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.

• ✗

  Training models on a compact (small) dataset with fewer than 50 images

  Why it's wrong here

  The word 'compact' in compact domains refers to the resulting model's size and computational footprint, not to the number of training images. Custom Vision does have dataset minimums (typically a handful of images per tag), but 50 is an arbitrary threshold and far below typical recommendations for good accuracy. A compact domain can be trained on dozens or hundreds of images, just like any other domain; the model-file size is what is minimized through architecture choices like depthwise separable convolutions.

• ✓

  Producing exportable models optimized for deployment on edge devices with limited compute

  Why this is correct

  Compact domains are a family of Custom Vision model architectures specifically designed for export to lightweight inference formats such as ONNX, TensorFlow Lite, and CoreML. These models are significantly smaller and require fewer compute resources, allowing them to run offline directly on edge devices like cameras, drones, or IoT hardware with limited memory and CPU. This is the intended purpose of compact domains: trade a bit of accuracy for a deployment-ready model that performs local, low-latency inference without a cloud round-trip.

• ✗

  Creating more compact API responses with less metadata

  Why it's wrong here

  Compact domains are designed for edge export, not for changing the REST API response shape of cloud-hosted Custom Vision endpoints. API response metadata is determined by the prediction endpoint's schema, not by the model's internal architecture. In fact, compact domains enable on-device inference, often eliminating API calls altogether rather than simply trimming their payloads.

• ✗

  Training models that use less storage in Azure blob containers

  Why it's wrong here

  Compact domains in Custom Vision produce small, exportable model files, but reducing Azure Blob storage consumption during training is not their purpose. Training images and checkpoints stored in blob containers are data-management artifacts; compactness applies to the inference model's architecture and export format, not to where training data is held. Even if a compact model uses less memory at runtime, blob storage costs are unrelated to domain selection.

Go deeper

Related to this question