Developing artificial intelligence systems that can interpret images

www.physorg.com/news/2011-12-artificial-intelligence-images.html/

This article was a very interesting read. The article profiles Antonio Torralba  and his work into the field of artificial intelligence. Recently, Mr. Torralba developed systems that allowed artificial intelligence to recognize individual objects within a scene based on the context of those objects. For example, this new technology would allow artificial intelligence to recognize things like magazines on top of a coffee table or a toaster on a countertop.

            I believe this new technology will benefit us both in the short term and in the long term. In the short term, this technology could allow for more searchability of images on the Internet or other databases. The technology would recognize individual objects in the scene and then proceed to tag the image according to the bigger picture of the scene. For instance if you needed to find a picture of a cat juggling in front of a house with white shutters, you could simply type into a search engine using this technology “cat juggling house white shutters” and the results are far more likely to be accurate than without this technology.

            On the other hand, another benefit of this new technology could be in robotic applications. Being able to recognize objects in the room could allow for robots to have more situational awareness. The robot would be able to use this technology to recognize where in a house it is or to locate a specific object in a room and use it to complete a task.

            Whatever the purpose, this new technology is sure to impact us as any other new technology does; slowly and behind the scenes. It’s these emerging behind the scenes technology however that drives the digital revolution that we find ourselves in the middle of.

Eric Weiss