Amazon’s new smart video camera helps developers explore machine learning

Some 43,000 attendees are in Las Vegas this week for Re:Invent, Amazon’s annual conference about its ever-expanding range of Amazon Web Services offerings. Most of the news during AWS CEO Andy Jassy’s keynote has been about databases, containers, and other technical underpinnings of net-based services. But he also unveiled a piece of hardware: DeepLens, a video camera with built-in smarts relating to the AI field known as machine learning.

DeepLens isn’t designed for companies to deploy in the field. Instead, Amazon means for engineers to experiment with it as they dabble with machine learning, which allows computers to perform tasks such as identifying items in photographs or translating languages. It works in conjunction with a new AWS service, SageMaker, which is aimed at developers who may not have the level of expertise required to work with low-level machine-learning software such as TensorFlow.

Matt Wood, AWS’s general manager of AI, demoed DeepLens onstage by holding up vinyl albums and reacting to them–smiling as he brandished Pink Floyd’s Dark Side of the Moon and frowning for a Rick Astley LP. The camera and SageMaker figured out the records in question from their cover art and detected Wood’s expression, then used that data to suggest other music he might like, such as a Fleetwood Mac album.

DeepLens will cost $249 and is scheduled to ship next April. Amazon is currently taking preorders.

AWS DeepLens