Streaming Media

Streaming Media on Facebook Streaming Media on Twitter Streaming Media on LinkedIn Streaming Media on YouTube

Amazon Expands Rekognition Image Detection Service to Video
With this cloud service, developers can build face, object, and action recognition into systems that work with stored or live video.
Learn more about the companies mentioned in this article in the Sourcebook:

At its AWS Re:Invent conference yesterday in Las Vegas, Nevada, Amazon Web Services announced Amazon Rekognition Video, an expansion of its Rekognition Image photo detection service. Rekognition Image, introduced at the 2016 conference, is a cloud service that lets developers build object and scene detection into applications. With Rekognition Video, AWS provides the same abilities to S3-stored videos and live streams.

Rekognition Video is able to track people across videos, detect activities, and identify faces and objects. Celebrity identification is built in. It identifies faces even if they're only partially in view, provides automatic tagging for locations and objects (such as beach, sun, or child), and tracks multiple people at once. The service goes beyond basic object identification, using context to provide richer information. The service is available today.

In a press release, Dan Law, chief data scientist for Motorola, suggested that Rekognition Video could be used to search historical and real-time videos for persons-of-interest, automating a task typically done by humans.

Amazon also announced AWS DeepLens, a video camera intended to give developers hands-on experience with machine learning. This HD camera can run over 100 billion deep learning operations each second. Providing examples of how DeepLens could be used, Amazon suggested developers could create systems that recognize the numbers on a license plate then automatically open a garage door, or automatically send a text to a dog owner when the dog is on the couch. DeepLens comes with sample projects and code, and Amazon says developers with no experience can run a deep-learning model in under 10 minutes. The camera is available for pre-order for $249 (on Amazon, naturally) and will be released on April 14, 2018.

On Monday, AWS announced five AWS Elemental media services intended to simplify video offering creation.

Related Articles
Spend less time creating or managing infrastructure and more time innovating, touts AWS. Its services simplify live and on-demand streaming.