AI-102 Implement computer vision solutions — 15 Questions

Question 1

A hospital uses Azure Custom Vision to classify X-ray images as normal or abnormal. The model achieves 98% accuracy on the test set. However, during deployment, the model misclassifies many abnormal cases as normal, causing missed diagnoses. The hospital has a class imbalance where abnormal cases are only 5% of the data. What should the data scientist do first to address this?

Choose an answer to begin — your selection is scored in the full session.

Accepted Answer

Use oversampling or class-weight techniques to balance the training.. The primary issue is class imbalance, where abnormal cases constitute only 5% of the data. Oversampling (e.g., SMOTE) or class-weight techniques adjust the training process to give more importance to the minority class, directly addressing the model's bias toward the majority class and reducing false negatives. This is a standard preprocessing step in Custom Vision and other ML frameworks before tuning hyperparameters or changing algorithms.

Answer

Increase the number of training epochs.

Answer

Add more normal X-ray images to the dataset.

Answer

Switch to a different object detection algorithm.

Question 2

A developer is building an application to extract text from scanned invoices using Azure Computer Vision's Read API. The invoices contain a mix of printed and handwritten text. The developer needs to ensure the highest accuracy for both types. Which parameter should they set in the API call?

Accepted Answer

No special parameter; the Read API automatically handles both.. The Read API in Azure Computer Vision is designed to extract text from images and documents, and it automatically handles both printed and handwritten text without requiring any special parameter. Setting the 'language' parameter to 'en' is optional and only improves accuracy for language-specific text, but it does not enable or disable handwriting recognition. Therefore, no additional parameter is needed to achieve the highest accuracy for both types.

Answer

Set the 'language' parameter to 'en' for English handwriting.

Answer

Specify the 'model-version' as '2022-04-30'

Answer

Use the 'mode' parameter set to 'Handwriting'

Question 3

You are building an application that processes scanned invoices to extract key fields such as total amount, invoice date, and vendor name. The application uses Azure AI Document Intelligence. You need to ensure high accuracy for field extraction without manual labeling. Which feature should you use?

Accepted Answer

Use the General Document model. The General Document model in Azure AI Document Intelligence can extract common fields like total amount, invoice date, and vendor name from invoices without requiring any labeled training data. While a custom neural model would offer higher accuracy, it requires manual labeling of sample documents, which contradicts the requirement of no manual labeling. Therefore, the General Document model is the appropriate choice.

Answer

Use the Read API

Answer

Use a custom neural model

Answer

Use the Layout API

Question 4

A retail company uses Azure AI Vision to analyze store shelf images for product availability. The solution uses an object detection model trained on custom products. Recently, the model's performance dropped significantly due to new packaging designs. You need to improve the model's accuracy with minimal manual effort. What should you do?

Accepted Answer

Use Azure AI Vision model customization with active learning. Azure AI Vision model customization with active learning is the correct choice because it allows the model to automatically identify images where it is uncertain (low confidence predictions) and prioritize those for labeling and retraining. This minimizes manual effort while directly addressing the performance drop caused by new packaging designs, as the model iteratively improves on the specific data distribution shift without requiring a full retraining from scratch.

Answer

Adjust the confidence score threshold to reduce false negatives

Answer

Use the Image Analysis 4.0 dense captioning feature

Answer

Collect a new set of images and retrain the model from scratch

Question 5

A healthcare organization uses Azure AI Health Insights to extract medical insights from unstructured clinical notes. The solution must comply with HIPAA. Which configuration is required?

Accepted Answer

Configure the Azure AI services resource to use a private endpoint and disable public network access. HIPAA compliance for Azure AI Health Insights requires network isolation to prevent unauthorized access to protected health information (PHI). A private endpoint assigns the Azure AI services resource a private IP address within the customer's virtual network, and disabling public network access ensures all traffic stays within the Microsoft backbone, eliminating exposure to the public internet. This configuration meets the HIPAA Security Rule's requirement for technical safeguards, specifically access control and transmission security.

Answer

Store the clinical notes in Azure Storage with Azure AD authentication only

Answer

Enable public network access but use a firewall rule to restrict IP addresses

Answer

Use a system-assigned managed identity for authentication

Question 6

You deploy a custom vision model for defect detection on a manufacturing line. The model runs on an Azure IoT Edge device. You notice that inference latency is too high for real-time detection. Which action should you take to reduce latency?

Accepted Answer

Convert the model to TensorFlow and use the Azure IoT Edge Deep Learning module with hardware acceleration. Converting the model to TensorFlow enables compatibility with the Azure IoT Edge Deep Learning module, which can leverage hardware acceleration (e.g., Intel Movidius or NVIDIA GPUs) to significantly reduce inference latency. This approach keeps inference on the edge device, avoiding network round-trips, and optimizes the model for real-time defect detection.

Answer

Move inference to Azure Functions in the cloud

Answer

Retrain the model with more defect images

Answer

Increase the resolution of input images

Question 7

You are designing a computer vision solution that must detect and redact personally identifiable information (PII) from images of ID cards. The solution must comply with GDPR. Which THREE components should you include? (Select three.)

Accepted Answer

Azure AI Document Intelligence redaction. Correct options: A, C, D. Azure AI Vision OCR (C) extracts text from ID card images. Azure AI Language PII detection (D) identifies PII entities in the extracted text. Azure AI Document Intelligence redaction (A) can redact the identified PII from the document. Option B (Azure AI Custom Vision) is used for custom image classification, not for text-based PII detection. Option E (Azure AI Video Indexer) is for analyzing video content, not static ID cards.

