Community

Enable input / upload to API directly from Azure Storage accounts

As resent as January of this year (2018), it appears that your system expects files to come from a local file system. This notion breaks "cloud"!

https://forum.ocrsdk.com/thread/can-input-and-output-be-azure-blob-storage/?order=all#comment-2ec0aa1b-5d9e-461f-9a50-a85d005384fb

Please enable the ability to upload / process / input images that are stored as blobs in Azure Storage Accounts.

Thanks

0

Comments

3 comments

  • Avatar
    Oksana Serdyuk

    Hello,

    Unfortunately, we do not plan to implement this possibility in the nearest future.

    0
    Comment actions Permalink
  • Avatar
    Adrian

    Thanks for the PM. This method may work.

    It is strange that you don't see this as a nessesary requirement and I'd love to find out more on the reasoning behind this decision.

    My only thought is that you assume all images to be OCR'd come from an MFD.

    For me, I am looking to use Azure Function Apps and Logic Apps. Both serverless features, that don't have a file system. I know I can create custom connectors to read the local file system, but that only gets the file into the storage account where the file then has to be processed by your API.

    Cost limitations require "consumption" plans that ideally trigger only when a new blob appears. Azure Functions include this capability, almost out-the-box. Once the file is OCR'd, it needs to be indexed and stored in SharePoint online. My intended workflow is pure PaaS, and having to create an IaaS VM to run your API is nonsensical and probably will result in the solution being cost-prohibitive.

    As I understand it, your API is hosted in Azure, and the processed image is downloaded from a storage account, therefore I don't see you would have a technical challenge in implementing this.

    Thanks.

    0
    Comment actions Permalink
  • Avatar
    Oksana Serdyuk

    Hi,

    Thank you for so detailed description of your usage scenario. I have forwarded this information to our product analysts, they have promised to think about it. In general, we plan to expand the functionality of our processing methods in future, but at the moment we have more prioritized features for implementation. So far we cannot provide you with information about any estimated dates.

    0
    Comment actions Permalink

Please sign in to leave a comment.