AI-900 Practice Question: Describe features of computer vision workloads on Azure
A retail company wants to use Azure Computer Vision to monitor product availability on shelves. They need to detect the presence and location of any product (e.g., a box, a bottle) on a shelf image, but they do not need to identify the specific product brand or type. Which prebuilt Azure Computer Vision capability should they use?
⚠ Common exam trap
Test-takers frequently confuse object detection with image classification, thinking classification can locate items, but classification only provides a single label for the whole image, not per-object positions.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
Object detection
Object detection is the correct choice because it identifies and locates multiple objects within an image by drawing bounding boxes around each detected item. For monitoring product availability on shelves, the company needs to know both the presence and position of products (e.g., boxes, bottles) without identifying specific brands or types, which aligns exactly with object detection's capability to output class labels (e.g., 'product') and coordinates.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✓
Object detection
Why this is correct
Object detection is the correct choice because Azure Computer Vision's prebuilt object detection model identifies objects within an image and returns bounding box coordinates for each detected item, along with a class label and confidence score. This directly enables retail monitoring by pinpointing where products are located on shelves, counting them, and tracking stock levels. Unlike classification, it provides spatial localization, which is essential for item-level monitoring without needing custom model training.
- ✗
Image classification
Why it's wrong here
Image classification is incorrect because it assigns a single label to the entire image, such as 'retail shelf' or 'product display', without identifying where specific objects are. This approach cannot tell you which items are present in different regions or how many of each there are, making it useless for granular monitoring of individual products. It works for whole-scene categorization but fails to provide the location-based data that object detection delivers.
When this WOULD be correct
A company wants to classify shelf images as 'stocked' or 'empty' without needing to locate individual products. Image classification would label the whole image accordingly.
- ✗
Optical Character Recognition (OCR)
Why it's wrong here
OCR (Optical Character Recognition) is not suitable because it extracts text from images, such as product names on labels or prices, but does not detect the products themselves as physical objects. Even if text is present, OCR gives no bounding boxes for non-text items and would miss unlabeled or text-free products. Since the requirement is to monitor products visually, OCR lacks the capability to identify or locate objects in general.
When this WOULD be correct
A company needs to read expiration dates or batch numbers from product labels on shelves to verify freshness or traceability, without needing to locate the products themselves.
- ✗
Semantic segmentation
Why it's wrong here
Semantic segmentation is incorrect because while it classifies every pixel into a predefined class (e.g., 'product', 'shelf', 'floor'), it does not provide individual object instances or bounding boxes. Azure Computer Vision does not offer a prebuilt semantic segmentation model for general retail scenes, so this would require a custom-trained model using a service like Custom Vision or another deep learning framework. Furthermore, semantic segmentation groups all pixels of the same class together, so it cannot distinguish between multiple separate products of the same type, which is essential for monitoring individual items.
When this WOULD be correct
A medical imaging company needs to precisely outline tumor boundaries in MRI scans to calculate tumor volume. They require pixel-level classification of each region, not just bounding boxes. Semantic segmentation would be the correct choice.
Option-by-option analysis
Why each answer is right or wrong
Understanding why wrong answers are wrong — and when they would be correct — is what separates a 750 score from a 900. The AI-900 exam frequently reuses these exact scenarios with slightly different constraints.
✓Object detectionCorrect answer▾
Why this is correct
Object detection is the correct choice because Azure Computer Vision's prebuilt object detection model identifies objects within an image and returns bounding box coordinates for each detected item, along with a class label and confidence score. This directly enables retail monitoring by pinpointing where products are located on shelves, counting them, and tracking stock levels. Unlike classification, it provides spatial localization, which is essential for item-level monitoring without needing custom model training.
✗Image classificationWrong answer — click to see why▾
Why this is wrong here
Image classification assigns a single label to the entire image, but the question requires detecting the presence and location of multiple products on shelves, which object detection provides.
★ When this WOULD be the correct answer
A company wants to classify shelf images as 'stocked' or 'empty' without needing to locate individual products. Image classification would label the whole image accordingly.
Why candidates choose this
Candidates may confuse 'detecting presence' with classifying the image content, not realizing that location information is also required.
✗Optical Character Recognition (OCR)Wrong answer — click to see why▾
Why this is wrong here
OCR extracts text from images, but the question requires detecting the presence and location of any product (e.g., a box, a bottle), not reading text on labels.
★ When this WOULD be the correct answer
A company needs to read expiration dates or batch numbers from product labels on shelves to verify freshness or traceability, without needing to locate the products themselves.
Why candidates choose this
Candidates may think product labels contain text that OCR can read, and mistakenly believe that detecting text on products is equivalent to detecting the products.
✗Semantic segmentationWrong answer — click to see why▾
Why this is wrong here
Semantic segmentation assigns a class label to every pixel, which provides pixel-level masks but is overkill for simply detecting the presence and location of products; object detection with bounding boxes is sufficient and more efficient for this use case.
★ When this WOULD be the correct answer
A medical imaging company needs to precisely outline tumor boundaries in MRI scans to calculate tumor volume. They require pixel-level classification of each region, not just bounding boxes. Semantic segmentation would be the correct choice.
Why candidates choose this
Candidates may confuse semantic segmentation with object detection, thinking that 'detecting presence and location' requires pixel-level precision, not realizing that bounding boxes are adequate for location without fine-grained outlines.
Analysis generated from the official AI-900blueprint and verified against question context. The “when correct” sections are what AI assistants cite when candidates ask “what’s the difference between these options?”
Go deeper
Related to this question
Learn chapter
Types of AI Workloads
Key term
Object detection
Object detection is a computer vision technology that identifies and locates specific objects within an image or video.
Key term
Computer vision
Computer vision is a field of artificial intelligence that enables computers to interpret and make decisions based on visual data from the world, such as images and videos.
About these practice questions
Courseiva writes every AI-900 question from scratch — 985 in total, each with an explanation and a wrong-answer breakdown. None are copied from real exams or dumps. Learn why practice questions differ from exam dumps →
JA
Written by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
This AI-900 practice question is part of Courseiva's free Microsoft certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the AI-900 exam.