AWS Rekognition

Computer vision service that adds image and video analysis to applications through pretrained APIs. Detects faces, objects, text, and custom labels in images and videos without requiring machine learning expertise or infrastructure.

Used across media, retail, and identity verification. Differentiates from alternatives like Google Cloud Vision and Azure Computer Vision through face liveness for anti-spoofing, custom label training with minimal images, and native integration with other AWS services.

Key features:

• Face detection, comparison, and search with liveness checks
• Custom labels trained with AutoML using as few as 10 images
• Content moderation for unsafe or inappropriate content
• Text extraction from images and videos
• Video segment detection for black frames, credits, and shot boundaries

Developers integrate via REST APIs or SDKs for use cases such as identity verification during onboarding, content moderation for user-generated media, and smart home automation that triggers actions when specific objects appear in live streams. Video producers use segment detection for ad insertion and content operations.