AI service
Amazon Rekognition
Image and video analysis for labels, moderation, faces, celebrities, text, and custom labels.
Key points
- Provides computer vision APIs for images and videos.
- Can detect labels, objects, scenes, faces, celebrities, unsafe content, and text in images.
- Supports face comparison and face search workflows where allowed by policy and regulation.
- Custom Labels can train vision models for domain-specific image classification or object detection.
- Works on visual media rather than document form structure or plain text NLP.
When to use it
- Choose Rekognition for image moderation, object detection, face analysis, or video labeling.
- Use Custom Labels when a business needs to detect its own product or defect classes in images.
- Use it for media libraries that need visual search or automated tagging.
Exam tips
- Rekognition analyzes images and video; Textract extracts text, forms, and tables from documents.
- Lookout for Vision focuses on industrial defect detection, while Rekognition is the broader vision API.
- Comprehend is for natural language text, not image labels.
- Face capabilities require careful privacy, consent, and governance controls.