Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Amazon Rekognition Image Analysis Overview

Amazon Rekognition Image Analysis Overview

Learn about the newly announced Amazon Rekognition with a hands-on session by Alex Casalboni. Find out how deep learning is powering Amazon's Image Analysis service and which use cases are supported, with respect to other frameworks and APIs.

Alex Casalboni

December 22, 2016
Tweet

More Decks by Alex Casalboni

Other Decks in Technology

Transcript

  1. About  Me @alex_casalboni clda.co/rekogni.on Bachelor  in  Computer  Science Master  in

     Sound  &  Music  Engineering Web  Developer  (6+  years) Cloud  Evangelist  @  Cloud  Academy
  2. Image  Analysis  Context clda.co/rekogni.on Object  Detec?on Extrac.on  of  labels Tagging

     w/  Confidence Facial  Analysis Face  detec.on Face  comparison Face  search
  3. What  is  NOT  supported  (yet)? clda.co/rekogni.on Op.cal  Character  Recogni.on  (OCR)

    Explicit  content  (nudity,  violence,  etc.) Logo  detec.on  (popular  products) Image  rendering  (landmarks,  boxes,  etc.)
  4. Facial  Analysis clda.co/rekogni.on Face  Detec?on != Face  Recogni?on Image  scan

    Loca.on  &  features Confidence  ~  Detec.on Numerical  representa.on Vectors  comparison Confidence  ~  Similarity
  5. Face  Indexing  and  Collec?ons clda.co/rekogni.on User B Face  Collec.on Search

      by  Face dc090f86-48a4-5f09-905f-44e97fb1d455 bh020ftl-12a3-ho09-1355-21blpfb42acd 011bdsa1-op5j-gf14-bhg2-1p7agds6mca9 ………………..………………..………………..…………… ………………..………………..………………..…………… ………………..………………..………………..…………… ………………..………………..………………..…………… ………………..………………..………………..…………… ………………..………………..………………..…………… ………………..………………..………………..…………… ………………..………………..………………..…………… ………………..………………..………………..…………… ………………..………………..………………..…………… ………………..………………..………………..…………… User A Index  new   Images
  6. S3  Events  &  Rekogni?on  Architecture clda.co/rekogni.on User S3 Bucket Lambda

    DynamoDB CloudSearch Rekognition or Upload Trigger Index Put  Item Commit
  7. Thank  you!                

    12/22/2016 clda.co/rekogni.on Q  &  A