Ce diaporama a bien été signalé.
Nous utilisons votre profil LinkedIn et vos données d’activité pour vous proposer des publicités personnalisées et pertinentes. Vous pouvez changer vos préférences de publicités à tout moment.

NEW LAUNCH! Introducing Amazon Kinesis Video Streams - ABD216 - re:Invent 2017

3 618 vues

Publié le

Amazon Kinesis Video Streams makes it easy to securely stream video from connected devices to AWS for analytics, machine learning (ML), and other processing. In this session, we introduce Kinesis Video Streams and its key features, and review common use cases including smart home, smart city, industrial automation, and computer vision. We also discuss how you can use the Kinesis Video Streams parser library to work with the output of video streams to power popular deep learning frameworks. Lastly, Abeja, a leading Japanese artificial intelligence (AI) solutions provider, talks about how they built a deep-learning system for the retail industry using Kinesis Video Streams to deliver better shopping experience.

  • Soyez le premier à commenter

NEW LAUNCH! Introducing Amazon Kinesis Video Streams - ABD216 - re:Invent 2017

  1. 1. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. AWS re:INVENT Introducing Amazon Kinesis Video Streams R o g e r B a r g a , G e n e r a l M a n a g e r o f A m a z o n K i n e s i s , A W S A d i K r i s h n a n , H e a d o f A m a z o n K i n e s i s V i d e o S t r e a m s , A W S Y o u s u k e O k a d a , F o u n d e r & C E O , A B E J A , I n c . T o s h i y a K a w a s a k i , P l a t f o r m E n g i n e e r i n g H e a d , A B E J A I n c . A B D 2 1 6
  2. 2. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. What to expect from the session • Overview of Amazon Kinesis Video Streams • Key use cases • Product details • Customer case study and demo: ABEJA Inc.
  3. 3. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Kinesis – real-time analytics Easily collect, process, and analyze real-time, streaming data Load data streams into AWS data stores Analyze data streams with SQL Build custom applications that analyze data streams Kinesis Data Streams Kinesis Data Firehose Kinesis Data Analytics SQL
  4. 4. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Video is critical to many applications Smart Home Security Monitoring Smart City Industrial Automation Computer Vision
  5. 5. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Building a video ingestion system is complex • Cumbersome to rewrite code for various devices and development environments • Difficult to elastically scale to millions of devices • Requires support for cadence, latency, and jitter on streams • Requires secure, highly scalable, available storage • Need easy-to-use APIs to retrieve, process, and replay video
  6. 6. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Stream video from millions of devices Easily build vision-enabled apps Secure Durable, searchable storage Fully managed Amazon Kinesis Video Streams S t r e a m v i d e o a n d t i m e - e n c o d e d d a t a f o r a n a l y t i c s Kinesis Video Streams
  7. 7. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Key Use Cases
  8. 8. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Use case: Smart Home Ex amp le : Pe t Monitor
  9. 9. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Use case: Smart City Ex amp le : Amb e r Al e rt S yste m
  10. 10. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Use case: Industrial Automation Ex amp le : Equ ip me nt Pre ve ntive Mainte nance
  11. 11. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Kinesis Video Streams Product Details
  12. 12. Amazon Kinesis Video Streams Concepts Amazon Confidential
  13. 13. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Confidential • Flexible SDK for integration with on- device hardware media pipelines • Out-of-the-box integration with AWS Auth mechanism: SigV4 • Flexible Put API to stream Frames or buffered Fragments or after-the-fact uploads • Build custom integrations with diverse camera and device types • Secure AuthN and AuthZ to connect devices w/ image sensors to AWS • Stream video data based on the preferred transmission scenario Video Streams SDK offers … … that enables developers to Kinesis Video Streams Producer SDK E a s i l y c o n n e c t a n d s t r e a m f r o m c a m e r a s o u r c e s
  14. 14. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Kinesis Video Streams Producer SDK E a s i l y c o n n e c t a n d s t r e a m f r o m c a m e r a s o u r c e s Amazon Confidential Platform independent layer (C) OO Wrapper Layer (C++ and Java) Android Linux For app devs to install on target OS to build applications. Does not satisfy all underlying hardware media pipeline, source listing but makes it easy to do so. For camera-level integrators who want an object-oriented, high-performance integration model that satisfies the service calls but is flexible to customize. For manufacturers who want full flexibility to integrate at source and firmware level for different hardware platform types.
  15. 15. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. • CreateStream: Creates a new video stream • DeleteStream: Deletes the specified stream • DescribeStream: Returns the most current information: version, status, data retention, etc. about the specified stream • ListStreams: Returns an array of StreamInfo objects. Each object describes a stream • GetDataEndpoint: Gets an endpoint for a specified stream for either reading or writing Control Plane APIs (5 of 10)
