container. Currently, the Receipt, Business Card and ID Document containers need the Read OCR container which are mentioned as part of pre-reqs of running the form recognizer containers. The docker compose files for all these setups use this container to setup the prebuilt API setup for Receipt, Business Card and ID Document.

I am checking internally to confirm if you can use the same configuration as CV container to use the prebuilt-read API as a container. Thanks!!

azureuser123 1 Reputation point

2022-11-08T11:40:21.007+00:00

Thanks for your answer, @romungi-MSFT !

Were you able to confirm this internally?

romungi-MSFT 42,286 Reputation points Microsoft Employee

2022-11-14T04:24:03.07+00:00

@azureuser123 The container for CV is same as the prebuilt-read. Please try to pull the same from the mentioned registry and configure the same for your other prebuilt models. Thanks!!

mcr.microsoft.com/azure-cognitive-services/vision/read:3.2-model-2022-04-30

azureuser123 1 Reputation point

2022-11-14T09:07:21.673+00:00

Thank you for clarifying this!
Sign in to comment