Amazon Rekognition
AWS cloud APIs for image and video: face detection, face search (user-maintained collections), celebrity recognition, and related computer-vision tasks—usable by any developer with an account.
Background
Rekognition is sold as metered cloud infrastructure, not a single “police product.” Customers upload images or stream video and call APIs for labels, faces, text, and moderation. Face search lets an organization index faces into a private collection and query new images against it. Amazon has changed some sales policies for law enforcement face capabilities over time; capabilities and terms remain productized and widely available globally.
Harms and concerns
Lowering the cost and skill barrier to face recognition increases deployment in schools, stores, and public space. Studies and audits have highlighted demographic performance differences in commercial APIs. Cloud processing can centralize sensitive imagery with a major tech vendor. Customers may lack strong governance—collections can be built from CCTV without subjects’ knowledge.
Capabilities (overview)
- Face detection & comparison – locate faces and score similarity between images
- Face search – index and search private face collections
- Video & streaming – analyze stored or live video for faces and other objects
- Other vision APIs – labels, text-in-image, moderation, custom models (separate paths)