Answer

Azure AI Custom Vision

Answer

Azure AI Video Indexer

Question 8

A company uses Azure AI Custom Vision to classify images of products. The model is deployed to a mobile app. The app sends images to the Custom Vision prediction endpoint. Users report that the app is slow when the network is poor. You need to enable offline inference on the mobile device. What should you do?

Accepted Answer

Export the model as TensorFlow Lite and integrate it into the mobile app. Exporting the Custom Vision model as TensorFlow Lite allows it to run directly on the mobile device, enabling offline inference without requiring network connectivity. This eliminates latency from poor network conditions by processing images locally, which is the only way to achieve true offline inference in a mobile app.

Answer

Use the Custom Vision compact model with smaller image size

Answer

Increase the prediction API timeout

Answer

Deploy the model to an Azure IoT Edge device

Question 9

A bank uses Azure AI Document Intelligence to process loan applications. The solution must extract data from scanned PDFs and validate it against a database. The bank requires that all extracted data be encrypted at rest and in transit. Which security measure should you implement?

Accepted Answer

Enable customer-managed keys (CMK) with Azure Key Vault for the Document Intelligence resource. Customer-managed keys (CMK) with Azure Key Vault provide the ability to control the encryption keys used to protect data at rest in Azure AI Document Intelligence. This, combined with the platform's default encryption in transit (TLS), satisfies the requirement for encrypting extracted data both at rest and in transit. CMK is the only option that directly addresses encryption key management for data at rest.

Answer

Use a system-assigned managed identity for the application

Answer

Configure a private endpoint for the Document Intelligence resource

Answer

Use Azure RBAC to restrict access to the Document Intelligence resource

Question 10

You are building an application that analyzes images of handwritten notes. The application must extract the handwritten text and preserve the original layout, including line breaks and indentation. Which Azure AI service should you use?

Accepted Answer

Azure AI Document Intelligence Read model. The Azure AI Document Intelligence Read model is specifically designed to extract printed and handwritten text from documents while preserving the original layout, including line breaks and indentation. It uses advanced OCR capabilities optimized for document-like content, making it the correct choice for this scenario.

Answer

Azure AI Vision OCR

Answer

Azure AI Document Intelligence Layout model

Answer

Azure AI Translator

Question 11

An application uses Azure AI Vision to analyze images and extract text. The application crashes when processing images with embedded barcodes. You suspect the issue is related to the image pre-processing. Which step should you add to the pipeline to resolve the issue?

Accepted Answer

Detect and remove barcodes from the image before OCR. Azure AI Vision's OCR engine is designed to extract text from natural images and documents, but embedded barcodes can introduce noise or unexpected patterns that interfere with the text detection algorithm. By detecting and removing barcodes from the image before OCR, you eliminate this interference, allowing the OCR engine to focus on textual content without crashing or producing erroneous results.

Answer

Increase the image contrast before sending to the OCR engine

Answer

Add more training data with barcodes to the OCR model

Answer

Resize the image to a smaller resolution to reduce barcode impact

Question 12

You need to analyze a live video stream from a security camera to detect people entering a restricted area. Which Azure AI service should you use?

Accepted Answer

Azure Video Indexer. Azure Video Indexer supports live video stream analysis for detecting people in restricted areas using pre-built AI models. It can process live streams and generate alerts in real time. Azure Video Analyzer for Media is a deprecated predecessor and is not the optimal choice. Custom Vision requires custom model training, and Face API focuses on facial recognition rather than general person detection.

Answer

Azure AI Custom Vision

Answer

Azure Video Analyzer for Media (deprecated)

Answer

Azure AI Face API

Question 13

You are designing a solution that extracts information from scanned invoices. The solution must automatically classify invoices by vendor and extract key fields (total amount, date, invoice number). Which combination of Azure AI services should you use?

Accepted Answer

Azure AI Document Intelligence custom classification and extraction models. Azure AI Document Intelligence (formerly Form Recognizer) allows you to build custom classification models to identify the vendor from the invoice layout, and custom extraction models to pull key fields like total amount, date, and invoice number. This combination handles both the classification and extraction requirements in a single service, using labeled training data for high accuracy.

Answer

Azure AI Vision OCR and Azure AI Language

Answer

Azure AI Form Recognizer (deprecated) with custom models

Answer

Azure AI Document Intelligence pre-built invoice model

Question 14

An application uses Azure AI Face API to perform face detection and verification. The application must ensure that only users with verified identities can access sensitive data. Which additional Azure service should you integrate to comply with Microsoft's Responsible AI standards for facial recognition?

Accepted Answer

Microsoft Entra ID and Face API Limited Access. D is correct because Microsoft's Responsible AI standards for facial recognition require that applications using the Face API for identity verification must be backed by a verified identity system like Microsoft Entra ID (formerly Azure AD) and must apply for the Face API Limited Access policy. This ensures that only authorized users with verified identities can access sensitive data, aligning with Microsoft's ethical guidelines for facial recognition technology.

Answer

Azure Video Indexer

Answer

Azure AI Content Safety

Answer

Azure AI Vision Image Analysis

Question 15

Which TWO Azure AI services provide capabilities to detect and analyze faces in images? (Choose two.)

Accepted Answer

Azure AI Face API. Azure AI Face API (B) is correct because it is specifically designed for face detection, recognition, and analysis, including attributes like age, emotion, and landmarks. Azure AI Vision Image Analysis (D) is correct because it includes the 'Detect Faces' capability within its broader image analysis feature set, allowing extraction of face bounding boxes and attributes without requiring a dedicated Face API resource.

Answer

Azure AI Custom Vision

Answer

Azure AI Document Intelligence

Answer

Azure Video Indexer