Google Visual Rank – picture search to get smarter

google image search for "ape"
Searching for images on the web is still a hit and miss affair.

Efforts from many organisations to make the task of finding images and video on the web easier are really only just getting underway.

Currently all mainstream search providers rely on text descriptions of the image known as ‘ meta data ‘ or ‘ tagging ‘ to return results to our queries. Adding these descriptions to the images in most cases is a human task,time consuming, prone to error / interpretation and can very laborious marketing technology.

Google last year added ‘ face ‘ recognition to its search algorithms – however exactly whose face it was still relied on good old fashioned tagging ( see above ).

Taking this one stage further they recently published documents outlining trials of “ Visual Rank “ which combines image recognition technology with their ranking algorithms, ranking and returning pictures based on similarities found across a set of images.

Initially focusing its efforts on product images the search giant is expanding the trials to take in the travel related queries.

From a marketing perspective it adds a new dimension, now not only do advertisers have to grapple with ensuring keywords and phrases are associated to their products, the product imagery itself must be optimised.

Official Google Research Blog: VisualRank: "The idea was simple: find common visual themes in a set of images, and then find a small set of images that best represented those themes"

No comments: