Polar Rose, a Swedish startup which aims to revolutionise image search on the internet, with its facial recognition technology, has announced more details of its plans.

I mentioned Polar Rose last month after they had announced that they had raised $5.1 million in VC funding. Now they have announced more details of their image search project.

Currently, web image searches can only search the text which accompanies the picture, so your search is only as accurate as the tags which have been applied to the picture.

Polar Rose aims to solve the problem by improving tagging, but, more importantly, by analysing the images themselves, and automatically recognising the subject.

The company’s technology turns a 2D image, as displayed on a computer screen, into a 3D image which, after compensating for light and pose variations, can be used to search for related photos.

Polar Rose’s technology will be offered as a free web browser plug-in and through royalty-free APIs for partners, including photo sharing sites. A contextual advertising program is also planned for 2007.

The technology has a range of applications – you could use it simply to help organise your photo files, or to search for pictures of a particular person.

When used in conjunction with a news site, for instance, Polar Rose’s technology could be used to find background information on anyone photographed in the article. Likewise, the technology has applications for online dating services.

As the technology is used over time, Polar Rose hopes to harness ‘collective social intelligence’ in order to improve the tagging of photos. Ah, the wisdom of crowds. There are also future plans to apply the technology to video.

Polar Rose will launch its beta trial in Q1 2007. You can see a video sample here.