Image recognition software is already used by several social media groups for a variety of purposes, such as the Facebook feature that automatically recognizes faces to tag pictures of your friends. However, more of them are starting to use the software for recognizing logos and other brand-related data. A recent agreement between Tumblr and a firm that provides such social media image analysis called Ditto Labs is now aimed at identifying images that contain brand-related data to show businesses how they are perceived on Tumblr.
While the potential exists to turn this brand-name recognition into an ad machine that targets audiences based on what shows up on their photos, Tumblr is not seeking that route. “Right now, we're not planning to do anything ad-related.” said Tumblr's head of business development, T.R. Newcomb. Instead, Tumblr wants to grant advertisers the opportunity to thumb through images that contain their product and understand part of the nature of the conversation about them.
Of course, Tumblr is not the only one using this technology. In January, the image board Pinterest bought a two-person firm called VisualGraph that operates very similarly. However, the company stayed quiet about what exactly the group would be working on for the company, which could mean they could be introducing more focused ad-targeting. Facebook's own Mark Zuckerberg also invested in an image-recognition artificial intelligent firm known as Vicarious, which could offer startlingly powerful image recognition techniques. All of these could potentially lead to ad generation.
But so far, Tumblr's agreement seems to be limited to offer information to advertising partners, instead of directly enhancing ad-targeting. Despite the fact that doing so would be simple enough to code, Tumblr appears to be taking things slow, and would rather use the technology on a low-key basis instead of doing something too intrusive for users.
Edited by Stefania Viscusi