Evolution Robotics will soon bring visual object recognition to your iPhone. The company has said that they will launch the new application in June. Using the ViPR visual object recognition technology, iPhone owners will be able to snap a photo of a book or CD cover and then mail it to the Evolution servers for image processing. If the object is properly identified, then a reply email will be issued with additional information. The demo video below shows how this technology can be used to find information about the Nemo DVD and movie soundtrack; the software allows users to purchase songs directly from iTunes. This e-commerce component is probably how the company plans to make money from their mobile phone version of their object recognition technology. More of my thoughts about this system and possible future applications after Evolution's demo video of the iPhone visual object recognition system.
Just a few months ago, we reported on a similar system for mobile phones available in Japan (by KDDI and Bandai Networks) that also utilizes Evolution's ViPR technology. In the past, the same object recognition algorithm powered the visual capabilities of SONY's AIBO robot dog. Evolution also has partnered with WowWee Robotics to bring the visual recognition software to their line of entertainment robots.
I have said in the past that Evolution's business model is in danger because of Microsoft's entry in the robot control software. The object recognition technology that Evolution utilizes in their products, however, is state-of-the-art and has withstood the test of time for nearly a decade. So far, no other recognition system can do much better than ViPR's underlining visual recognition algorithm so Evolution still has something very valuable in its procession.
So, if this visual recognition technology has been around for nearly a decade, why are we only seeing it becoming available to consumers today? There are a few reasons for that. Fist, the fact that the cameras attached to mobile devices such as phones and PDAs have become so much better makes the usage of these computer vision algorithms possible under difficult, everyday conditions. Moreover, cheap bandwidth is allowing companies such as Evolution to move complex visual processing far from the mobile device with its limited computational resources and to remote servers that can execute the computer vision algorithms in real-time and store the large amounts of image data necessary to make them work.
I expect to see more applications such as this one becoming a reality in the next few years. Soon, we will be able to snap a photo of a downtown corner in any city, send it to a remote server and then get back information about our precise location, a kind of visual GPS. Similar systems will automatically process the enormous amount of video posted online daily and automatically link the videos together in a meaningful way or even embed advertising in the videos without the need for a human editor similar to how Google's AdSense works for text-based content today.
0 comments:
Post a Comment