AI-900 Practice Question: Describe features of computer vision workloads on Azure

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✗

  A cloud-only AI service for high-accuracy computer vision inference

  Why it's wrong here

  Azure Percept was explicitly an edge-first platform, with its neural processing unit (NPU) and VPU enabling on-device inference; it was not a cloud-only service. The core value proposition of Azure Percept was to run vision and speech models locally so that devices could operate with low latency and maintain privacy even without internet connectivity. Cloud services such as Azure AI Vision (Computer Vision) provide high-accuracy inference in the cloud, but that is a separate offering, not Azure Percept, which did support cloud connectivity for model management but did not require it for inference.

• ✓

  An edge AI hardware platform for deploying vision and speech AI models locally on devices

  Why this is correct

  Azure Percept was a hardware and software platform designed to run computer vision and speech AI models directly on edge devices. Its development kit included a vision system-on-module (SoM) with an Intel Movidius Myriad X VPU for local neural network inference, plus a four-microphone audio SoM for voice scenarios. By processing audio and video locally without a persistent internet connection, it enabled low-latency, privacy-preserving AI in industrial and retail scenarios while still allowing cloud-based model management and updates.

• ✗

  A perception layer in the Azure networking stack for monitoring packet loss

  Why it's wrong here

  Monitoring packet loss and network health is the domain of Azure Network Watcher, a service that provides tools like Connection Monitor, packet capture, and topology visualization. 'Perception layer' in a networking context refers to network telemetry/observability, which is conceptually unrelated to Azure Percept's edge AI inference. Azure Percept operated as physical intelligence on devices, not as a network monitoring component within the Azure backbone, so this option confuses the platform's role with Azure's networking diagnostics services.

• ✗

  A service for perceiving user intent from mouse movements and keyboard patterns

  Why it's wrong here

  This option describes user behavior telemetry, where mouse movements, clicks, and keyboard typing patterns are analyzed to infer engagement or intent—commonly referred to as UX analytics. Azure Percept, in contrast, was purpose-built for sensing the physical environment through cameras and microphones to run vision and speech models at the edge, not for analyzing host input peripherals. The platform's sensor modules did not capture mouse or keyboard data, making this a fundamental mischaracterization of its purpose.

Go deeper

Related to this question