Edit

Share via

Azure AI Vision vectorizer

  • Article

Important

This feature is in public preview under Supplemental Terms of Use. The 2024-05-01-Preview REST API supports this feature.

The Azure AI Vision vectorizer connects to an Azure AI Vision resource to generate embeddings at query time using the Multimodal embeddings API. Refer to the region availability for multimodal embeddings to determine the regions where this model is accessible. Your data is processed in the Geo where your model is deployed.

Note

This vectorizer is bound to Azure AI services. Execution of the vectorizer is charged at the existing Azure AI services pay-as-you go price.

Vectorizer parameters

Parameters are case-sensitive.

Parameter name Description
resourceUri The URI of the AI Services resource.
apiKey The API key of the AI Services resource.
modelVersion (Required) The model version to be passed to the Azure AI Vision API for generating embeddings. It's important that all embeddings stored in a given index field are generated using the same modelVersion. For information about version support for this model refer to multimodal embeddings.
authIdentity A user-managed identity used by the search service for connecting to AI Services. You can use either a system or user managed identity. To use a system manged identity, leave apiKey and authIdentity blank. The system-managed identity is used automatically. A managed identity must have Cognitive Services User permissions to use this vectorizer.

Supported vector query types

The Azure AI Vision vectorizer supports text, imageUrl, and imageBinary vector queries.

Expected field dimensions

A field configured with the Azure AI Vision vectorizer should have a dimensions value of 1024.

Sample definition

"vectorizers": [
    {
        "name": "my-ai-services-vision-vectorizer",
        "kind": "aiServicesVision",
        "aiServicesVisionParameters": {
            "resourceUri": "https://westus.api.cognitive.microsoft.com/",
            "apiKey": "0000000000000000000000000000000000000",
            "authIdentity": null,
            "modelVersion": "2023-04-15"
        },
    }
]

See also