Azure AI Vision Image Analysis SDK for Python

Client library for Azure AI Vision 4.0 image analysis including captions, tags, objects, OCR, and more.

Installation

pip install azure-ai-vision-imageanalysis

Environment Variables

VISION_ENDPOINT=https://<resource>.cognitiveservices.azure.com  # Required for all auth methods
AZURE_TOKEN_CREDENTIALS=prod # Required only if DefaultAzureCredential is used in production
VISION_KEY=<your-api-key>  # Only required for the legacy API-key auth path below

Authentication & Lifecycle

🔑 Two rules apply to every code sample below:

1. Prefer DefaultAzureCredential. It works locally (Azure CLI / VS Code / Developer CLI) and in Azure (managed identity, workload identity) with no code change. Avoid connection strings, account/API keys — they bypass Entra audit and rotation.
   • Local dev: DefaultAzureCredential works as-is.
   • Production: set AZURE_TOKEN_CREDENTIALS=prod (or AZURE_TOKEN_CREDENTIALS=<specific_credential>) to constrain the credential chain to production-safe credentials.
2. Wrap every client in a context manager so HTTP transports, sockets, and token caches are released deterministically:
   • Sync: with <Client>(...) as client:
   • Async: async with <Client>(...) as client: and async with DefaultAzureCredential() as credential: (from azure.identity.aio)

Snippets may abbreviate this setup, but production code should always follow both rules.

import os
from azure.identity import DefaultAzureCredential, ManagedIdentityCredential
from azure.ai.vision.imageanalysis import ImageAnalysisClient
from azure.ai.vision.imageanalysis.models import VisualFeatures

# Local dev: DefaultAzureCredential. Production: set AZURE_TOKEN_CREDENTIALS=prod or AZURE_TOKEN_CREDENTIALS=<specific_credential>
credential = DefaultAzureCredential(require_envvar=True)
# Or use a specific credential directly in production:
# See https://learn.microsoft.com/python/api/overview/azure/identity-readme?view=azure-python#credential-classes
# credential = ManagedIdentityCredential()

with ImageAnalysisClient(
    endpoint=os.environ["VISION_ENDPOINT"],
    credential=credential,
) as client:
    result = client.analyze_from_url(
        image_url="https://aka.ms/azsdk/image-analysis/sample.jpg",
        visual_features=[VisualFeatures.CAPTION],
    )

Legacy: API Key (existing keyed deployments)

New code should use DefaultAzureCredential above. Use AzureKeyCredential only if you have an existing keyed deployment that hasn't been migrated to Entra ID yet — for example, regulated environments still completing their Entra rollout.

import os
from azure.core.credentials import AzureKeyCredential
from azure.ai.vision.imageanalysis import ImageAnalysisClient
from azure.ai.vision.imageanalysis.models import VisualFeatures

with ImageAnalysisClient(
    endpoint=os.environ["VISION_ENDPOINT"],
    credential=AzureKeyCredential(os.environ["VISION_KEY"]),
) as client:
    result = client.analyze_from_url(
        image_url="https://aka.ms/azsdk/image-analysis/sample.jpg",
        visual_features=[VisualFeatures.CAPTION],
    )

Analyze Image from URL

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

image_url = "https://example.com/image.jpg"

result = client.analyze_from_url(
    image_url=image_url,
    visual_features=[
        VisualFeatures.CAPTION,
        VisualFeatures.TAGS,
        VisualFeatures.OBJECTS,
        VisualFeatures.READ,
        VisualFeatures.PEOPLE,
        VisualFeatures.SMART_CROPS,
        VisualFeatures.DENSE_CAPTIONS
    ],
    gender_neutral_caption=True,
    language="en"
)

Analyze Image from File

with open("image.jpg", "rb") as f:
    image_data = f.read()

result = client.analyze(
    image_data=image_data,
    visual_features=[VisualFeatures.CAPTION, VisualFeatures.TAGS]
)

Image Caption

result = client.analyze_from_url(
    image_url=image_url,
    visual_features=[VisualFeatures.CAPTION],
    gender_neutral_caption=True
)

if result.caption:
    print(f"Caption: {result.caption.text}")
    print(f"Confidence: {result.caption.confidence:.2f}")

Dense Captions (Multiple Regions)

result = client.analyze_from_url(
    image_url=image_url,
    visual_features=[VisualFeatures.DENSE_CAPTIONS]
)

if result.dense_captions:
    for caption in result.dense_captions.list:
        print(f"Caption: {caption.text}")
        print(f"  Confidence: {caption.confidence:.2f}")
        print(f"  Bounding box: {caption.bounding_box}")

