When preparing a dataset for training a custom object detection model in Azure Custom Vision, what is the correct way to label images?

Provide a text description of the objects in the image

Assign a single tag to the entire image

Draw bounding boxes around each object of interest and assign a tag to each box

Upload images without any labels and let the service auto-label them

A developer has a published custom vision image classification model that identifies different types of flowers. The developer wants to write a Python script that sends a photo of a flower to the model and prints the predicted class and confidence score. Which code snippet should the developer use?

Use the TrainingClient from the same package, call train_project, and print the results.

Use the PredictionClient from the azure-cognitiveservices-vision-customvision package, call classify_image with the image file, and print the predictions.

Use the Computer Vision SDK's analyze_image method.

Use the Azure Blob Storage SDK to upload the image and get predictions.

A medical imaging team trained a custom vision model to detect tumors in X-ray images. The evaluation shows a precision of 0.98 and a recall of 0.70 for the 'tumor' class. The clinical workflow requires that no tumor is missed, and false positives are acceptable because they will be reviewed by a radiologist. Which metric should the team optimize, and what is the best action?

Optimize accuracy; balance the dataset with more non-tumor images.

Optimize recall; add more tumor images and retrain the model.

Optimize precision; switch to object detection to get bounding boxes for tumors.

Optimize precision; add more non-tumor images to reduce false positives.

A custom vision model for detecting tumors in X-ray images has high accuracy (0.97) but low recall (0.55) for the tumor class. The team is concerned about missing tumors. They have limited ability to collect more training data. What is the best approach to improve recall without sacrificing too much precision?

Switch to an object detection model

Lower the classification threshold and then evaluate the precision-recall trade-off

Raise the classification threshold to increase precision

Retrain the model with only tumor images

A developer needs to create a custom vision model to classify images of different bird species. The developer wants to write code that creates the project, uploads images with tags, trains the model, and publishes it — all without using the web portal. Which SDK should the developer use?

Azure Cognitive Services Language SDK

Microsoft Certified:Azure AI Engineer Associate

Domain 4Objective 2

Implement Custom Vision Models AI-102 Practice Questions (Page 2)

Part of the Implement computer vision solutions domain, which accounts for 10-15% of the AI-102 exam. Microsoft does not publish an official question count, but from its 100-minute exam (~40–65 total, ~4–10 in this domain), expect 1–3 from this objective — we provide 36 practice questions to prepare you well beyond it. (estimate)

36questions here

8free pages

7concepts

10-15%of the exam

Questions 6–10

6foundation · easy
When preparing a dataset for training a custom object detection model in Azure Custom Vision, what is the correct way to label images?
Select an answer first
7application · medium
A developer has a published custom vision image classification model that identifies different types of flowers. The developer wants to write a Python script that sends a photo of a flower to the model and prints the predicted class and confidence score. Which code snippet should the developer use?
Select an answer first
8application · medium
A medical imaging team trained a custom vision model to detect tumors in X-ray images. The evaluation shows a precision of 0.98 and a recall of 0.70 for the 'tumor' class. The clinical workflow requires that no tumor is missed, and false positives are acceptable because they will be reviewed by a radiologist. Which metric should the team optimize, and what is the best action?
Select an answer first
9expert · hard
A custom vision model for detecting tumors in X-ray images has high accuracy (0.97) but low recall (0.55) for the tumor class. The team is concerned about missing tumors. They have limited ability to collect more training data. What is the best approach to improve recall without sacrificing too much precision?
Select an answer first
10application · medium
A developer needs to create a custom vision model to classify images of different bird species. The developer wants to write code that creates the project, uploads images with tags, trains the model, and publishes it — all without using the web portal. Which SDK should the developer use?
Select an answer first

Finished these 5 questions?

Review the revealed explanations, or continue through the curriculum.

Next 5 questions →Back to AI-102

Free Basic Practice is a study aid with revealable answers — not a scored exam. Examers.io is independent and not affiliated with or endorsed by Microsoft. “AI-102” is a trademark of its owner, used for identification only.

Implement Custom Vision Models AI-102 Practice Questions (Page 2)

Questions 6–10

When preparing a dataset for training a custom object detection model in Azure Custom Vision, what is the correct way to label images?

A developer has a published custom vision image classification model that identifies different types of flowers. The developer wants to write a Python script that sends a photo of a flower to the model and prints the predicted class and confidence score. Which code snippet should the developer use?

A developer needs to create a custom vision model to classify images of different bird species. The developer wants to write code that creates the project, uploads images with tags, trains the model, and publishes it — all without using the web portal. Which SDK should the developer use?

Keep studying — Implement computer vision solutions