AI-102 Question 158
Single answerYou have trained a custom vision model using Azure Custom Vision and published it to an endpoint. You want to integrate this model into an application to classify images provided by users. Which step is necessary to consume the custom vision model in your application?
- A
Obtain the prediction URL and key from the Custom Vision portal.
- B
Export the model to ONNX format and integrate it directly into your application without using the endpoint.
- C
Create a Logic App to call the model automatically.
- D
Use Azure Functions to deploy the model to a different endpoint before consuming it.
Show answer and explanation
Correct answer: A
Explanation
To consume a custom vision model in an application, you must retrieve the prediction URL and key from the Custom Vision portal. These credentials allow the application to send prediction requests to the model's endpoint. Other options, such as exporting the model or using Azure Functions, are not required for this specific scenario.
- A. Correct.
Correct. To consume a custom vision model, you need the prediction URL and key provided in the Custom Vision portal. These are required to make API calls to the model's endpoint.
- B. Incorrect.
Incorrect. Exporting to ONNX is optional and not necessary for consuming the deployed model via the Custom Vision endpoint.
- C. Incorrect.
Incorrect. While Logic Apps can be used for automation, they are not a required step for consuming the model directly in your application.
- D. Incorrect.
Incorrect. Azure Functions can be used for other use cases, like adding preprocessing logic, but deploying the model to another endpoint is unnecessary.