Follow Datanami:
August 21, 2014

Selfies Spawn Photo Analytics

An estimated 1.8 billion photos are shared on social media each day. Current “social listening tools” can’t analyze all this data.

Enter Ditto Labs Inc., which touts its proprietary “visual recognition engine” that locates brand names in social media posts, then turns the information over to marketing geniuses.

It’s the brave new world of “photo analytics,” according to Ditto, that uses its recognition engine to scan photos for corporate logos that might hint at how consumers “interact” with products and brands.

Ditto claims its solution outperforms hashtag and text-based searches because it is able to discover and analyze photos of logos and brands. The photo analytics vendor claims it can answer questions like where photos were taken and when did the consumers in the photo use a particular product.

Ditto announced a “firehose partnership” with the Yahoo-owned social media site Tumblr on August 19. Tumblr users post an estimated 130 million photos a day. “The deal will help companies better understand consumer preferences for products and services as well as affinities by looking at the brands people include in their photos,” Ditto noted in a statement announcing the deal.

Ditto has similar deals with Twitter and Instagram for “brand photo listening capabilities.” The company also offers “photo-based social ad-targeting” for retailers.

Among Ditto’s other analytic tools are an API designed to combine data gleaned from posted photos with text analytics.

Tumblr told the web site Mashable that it has no immediate plans to start sending out targeted advertising. Instead, it reportedly will share Ditto data with advertisers who want to know how they are perceived on hip web sites like Tumblr.

Ditto also makes photo analytics from Tumblr and other social media clients available to its customers.

Ditto’s visual recognition engine is based on image recognition technology long used to acquire, analyze and understand images. The company leverages pattern recognition in particular to spot generic objects in a photo like a bicycle or the pixel pattern of the Coca-Cola logo.

Pattern recognition allows users to “simply upload an image of the Coca-Cola logo to a search algorithm, which can find that image in other images,” the company noted in a blog post. “It is a more natural and intuitive way to search the ever-expanding visual web.”

Ditto appears to have the emerging photo analytics market largely to itself, claiming that it proprietary computer vision technology provides “structure to social media’s unstructured visual content.”

The privately-held company based in Cambridge, Mass., closed a second funding round in May that raised $2.2 million. The lead investor is Cue Ball Capital. Many of the company’s executives are MIT-trained computer scientists. Among its strategic advisors is Nicholas Negroponte, co-founder and director of the MIT Media Lab.

Recent items:

GPU-Powered Tagging Service ‘Gets’ the Picture

Twitter Looks to Leverage Social Data with Gnip Deal

Datanami