Sightcorp Delivers Computer Vision Technology for Real-Time Face Analysis Hosted at the Edge

TL; DR: Sightcorp’s advanced AI technology, first conceived at the University of Amsterdam, equips those in the retail and digital signage industries with actionable, real-time insights designed to maximize ROI. Innovative solutions like the DeepSight Toolkit leverage computer vision and deep learning to analyze faces in images, videos, and real-life environments. By hosting its technology at the edge — in local retail stores and other settings — Sightcorp is working to digitize the offline world.

Like it or not, modern life involves a great deal of staring at screens.

According to a recent Nielsen audience report, Americans spend more than 10 hours each day doing so — and that’s just to consume media. Even when we’re out and about, we’re using self-service kiosks at banks, ordering and paying via touch-screen panels at restaurants, and viewing digital signage at supermarkets and shopping centers.

From an online retailer’s point of view, ever-increasing screen time provides the opportunity to track, collect, and analyze user data to benchmark performance and drive conversions. Now, taking advantage of the fact that screens are increasingly prevalent in brick-and-mortar environments, one company is working to bring that same opportunity to physical stores.

“We’re essentially bringing Google Analytics to the offline world,” said Joyce Caradonna, CEO of Sightcorp. “If you don’t know how many people go into your store a day or what their demographic profiles are, how can you make sure that the products at your shop are the right ones? How can you ensure the staff that works in the shop represent the demographic profile of your shopping audience?”

Considerations like these are vital for retailers looking to optimize the customer experience and drive sales. Sightcorp’s technology, hosted at the edge in retail stores and other settings, helps brands with a physical presence keep up with the ecommerce world in this respect.

“In the online domain, there are jobs where people focus on nothing other than optimizing the locations of buttons on a page to make sure that as many people as possible convert,” Joyce said. “The same thing hasn’t been happening at all in the offline domain, but with our type of technology, it can.”

Sightcorp’s founding team includes current Chief Technology Officer Dr. Roberto Valenti and Chief Scientific Officer Prof. Dr. Theo Gevers. Dr. Roberto Valenti completed his doctorate in machine vision at the Intelligent Systems Lab Amsterdam at the University of Amsterdam, and the results of his research are currently taught at institutions around the world.

Prof. Dr. Theo Gevers teaches at the Computer Vision Center (UAB) in Barcelona, Spain, as well as the University of Amsterdam, where he is director of the school’s Master of Science in Artificial Intelligence. He holds a VICI Award from the Dutch Organisation for Scientific Research, among other designations.

Both co-founders used their vast experience in machine vision and artificial intelligence to form the foundation for Sightcorp.

With Sightcorp’s crowd analysis solution, DeepSight, users can anonymously aggregate real-time information on the number of faces from any angle and distance, as well as demographics such as age group, gender, behavior patterns, and even mood.

“We’re giving users valuable information in regard to the effectiveness of campaigns, as well as using the information that we aggregate to optimize messaging for the audiences that stand in front of screens,” Joyce said. “Let’s say you’re in a supermarket, you pass the screen, and we see that you’re a female in the 30 to 40 age group. We can use that information to trigger, in real time, a campaign that is more suited or targeted at your persona group.”

This highly accurate demographic analysis is hosted locally on the edge, on-premises, and Sightcorp is careful not to violate data privacy rights. All data is translated in real-time into binary code, and video or picture footage is never stored or shared.

“The only information that does get shared to any cloud point location is the binary code strings — the zeros and the ones that make up data on demographic profiles,” Joyce said.

As part of its commitment to protect the privacy of customers and audiences, Sightcorp also includes a face-blurring feature designed for compliance with data regulations.

Between fostering better engagement and boosting conversion rates, the technology provides a clear return on investment for retailers.

“We help improve customer experiences and conversion rates by pushing certain products that otherwise might have been more difficult to sell,” she said. (To put this in context for our hosting-obsessed readers, the technology automates work that would otherwise be nearly impossible, much like the way hosted site builders use AI to take the sting out of development and design).

And while Sightcorp is not the only company in the world to offer such a service, it is the only one innovative enough to leverage the power of deep learning. “We are by far the most accurate and affordable solution out there, and we built everything ourselves, in-house,” Joyce said.

As businesses and stores slowly reopen amid the ongoing COVID-19 pandemic, Sightcorp is stepping up with ready-to-use solutions to ease the transition.

“Based on a lot of requests coming in from our customers and wanting to help as much as we could with reopening of stores, we added face mask detection to DeepSight,” Joyce said. “It’s essentially fool-proof: It can see if you’re putting your hand or a scarf in front of your face, and it uses the information to trigger specific content on screens when you enter any retail environment.”

The company is also offering tools for occupancy and entrance flow management, temperature sensing, and density management. Joyce said all of these solutions can replace the need to post expensive security guards at the front of stores.

“With a human, you won’t have full coverage at all given times due to bathroom and lunch breaks,” she said. “Using technology to trigger warning messages or sounds on screens in the event that someone is not wearing a mask is a more sustainable and accurate approach.”