Google’s AI ‘Clips’ camera goes on sale in US
Image credit: Dreamstime
A machine learning-enabled camera designed by Google to take photographs and short videos of interesting activity on its own has gone on sale in the US.
The camera was available to US customers only for $249 (£176). Due to high demand, Google stopped taking orders for the camera within hours and is now redirecting customers to a waitlist.
Delivery times have been stretched out as far as early March.
The camera was first unveiled in October 2017, alongside Google’s Pixel 2 smartphone, and is targeted at parents who enjoy recording their children’s lives extensively through photography and video. The camera does not have a screen, so the captured photographs and videos must be shared with a phone via an app for viewing.
The clip-on camera - which is no larger than a standard teabag - is intended to be placed in a room and left to its own devices. As it tracks activity in the room, it can learn to recognise members of the household, pets and certain actions using machine-learning software run on an Intel Movidius chip. It will capture spontaneous photographs and seconds-long videos when it spots something interesting or unusual happening.
This removes the distraction of taking out a camera or smartphone and posing the subjects of the photograph.
According to a blog post, Google hired a documentary maker, a photojournalist and a fine arts photographer to help provide visual data with which to train the machine-learning software how to recognise “memorable moments” worthy of capturing.
In response to some concerns by privacy campaigners uncomfortable with the idea of a camera watching their every move within a certain room, Google has given reassurance that the device was built with user privacy as a priority. A light appears during recording and all processes are carried out on its own hardware, with no personal information stored on external servers.
The household device is similar in some ways to Kuri, a “companion” robot designed by Mayfield Robotics. The self-driving robot records video footage within its household, and uses machine learning to recognise interesting interactions and other activity.
Meanwhile, Amazon Web Services, Amazon’s business arm, is to launch a similar small AI-equipped video camera in April this year, reportedly for the same price as Clips. The camera - named DeepLens - is intended for business rather than personal use and is aimed at developers exploring vision-based deep learning.