Azure AI-900 Vision Tasks Guide

AI-900 computer-vision questions are easier when you identify the expected output. Is the system assigning a label to the whole image, locating objects, reading text, or working with faces? That output usually tells you the task.

Core task types by output

OCR vs document processing

OCR is about extracting text from an image or document. Document processing often goes one step further and cares about document structure such as fields, tables, or form layout. AI-900 can test this boundary even if the formal computer-vision service bullets stay centered on visual capabilities.

High-value scenario clues

What strong answers usually do

• decide whether the output is a label, location, extracted text, or face-related result
• keep OCR separate from generic tagging
• remember that object detection answers both what and where
• avoid confusing face-specific tasks with broad scene analysis

Decision order that usually wins

1. First ask whether the task is label the whole image, locate objects, or read text from an image.
2. If the output is one category for the whole image, think image classification.
3. If the output must identify objects and their positions, think object detection.
4. If the task is extracting printed or handwritten text, think OCR.
5. AI-900 usually rewards identifying the task shape before naming a service.

Quiz