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.

• ✗

  Captioning adds user-written descriptions; tagging uses AI to detect objects automatically

  Why it's wrong here

  The statement incorrectly assumes that captions are human-authored and tags are machine-generated. In Azure AI Vision, both captioning and tagging are fully automated outputs of the same image analysis model family; no user-written text is added during inference. A user can optionally provide an external description to influence some custom models, but the standard captioning API generates its own sentence using an encoder-decoder neural network, just as tagging generates its own labels with a classifier. Therefore, the real difference is output type—sentence versus keyword list—not the source of the content.

• ✓

  Captioning generates a natural language sentence describing the scene; tagging returns individual concept keywords

  Why this is correct

  Captioning in Azure AI Vision uses an image-to-text model to produce a grammatically complete sentence that describes the whole scene, such as “A cat sitting on a sofa.” Tagging instead runs a multi-label classification model that returns a discrete list of concept keywords (for example, cat, sofa, indoor) with confidence scores. The sentence gives narrative context for accessibility or storytelling, whereas the keyword list is optimized for indexing, search, and automated filtering. This difference in output format—one generated sentence versus many discrete labels—is the core definitional distinction.

• ✗

  Captioning works on video; tagging works only on still images

  Why it's wrong here

  This claim confuses the supported input media with the output modality. Azure AI Vision’s image captioning and image tagging features both operate on still images; neither the Image Analysis API function is video-only, and tagging is not restricted to still frames. While video understanding tools can extract frames and run these same models on each frame, the API itself does not require or distinguish media type this way. The actual difference between the features remains that captioning emits a sentence and tagging emits keyword labels.

• ✗

  Image tagging is more accurate than captioning because it uses simpler classification

  Why it's wrong here

  Accuracy cannot be compared as a single metric because captioning and tagging are fundamentally different machine learning tasks with different evaluation criteria. Tagging is a multi-label recognition problem scored by how confidently known classes are assigned, while captioning is a sequence generation problem that must synthesize grammar, context, and object relationships to create a novel sentence. A simpler classification pipeline can make tags precise for vocabulary items, but it cannot express scene semantics, and a captioning model can be more expressive yet still make grammatical errors or hallucinate details. Neither is strictly more accurate; they fail in different ways and serve different downstream purposes.

Go deeper

Related to this question