Microsoft Certified: Azure AI Engineer AssociateImplement image and video processing solutionsMedium

A security firm is developing an automated surveillance system for a remote industrial site. The system needs to detect trespassing events (e.g., unauthorized person entering a restricted area) in real-time. Due to limited internet connectivity at the site, processing must occur locally with minimal latency. Which Azure AI deployment option for a custom vision model should be chosen?

  1. ACloud endpoint deployment
  2. BIoT Edge deployment
  3. CAzure Functions with Computer Vision SDK
  4. DAzure Batch AI
Show answer & explanation

Correct answer: B. IoT Edge deployment

IoT Edge deployment allows custom vision models to run directly on local devices (edge devices) at the site, enabling real-time processing with low latency and reducing reliance on continuous cloud connectivity, which is critical for remote sites with limited internet.

Why the other options are wrong

  • A. Cloud endpoint deployment requires constant, reliable internet connectivity for every inference.
  • C. Azure Functions still execute in the cloud, requiring internet access for processing.
  • D. Azure Batch AI is for large-scale training jobs, not real-time inference at the edge.

Custom Vision IoT Edge Deployment

Deploying a custom-trained Azure Custom Vision model onto an IoT Edge device for local, low-latency inference.

  • Enables AI processing at the edge, close to data sources
  • Reduces latency and bandwidth usage
  • Allows for offline operation with intermittent connectivity
  • Requires an IoT Edge runtime environment

Memory trick: IoT Edge brings the AI 'brain' right to the device, not just the cloud.

More Implement image and video processing solutions questions