Yandex, Russia’s search giant, now enables its users to surf the web for images that are similar or identical to the ones that they have uploaded.
The Russian news agency RIA Novosti reported last week, citing Alexander Krainov, who is in charge of Yandex’s computer vision technology department, that users can now enjoy this function by either uploading an image or entering the image’s URL using a special ‘camera’ icon in the search box.
Content-based image retrieval (CBIR), the technology behind the innovation, is believed to be a cut above the prior method of searching for an image by entering a text or voice query.
The new function will not only help identify an image or find a website containing the image but is also expected to become a user’s guide for online retailers. “For example, you have found an image of an exclusive chair on the Internet, and our service will take you to the online store that sells such chairs,” Krainov explained.
In spite of this, Krainov said that Yandex had no plans for business relationships with e-commerce operators.
The inaugural version of the service is designed for desktop computers, but its mobile adaptation is also in the pipeline.
Computer vision technology is no news to Yandex. The Russian company has already applied it to recognizing faces or screening out adult content in its Yandex.Kartinki and Yandex.Fotki picture upload and storage services.
Last week, Yandex presented another new innovative feature, this time aimed at advertisers. Yandex launched its Look-alike, a new service that rivals Facebook’s purchase tracker tool and helps advertising firms pick target audiences for media ad displays based on whether a user’s behavior is similar to that of existing customers.
According to RIA Novosti, Look-alike is powered by Yandex’s Crypta, a proprietary behavior analytics technology using the company’s own machine learning method MatrixNet. With set terms of search, the system will analyze the behavior of an advertiser’s available client base – for example, a given group that shops online or downloads price lists – and look for other users that act similarly.
