Most search engines for images, i. e. Google, are in fact text based and require images being either named after depicted persons or marked with tags. The search engine Polar Rose uses computer vision technology for face recognition and image search using patented methods for extracting 3D data from 2D pictures.

“The first step is to locate faces. We have developed a fast method that uses a number of filters for searching. Results are then interpreted to establish whether there are faces present, or not,” says Jan Erik Solem, founder of Polar Rose, to Ny Teknik. 

“When a face is found, the corresponding 3D-form is computed. And based on that, images are matched,” continues Solem.

Polar Rose’s technology was originally intended for security applications. The company has now found a new market in categorizing and sorting images and making them searchable without having to be marked. 

The company is based in the Øresund region at Malmö Incubator (MINC). A recent addition to the company is CEO Nikolaj Nyholm, a Dane with a track record from IT start-ups such as Speednames.

More information at: Polar Rose