aceMedia Develops Image Content Recognition

November 17, 2006 – Systems content developer aceMedia is developing an universal method to automatically recognize and tag multimedia data, including photos and video.

As users increasingly upload more and more images to their desktops and online sharing sites, photographers are faced with the time-consuming task of manually entering metadata.  In addition, the metadata keywords are sometimes not recognized across all software programs.

aceMedia is looking to solve that problem for users of cameras, camcorders, and mobile phones by creating a single, harmonized system.  Still in its initial prototype stage, the system scans images for pixel clusters in still and moving images, according to a BBC News article published today.The technology recognizes the images and can then automatically annotate the image with a keyword. 

The system is based on description logic inference, otherwise known as DL reasoning.For instance, if an image showed a blue area next to a brown area, the technology might interpret the blue as sky and the brown as sand, with some level of accuracy, according to an aceMedia report. 

Penn State University researchers launched ALIPR (Automatic Linguistic Indexing of Pictures – real time) to make users’ images more visible on the Internet.  ALIPR, which is based on a statistical algorithm, automatically adds contextual text to color photos, with limits on art work and cartoons. 

aceMedia’s technology, still in its beginning stages, is currently limited to detecting and tagging certain scenes including tennis and beach images.aceMedia plans on applying the auto system not only to home photos, but news images, according to the BBC. 

"People don’t want to have to spend time managing their content manually, they just want to be able to view it whenever and however they want," said Dr. Paolo Hobson of aceMedia to the BBC.  "For that to happen, multimedia content needs to become intelligent."

Helsinki at the end of the month.