Google talks visual search

Google is working on visual search through "computer vision". Although at this stage the research is limited, it could hint towards the future of online search using images as input media, rather than text.

Revealed as a Google team is presenting a paper on landmark recognition at the Computer Vision and Pattern Recognition conference in Miami, Florida, the new technology enables computers to quickly and efficiently identify images of more than 50,000 landmarks from all over the world with 80% accuracy.

Google developed the technology using existing images of landmarks from 40 million geotagged pics on Picasa and Panoramio, then via image matching and clustering techniques developed a "highly efficient indexing system for fast image recognition".

Stating that this is a research paper, and "not a new Google product", Google says "but we still think it's cool".


>