A computer's ability to accurately identify images is a white whale for many technology companies, from Baidu to Google.
。 One Australian startup has found a corner of the market to dominate, winning contracts with the European Union Intellectual Property Office (EUIPO) and IP Australia for algorithms that can detect and compare logos. 。 SEE ALSO:Airbnb is getting into the airline booking disruption game with 'Flights' 。TrademarkVision, which has support from Australia's CEA Startup Fund, uses machine learning to support image searches that can identify similar trademarks.。 Having a unique trademark or logo is vital, but many intellectual property registration bodies often require outdated forms of non-visual search that make comparison difficult.。 Australia, for example, relies on keywords, Europe on Vienna codes and the U.S. on design codes.
。 "The Nike logo is protected with words like 'tick' to describe the image and you're hoping that someone will use the word 'tick' when they search to see if they're copying someone else accidentally," TrademarkVision COO Cameron Mitchell explained to
。 Mashable。Mashable。 .。 The deep learning system the EUIPO has rolled out allows users to upload an image and search it against the existing database. 。Mashable Top StoriesStay connected with the hottest stories of the day and the latest entertainment news.Sign up for Mashable's Top Stories newsletter
。By signing up you agree to our Terms of Use and Privacy Policy.
。By signing up you agree to our Terms of Use and Privacy Policy. 。 Thanks for signing up!
。For example, here are the results when you search Adidas' three stripe logo. 。 Credit: euipo 。 Mitchell said the aim is to develop algorithms that can think more and more like humans when they see an image.。 For example, a computer might look at a beach ball and a penguin and see that they're roughly the same shape, he suggested. The challenge is to make the computer contextualise the object, as well as see its alike-ness.。 "What we have to do is train the computer to identify the object in the image as well," he said. "So there's a weighting between the semantic meaning -- what the object is -- and the image similarity."。 While Mitchell couldn't share specific cases due to client confidentiality, he said their technology is also used by companies monitoring trademarks to ensure nothing is registered that's too close to their own. 。 "It's a shift in a very archaic industry," he said. "The deep learning models that we use in our machine learning umbrella are just scratching the surface."。 The company also plans to tackle other types of image data, including industrial design -- 3D drawings that represent a new object or structure.。 "We have to get user data to understand when people think two things are similar," he said. "In the trademark space, that's our biggest challenge, but our vision is much, much wider than trademarks.
。 "Our vision is to be the leader in image recognition, full stop."。 "Our vision is to be the leader in image recognition, full stop."。Featured Video For You。This smart exoskeleton could help paraplegics walk again。 |