Latest in Japan

Image credit:

SOINN gets intelligence boost, uses web-based image search to ID objects

Zach Honig
05.01.13
6 Shares
Share
Tweet
Share

Sponsored Links

Tokyo Institute of Technology's SOINN (Self-Organizing Incremental Neural Network) bot was impressive enough during our first introduction back in 2011, but the intelligent device can now tap the web for its latest trick: accurate object identification. The updated system, which appears to utilize a database akin to Google's image search, can scour the web for similar shots, making it possible to ID objects based on comparable structures published on the web. It can distinguish a box cutter from a knife, for example, or a rickshaw from a car. For now, SOINN is limited to identifying objects in images, including those captured in realtime with a camera, but its designers imagine that future revisions could enable content recognition in video streams, and audio clips, too. Our friends at DigInfo saw the update in action -- check it out for yourself in the video after the break.

All products recommended by Engadget are selected by our editorial team, independent of our parent company. Some of our stories include affiliate links. If you buy something through one of these links, we may earn an affiliate commission.
Comment
Comments
Share
6 Shares
Share
Tweet
Share

Popular on Engadget

Snapchat's latest AR filters turn the floor into lava

Snapchat's latest AR filters turn the floor into lava

View
Reuters: UK Google users will lose GDPR protections

Reuters: UK Google users will lose GDPR protections

View
HTC's expanded Vive Cosmos family may help lure more VR users

HTC's expanded Vive Cosmos family may help lure more VR users

View
If you bought Flywheel's home bike, you can trade it in for a free Peloton

If you bought Flywheel's home bike, you can trade it in for a free Peloton

View
Hasbro is relaunching classic Tiger Electronics gaming handhelds

Hasbro is relaunching classic Tiger Electronics gaming handhelds

View

From around the web

Page 1Page 1ear iconeye iconFill 23text filevr