You’re probably sick of hearing about companies in the ad tech space using new tactics to help advertisers reach consumers.
But what if I told you there is a company with technology capable of opening up an entirely new data pool that is at once growing exponentially and offering insights more intimate than any Facebook profile or hashtag can provide?
Believe it or not, that seemingly optimistic assessment is just the tip of the iceberg for Netra.
Netra is a visual intelligence company that uses cutting edge image recognition and artificial intelligence technologies to help brands understand and target customers.
“Right now, companies are trying to understand consumers based on things like cookie data and purchase history, not off of what they’re actually doing in photos,” Netra CEO Richard Lee says. “Our tech uses imagery as a proxy for consumer intent and interest to help companies better understand their customers.”
By scanning billions of photos on social media sites like Facebook, Twitter and Instagram, Netra can help companies analyze specific customer segments and understand how consumers are engaging with their brand.
The company’s technology is part of the much-hyped race to create systems that can recognize what's going on in images. Joining them in that race are web behemoths like Google, Microsoft, Amazon, and IBM.
While Netra, a company of just 15 employees, faces stiff competition, some prominent angel investors seem to believe the odds are in its favor, and the more you learn about the company, the easier it is to understand why.
“We believe this is best-in-class in the brand and logo detection market,” Lee says. “There’s only one other company that can do brands, logos, objects, scenes and humans, and that's Google. In our head to head tests we perform two times better than them.”
Shashi Kant, an artificial intelligence veteran from MIT, founded Netra in 2013 after raising $100,000 from celebrity angel investor Mark Cuban. Lee joined the company in June of 2015, just as Netra was accepted into Techstars.
At the time, the founders had a promising image recognition technology but not much else. Techstars gave them an opportunity to do market research and determine the technology’s best use case.
“We thought, ‘Where is the biggest volume of imagery out there today?’ and between three and four billion pictures are being shared everyday on the top five networks of social media,” Lee says. “Yet, none of it is searchable unless you caption or hashtag the image.”
Kant began adjusting Netra’s semi-supervised machine learning technique to get better at identifying brands and logos. The system was soon good enough to start pilot programs.
In May of 2016, Netra partnered with Kantar, the data and analytics arm of multinational advertising giant WPP. A month later, the company raised $1.8 million in a seed round involving Launchpad Venture Group, NXT Ventures and several angel investors including Cuban.
From there, the founders went to work assembling an international all-star team of developers and engineers.
“I’m proud of what we’ve accomplished with such little money, how we’re outperforming the big guys,” Lee says. “We’re building our own algorithms, so we’re doing the same heavy lifting as Google, Microsoft and IBM.”
Today, the Netra team has refined its brand recognition system and the results are truly impressive.
Netra has 1,700 brand logos in its database and is adding 500 more every month. The current process to train the system on a new logo involves a combination of synthetic imagery and human feedback. The team is hoping to make that process entirely automated within the next year, dramatically reducing the time it takes to onboard a logo.
Companies using Netra’s web interface can view brand engagement metrics based on context, location and the demographics of people posting the image. They can also search by image type to get data on different customer segments like pet owners or new parents.
“We’re creating the visual intelligence overlay on these photos, which oftentimes adds a lot to the text,” Lee says. “So if someone’s doing an activity like skiing, that could be of interest to marketers if they’re targeting outdoorsy people.”
Over the past year, Netra’s focus has been to improve and expand its image recognition technology. But the founders believe the time has come to invest more aggressively in growth. The company is currently working with five customers in scaling their core image recognition services, as well as running several proof of concept pilots for new use cases like targeting.
“We’re focused on getting our tech in people’s hands so they can see how it’s working and the value they get,” Lee says. “We’re on the precipice of stepping on the gas because the response so far has been really strong. It’s ready for primetime.”
Of course, powerful image recognition technology can be used for a lot more than helping advertisers. As the father of two young kids, Lee has experienced the need for visual search capabilities firsthand.
“I have thousands of pictures of them and it’s this huge pain point trying to find specific ones,” Lee says. “Our belief is that visuals and voice are the ways people will search for and consume information online in the future. The company that can develop that will be a big gatekeeper; so there’s the potential to create the next Google if you can solve visual search.”
Indeed, the potential applications for the leading image recognition service are staggering. Although today it’s a powerful force for marketers, tomorrow it could be driving the AR rig everyone’s walking around with.
All this just means Lee and his team have a lot of work to do.
“We can’t stop because everyone’s investing in this space, so we just need to prove out a couple of use cases then expand,” Lee says. “It’s pretty scalable.”
The founders hope to go international quickly and are considering expanding into various vertical and horizontal markets.
Regardless of what direction the Netra team goes in, it’s clear the software will be up for the challenge.
Images courtesy of Netra.