  16. 16. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. • UpdateDataRetention: Increases or decreases the stream's data retention period • UpdateStream: Updates stream metadata, such as the device name and media type • TagStream: Adds one or more tags (key-value pair) to a stream to help you define and categorize • ListTagsForStream: Returns a list of tags associated with the specified stream • UntagStream: Removes one or more tags from a stream Control Plane APIs (10 of 10)
  17. 17. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. • PutMedia: long-running streaming API to write media data to a video stream • GetMedia: retrieve media content from a video stream from the specified starting chunk continuously with low latency • GetMediaForFragmentList: retrieve media data for a list of fragments from the stored data in a video stream • ListFragments: returns a list of Fragments from the specified video stream and start location Data Plane APIs (4 of 4)
  18. 18. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Kinesis Video Streams Parser Library S o f t w a r e t h a t m a k e s i t e a s y t o w o r k w i t h G e t M e d i a o u t p u t • Get a frame-level object and its associated metadata • Extract and collect video streams-specific metadata attached for fragments • Merge consecutive fragments • Build into your custom ML or other video-processing applications
  19. 19. Kinesis Video Streams Pricing P a y f o r v o l u m e o f d a t a i n g e s t , s t o r e d , a n d c o n s u m e d Data Ingested (per GB) Data Consumed (per GB) Data Stored (per GB-Month) US EAST (N. Virginia) $ 0.00850 $ 0.00850 $ 0.0230 US WEST (Oregon) $ 0.00850 $ 0.00850 $ 0.0230 EU (Ireland) $ 0.00944 $ 0.00944 $ 0.0230 EU (Frankfurt) $ 0.01020 $ 0.01020 $ 0.0245 Asia Pacific (Tokyo) $ 0.01097 $ 0.01097 $ 0.0250
  20. 20. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Ecosystem
  21. 21. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Agent Vi: M a ke a n y s u r v e il l ance c a me ra s m a rt
  22. 22. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Health Deep Learnin g GPU Events Analytic s Search Admin Portal BW Optimai zer API Gatewa y Cache S3 RDS Agent Vi and Kinesis Video Streams Integration Agent Relay Kinesis Produce r Video Clip innoVi Micro Services Kinesis Gateway Clip Player Consumer SDK Producer SDK Kinesis Video Stream
  23. 23. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Veritone: B r i n g A I t o v i d e o a n a l yt ics
  24. 24. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved.
  25. 25. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Customer Profile: ABEJA
  26. 26. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. What is ABEJA Company ABEJA, Inc. Establishment September 2012 Capital 1,148,518,043 Yen HQ Tokyo, Japan & Singapore Mission Industrial Structure Transformation using Deep Learning Employee 70 People (70% Engineers) Business PaaS: ABEJA Platform Retail SaaS: ABEJA Platform for Retail Stock Holder NVIDIA, Salesforce, INCJ, NTT, Itochu, Toshiba-Tec and etc.
  27. 27. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. ABEJA Platform (IoT & AI Platform) Overview Cloud Inference Edge Inference Distributed Cloud Training Data Lake IoT Sensors IoT Actuators Other ServicesOther Services Data Analytics API Connect Stream & Batch Annotation to make Supervised Data A Data for Inference Model Deploy Model Deploy Real-time Stream Control API Connect Supervised Data Control Inferenced Data
  28. 28. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. DynamoDB + Tracking Mobile Devices Cameras (1/3) Kinesis Video Stream in ABEJA Platform Manufacturing Use Case VMD Optimizations Preventive Maintenance Visual inspection Retail Use Case Automatic Picking Measure the effect of Marketing
  29. 29. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. DynamoDB Kinesis Data Streams IoT Sensors + Tracking Mobile Devices Cameras (2/3) Kinesis Video Stream in ABEJA Platform Manufacturing Use Case VMD Optimizations Preventive Maintenance Visual inspection Retail Use Case Automatic Picking Measure the effect of Marketing
  30. 30. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. DynamoDB Kinesis Data Streams IoT Sensors + Tracking Mobile Devices Cameras On-premises GPU cluster/servers for training (3/3) Kinesis Video Stream in ABEJA Platform Manufacturing Use Case VMD Optimizations Preventive Maintenance Visual inspection Retail Use Case Automatic Picking Measure the effect of Marketing
  31. 31. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Demo
  32. 32. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Webcam MacOS streamer using Producer SDK in C++ Consumer fetches and decodes frames using parser library & applies deep-learning based model GStreamer based producer using C++ SDK on Linux sends processed stream to console Demo Architecture How does it work? 1 2 3 Kinesis Video web console
  33. 33. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. ABEJA’s impressions of Kinesis Video Streams • Do not have to worry about any infrastructure deployment or management for reliable streaming video ingestion and storage • Producer SDK enables streaming video to the cloud • Parser library enables access and processing of real-time frames • We focus on our core competency of building deep learning algorithms and technology
  34. 34. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Amazon Kinesis—real-time analytics Easily collect, process, and analyze video and data streams in real time Capture, process, and store video streams for analytics Load data streams into AWS data stores Analyze data streams with SQL Build custom applications that analyze data streams Kinesis Video Streams Kinesis Data Streams Kinesis Data Firehose Kinesis Data Analytics SQL New at re:Invent 2017 Amazon Confidential
  35. 35. © 2017, Amazon Web Services, Inc. or its Affiliates. All rights reserved. THANK YOU! V i s i t a w s . a m a z o n . c o m / k i n e s i s / v i d e o - s t r e a m s

×