Amazon Rekognition
Amazon Rekognition is a machine learning service from Amazon Web Services that automatically detects objects appearing in images, making it possible to organize, characterize, and search large collections of images.
AI Practitioner focus
- Choose Rekognition for managed image/video computer vision: labels, objects/scenes, faces, text, content moderation, celebrities, PPE, and related analyses.
- Image operations are commonly synchronous while stored-video analysis is asynchronous; use Amazon Textract when document structure, forms, tables, or handwriting are the real requirement.
Key points
- Provides automatic detection of objects appearing in pictures.
- Used to organize, characterize, and search large numbers of images.
- Also detects scenes and faces, extracts text, recognizes celebrities, and flags inappropriate content in images and video.
- Stores detected faces in server-side collections, letting an application search for and compare known faces across images and streaming video — e.g. facial identification for user verification.
- Sits alongside Amazon Polly, Amazon Comprehend, and Amazon Transcribe in the AWS Machine Learning services family.