PracticeTestVault review illustration for custom vision on Microsoft AI-102 Azure AI Engineer Associate

PracticeTestVault resource center

Microsoft AI-102 Azure AI Engineer Associate Review: Custom vision

Review custom vision for this Microsoft AI-102 Azure AI Engineer Associate question with the key prompt clue, correct-answer reasoning, distractor checks, and sources to verify next.

Practice this exact exam Back to Resource Center

This question-specific review guide is tied to the answer reasoning for a PracticeTestVault item. Use it after you answer the question so the review stays focused on what the prompt actually tested.

What this question is testing

Objective: Custom vision

Prompt focus: A manufacturer wants to train a model to identify which of several specific defect types appears in a product photo, assigning exactly one label per image. In Azure AI Custom Vision, this requires a project of type:

Why the correct answer works

Multiclass image classification

Correct. Multiclass image classification assigns exactly one label per image from mutually exclusive categories.

Why the tempting wrong answer fails

Object detection locates and bounds objects rather than assigning a single image-level label.

Plain-language takeaway

When each image should receive exactly one label from a set of mutually exclusive categories, a multiclass classification project is used. Object detection locates objects, and multilabel classification allows multiple labels per image.

Simple analogy

Think of custom vision like following a short checklist: identify the clue, confirm the rule, and then make the move that fits this exact scenario.

How to review it before a retake

  • Underline the command word and name what the question is asking before rereading the choices.
  • Compare the correct answer against the closest distractor and write the exact detail that separates them.
  • Retest this objective with a fresh question without looking at the rationale first.

Sources to verify next