Skip to main content

Base64 Input Support & Model Identification in Mindee V2 (.NET SDK)

Feature Request

Hi Mindee Team,

I’m using the Mindee V2 API with the C# (.NET) SDK.

  1. Is Base64 image input supported directly in V2? In the .NET SDK, I only see LocalInputSource and UrlInputSource. Currently, I have to convert Base64 → file → LocalInputSource. Is this the recommended approach or will direct Base64 support be added?

  2. In V2, is modelId the only way to identify invoice, receipt, and passport models? how to differentiate V2 models such as Invoice, Passport and Expense Receipt using the respective modelIds.

Thanks.

Posted by Kavya Kavya ·
Status: Completed7 comments

Log in to comment and vote

Comments7

  • Sebastian Olivera Silvera

    •

    Jan 6

    Hello,

    Thank you for your feedback.


    It would seem that base 64 support has never been asked for our .NET SDK before. I believe I can add it to the next release. In the meantime, converting to a file or stream seems like the best approach indeed.

    Regarding model classification, if you mean how can you differentiate between your models, you can either map them on your end, or ping https://api-v2.mindee.net/v2/search/modelswith an Authorize header containing your API key to get a list of your model IDs and names.

    The curl call should look something like:

    curl "https://api-v2.mindee.net/v2/search/models" -H 'Authorization: <your_api_key_here>

    You can also directly try here:

    https://api-v2.mindee.net/docs#/Inference/Search_Available_Models_v2_search_models_get


    Don't hesitate if you need anything else.

    • Sebastian Olivera Silvera

      •

      Jan 7

      Update: version 3.37.0 now supports native loading of base64-encoded strings:

      https://www.nuget.org/packages/Mindee/3.37.0

      Simply load your file using:

      var base64Input = new LocalInputSource(base64TextString, "myFileName.jpg");