Analyze Images with Azure AI Vision SDK for Java

This skill builds image analysis applications on Azure AI Vision using the Java SDK (azure-ai-vision-imageanalysis), authenticated via an API key with KeyCredential or via DefaultAzureCredential, using either a synchronous ImageAnalysisClient or an async ImageAnalysisAsyncClient. It covers seven visual features: CAPTION for a human-readable image description, optionally gender-neutral; DENSE_CAPTIONS for region-level captions across up to 10 regions with bounding boxes; READ for OCR that extracts text as blocks, lines, and words with confidence scores and bounding polygons; TAGS for content tags covering objects, scenes, and actions with confidence; OBJECTS for bounding-box object detection with tag and confidence; SMART_CROPS for smart thumbnail regions at requested aspect ratios; and PEOPLE for person detection with bounding box and confidence. Analysis runs against a local file via BinaryData.fromFile or a remote URL via analyzeFromUrl, and multiple features can be requested in a single call - for example CAPTION, TAGS, OBJECTS, and READ together with a language option. Async analysis uses reactive subscribe, error, and complete callbacks, and errors surface as HttpResponseException with a status code and message.

Use it when implementing image captioning, OCR, object or people detection, tagging, or smart cropping in a Java application via Azure AI Vision, triggered by phrases like "image analysis Java," "Azure Vision SDK," or "OCR image text extraction." Two hard constraints apply: images must be JPEG, PNG, GIF, BMP, WEBP, ICO, TIFF, or MPO, under 20MB, and between 50x50 and 16000x16000 pixels; and Caption and Dense Captions specifically require GPU-supported regions, so check Azure's supported-region list before deployment. Do not treat its output as a substitute for environment-specific validation or expert review.

Input is an image - a local file as BinaryData or a public URL - plus a list of requested VisualFeatures and optional ImageAnalysisOptions such as gender-neutral captions, smart-crop aspect ratios, or a language setting. Output is an ImageAnalysisResult exposing the requested feature results - caption text and confidence, dense captions with regions, OCR blocks/lines/words, tags, objects, people, or crop regions - or an HttpResponseException with a status code on failure.

<dependency>
    <groupId>com.azure</groupId>
    <artifactId>azure-ai-vision-imageanalysis</artifactId>
    <version>1.1.0-beta.1</version>
</dependency>

It requires a Vision resource endpoint and key, or DefaultAzureCredential for Entra ID authentication, configured through the VISION_ENDPOINT and VISION_KEY environment variables.

Java developers building image-understanding features - content moderation, accessibility captioning, document OCR, product or object detection, or automated thumbnail generation - who need production Azure Vision integration rather than a single-feature prototype.