Tags

result = client.analyze_from_url(
    image_url=image_url,
    visual_features=[VisualFeatures.TAGS]
)

if result.tags:
    for tag in result.tags.list:
        print(f"Tag: {tag.name} (confidence: {tag.confidence:.2f})")

Object Detection

result = client.analyze_from_url(
    image_url=image_url,
    visual_features=[VisualFeatures.OBJECTS]
)

if result.objects:
    for obj in result.objects.list:
        print(f"Object: {obj.tags[0].name}")
        print(f"  Confidence: {obj.tags[0].confidence:.2f}")
        box = obj.bounding_box
        print(f"  Bounding box: x={box.x}, y={box.y}, w={box.width}, h={box.height}")

OCR (Text Extraction)

result = client.analyze_from_url(
    image_url=image_url,
    visual_features=[VisualFeatures.READ]
)

if result.read:
    for block in result.read.blocks:
        for line in block.lines:
            print(f"Line: {line.text}")
            print(f"  Bounding polygon: {line.bounding_polygon}")
            
            # Word-level details
            for word in line.words:
                print(f"  Word: {word.text} (confidence: {word.confidence:.2f})")

People Detection

result = client.analyze_from_url(
    image_url=image_url,
    visual_features=[VisualFeatures.PEOPLE]
)

if result.people:
    for person in result.people.list:
        print(f"Person detected:")
        print(f"  Confidence: {person.confidence:.2f}")
        box = person.bounding_box
        print(f"  Bounding box: x={box.x}, y={box.y}, w={box.width}, h={box.height}")

Smart Cropping

result = client.analyze_from_url(
    image_url=image_url,
    visual_features=[VisualFeatures.SMART_CROPS],
    smart_crops_aspect_ratios=[0.9, 1.33, 1.78]  # Portrait, 4:3, 16:9
)

if result.smart_crops:
    for crop in result.smart_crops.list:
        print(f"Aspect ratio: {crop.aspect_ratio}")
        box = crop.bounding_box
        print(f"  Crop region: x={box.x}, y={box.y}, w={box.width}, h={box.height}")

Async Client

from azure.ai.vision.imageanalysis.aio import ImageAnalysisClient
from azure.identity.aio import DefaultAzureCredential

async def analyze_image():
    async with DefaultAzureCredential() as credential:
        async with ImageAnalysisClient(
            endpoint=endpoint,
            credential=credential
        ) as client:
            result = await client.analyze_from_url(
                image_url=image_url,
                visual_features=[VisualFeatures.CAPTION]
            )
            print(result.caption.text)

Visual Features

Feature	Description
CAPTION	Single sentence describing the image
DENSE_CAPTIONS	Captions for multiple regions
TAGS	Content tags (objects, scenes, actions)
OBJECTS	Object detection with bounding boxes
READ	OCR text extraction
PEOPLE	People detection with bounding boxes
SMART_CROPS	Suggested crop regions for thumbnails

Error Handling

from azure.core.exceptions import HttpResponseError

try:
    result = client.analyze_from_url(
        image_url=image_url,
        visual_features=[VisualFeatures.CAPTION]
    )
except HttpResponseError as e:
    print(f"Status code: {e.status_code}")
    print(f"Reason: {e.reason}")
    print(f"Message: {e.error.message}")

Image Requirements

• Formats: JPEG, PNG, GIF, BMP, WEBP, ICO, TIFF, MPO
• Max size: 20 MB
• Dimensions: 50x50 to 16000x16000 pixels

Best Practices

1. Pick sync OR async and stay consistent. Do not mix azure.ai.vision.imageanalysis sync clients with azure.ai.vision.imageanalysis.aio async clients in the same call path. Choose one mode per module.
2. Always use context managers for clients and async credentials. Wrap every client in with ImageAnalysisClient(...) as client: (sync) or async with ImageAnalysisClient(...) as client: (async). For async DefaultAzureCredential from azure.identity.aio, also use async with credential: so tokens and transports are cleaned up.
3. Select only needed features to optimize latency and cost
4. Use async client for high-throughput scenarios
5. Handle HttpResponseError for invalid images or auth issues
6. Enable gender_neutral_caption for inclusive descriptions
7. Specify language for localized captions
8. Use smart_crops_aspect_ratios matching your thumbnail requirements
9. Cache results when analyzing the same image multiple times

Reference Files

File	Contents
references/capabilities.md	Additional non-hero capabilities, operation-group coverage, and production checklists.
references/non-hero-scenarios.md	Dedicated non-hero examples for secondary/advanced scenarios.