Polar Rose Combines Image Search and Facial Recognition

Posted on January 4, 2007

A website called PolarRose.com is planning to launch an image search website that combines people's photographs with "face recognition algorithms." This may make it easy to search for a person's name and have the search engine return a bunch of photographs containing that person. Search Engine Watch blogs that the site will also rely on the help of its users to tag names on people.

Using what Tim O'Reilly calls 'bionic software', "Polar Rose relies on a combination of our unique face recognition algorithms and the collective intelligence of our users.
The face recognition technology used was originally developed by CTO Jan Erik Solem during his M.Sc and Ph.D. stints at the universities of Lund and Malm� in southern Sweden. It's unique in that we are able to extract 3D information from regular 2D images, an approach that radically improves the short-comings of existing face recognition approaches.
However, we don't and can't rely exclusively on face recognition, but also harness the collective intelligence of our users who help train our software and tag names on people we haven't seen before" according to their website.

Obviously, there are some major privacy concerns here for people who do not want to be found or do not want all their photographs discovered. There could also be concerns if the search engine were to incorrectly label a photograph as being person X when it fact it was not person X. There are also concerns that a search engine like PolarRose.com could one day be expaned and used to identify people using images and video from various public cameras. Polar Rose also offers a blog that will keep you informed of the latest developments with the innovate image search tool.