LAST week, Plink, an early-stage visual-search company behind the PlinkArt mobile application, which allows people to identify a work of art by taking a photo of it, was sold to Google, with its two Irish founders - James Philbin and Mark Cummins - now heading to work at Google Goggles.
For such a young outfit - Plink was set up in 2009 - the Google acquisition has sent ripples through the visual search community, with many sensing that Google is upping the pace by moving even further beyond simple text-based search on the mobile platform.
But what is visual search and what are the possibilities, particularly for the end user? For the online consumer, we have websites such as Riya.com where you can take a photo of a dress that has caught your eye in a magazine, for instance, and it will track down the designer or product for you. Quintura.com is another popular visual search engine.
However, according to Google, computer-vision technology is still in its infancy.
"Visual search means initiating a search using an image. People consume and process information visually, so there's no reason why computers shouldn't also do that," says a Google spokesperson.
"Google Goggles is a Labs product but our ultimate goal is to be able to recognise and provide relevant results for any image."
Explains Philbin of Plink: "Currently, people can search for specific categories of objects - book covers, wine labels, art. Eventually, I think the technology will be good enough to search for information on almost any object, acting as a kind of glue between objects in the offline world and information in the online world. But there's a lot of tough computer-vision problems to solve before we get there."
Some vibrant visual search projects are also coming to the fore in Ireland. Hilary Kenna is the principal investigator on a new National Digital Research Centre (NDRC)-funded project called SeeSearch.
Kenna, who is also a lecturer in design and digital media at Dun Laoghaire Institute of Art, Design & Technology (IADT), describes SeeSearch as a feasibility project that is funding the development of a visual-search user interface.
"My research is trying to come up with new ways of representing how search results are displayed to the user through visualisation. When you search through any of the search engines you basically get a text-based list and a lot of people don't go beyond the first or second page."
Focusing on the library sector initially, SeeSearch also has the involvement of a commercial partner called Interleaf.
"It's going to sit on top of library systems and present visualisations of the search results back. The aim is that it will be applicable to any kind of searching for data retrieval."
Simon Factor, founder of Moving Media, is also behind a visual search project called MetaLabs that is developing video analysis technology to drive video search and online advertising.
Also operating out of the NDRC, Factor says MetaLabs was born after he came across the research Prof Alan Smeaton, founding director of the Centre for Digital Video Processing, was leading at DCU. "I met with Alan and talked about how we could commercialise some of the work his research teams were doing.
"These days, when you look for video, you get a blended search return that includes whole videos. What we're doing is fragmenting video into its various component parts.
"Most objects online, be they images or web pages, are one-dimensional. Video adds the element of time, which introduces a range of considerations when you are talking about how you search for it. What we are creating is temporal metadata," says Factor.
And with image analysis being a really hot topic right now, he says MetaLabs is taking this a step further as it's going beyond the static image to the moving image.
Finally, Factor says the challenge for developers in this space boils down to video-analysis technology being processor intensive, as well as how quickly they can process video. MetaLabs, however, already has development agreements in place with major video publishing platforms and will have products to market within the year.
And referring to the key value of visual searching, Kenna from SeeSearch concludes: "Certainly one of the key aims of my research is to visually show the relationships between information, rather than to just show them in a linear and text-based way."
© Silicon Republic Ltd