ALIPR: Intelligent, Automated Tagging Coming to Flickr

Tagging can be tedious. People don’t like doing tedious task, wouldn’t you agree? It’s nice to know that you can search more quickly and broadly, but someone or something has to put the metadata in place for enhanced search options to be of any use, right?

Recently, two enterprising individuals launched a site that addresses the issue, with the recent introduction of ALIPR (Automatic Linguistic Indexing of Pictures), software which shows to be incredibly adept at sifting through thousands of images with astounding accuracy: 98% astounding.

The time it takes us uploaders to do the tagging is time wasted, say Jia Li, a mathematician at Pennsylvania State University, and colleague James Wang. They’ve have created and developed ALIPR to address that issue, so people could upload away and not worry about anything else.

ALIPR’s creators claim that the software analyses images pixel by pixel, focusing on the distribution of colors and texture while at the same time comparing the subject to an existing database of tagged files.

Li and Wang demonstrated ALIPR on images hosted by Flickr their software had not previously encountered. Results showed 51% of first words appropriated by ALIPR were duplicates of terms that already existed with the respective 5,411 photos beforehand. As for its overall accuracy, 98% isn’t bad at all.

ALIPR will launch the act of sharing and searching for photos to new heights, but it won’t stop there. ALIPR is a leap over a barrier that has proven to be quite the obstruction to productivity. Now, places like Flickr will be friendlier than ever, and I wouldn’t be surprised if ALIPR is adapted to help surmount roadblocks elsewhere in the future.