5.1.3. Object and People Detection

Detect and locate objects and people within images. Returns bounding box coordinates for each detection.

💡 First Principle: Humans glance at an image and instantly know "that's a dog in a park." Computer vision replicates this by converting pixels into semantic labels. But different questions require different analyses: "What's in this image?" → tags/objects. "Describe this image" → captions. "What text is visible?" → OCR. "Where is the dog?" → object detection with bounding boxes. The exam tests whether you can match the question type to the correct visual feature parameter.

🔧 Implementation Reference: Azure AI Vision

Visual features determine what the API extracts from an image. Select the features relevant to your scenario.

Visual Features:

The following diagram helps you select the right visual features for your scenario:

Testable Pattern:

from azure.ai.vision.imageanalysis import ImageAnalysisClient
from azure.ai.vision.imageanalysis.models import VisualFeatures

client = ImageAnalysisClient(endpoint=endpoint, credential=AzureKeyCredential(key))
result = client.analyze(
    image_data=image_bytes,
    visual_features=[VisualFeatures.CAPTION, VisualFeatures.TAGS, VisualFeatures.OBJECTS, VisualFeatures.READ]
)
caption = result.caption.text
tags = [tag.name for tag in result.tags.values]

Error Handling Pattern:

from azure.core.exceptions import HttpResponseError

try:
    result = client.analyze(image_data=image_bytes, visual_features=[VisualFeatures.READ])
except HttpResponseError as e:
    if e.status_code == 400:
        # Invalid image format or size
        logging.error("Invalid image")
    elif e.status_code == 415:
        # Unsupported media type
        logging.error("Unsupported format")

CLI Equivalent (REST):

curl -X POST "https://{endpoint}/computervision/imageanalysis:analyze?features=caption,tags&api-version=2024-02-01" \
  -H "Ocp-Apim-Subscription-Key: {key}" \
  -H "Content-Type: application/octet-stream" \
  --data-binary @image.jpg

⚠️ Exam Trap: OCR (read feature) is Computer Vision, not Language—input is image pixels.

Azure AI Vision Documentation