Entering metadata for each and every picture in our burgeoning collection of digital photos could be a royal pain. Imagine typing in all those texts for every single photo.
Systems content developer aceMedia is trying to alleviate that with their technology to automatically recognize and tag multimedia data, including photos and video. The system scans images for pixel clusters in still and moving images, recognizes the images and then automatically annotates the image with a keyword.
Itâ€™s all based on an algorithm known as DL reasoning. For example, 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. Compare this with Penn State Uâ€™s ALIPR method that we reported earlier.
aceMedia will do a presentation to explain their prototype keyword tagging method at Europeâ€™s Information Society Technologies (IST) 2006 Conference in Helsinki at the end of the month.