Conference goers know this refrain all too well. They hustle from their hotel to the conference entrance — only to get stuck in a long, annoying check-in line. It’s the type of bottleneck that puts a bad taste in attendee’s mouths and starts great events off on the wrong foot.

But new technology is aiming to reduce check-in times to seconds—facial recognition. The concept is simple. Look into the camera, get verified, grab your badge, and go. Not only is it faster, it’s more secure. It helps find people who try gaming the system, like passing their badges off to friends or colleagues. It also prevents someone conference crashers from, say, finding a badge in the trash can and using it to sneak in or sweet-talking the registration folks into printing up someone else’s badge. Facial recognition can also be used to restrict access to specific areas—like spaces for speakers or VIPs.

Just like any new technology, facial recognition comes with privacy concerns. Who controls the data being collected and what are they doing with it? Indeed, the combination of digital facial maps, headshot images, names, credit card numbers, addresses and email addresses is a powerful mix for a would-be identity thief. Also, researchers have found that facial recognition can discriminate against minorities. How can event organizers ensure that the technology works for all attendees—and assuages security concerns? As facial recognition and advanced computing become increasingly ingrained in the meetings and event space, organizers must stay ahead of trends to ensure that the technology is being deployed appropriately.

Avoiding Traffic Jams

Panos Moutafis, cofounder and CEO of facial recognition company Zenus, thinks of check-in queues like highway traffic—small delays or increases in volume can slow down the entire process.

“Small delays can build up. And if you have a line of people, those delays keep growing and growing,” said Moutafis. “If you can handle 200 people per hour but you get 210 people, you’ll have a huge line. If you are able to remove even a small portion of the slowdown, it makes things flow cleanly.”

The Zenus facial recognition process works like this: Attendees are asked to submit a photo along with their registration forms. The Zenus system measures the photo’s quality then asks attendees if they’d like to opt-in to using facial recognition at check in. As attendees become more familiar with the technology, they’re more likely to embrace it. Take real estate firm Keller Williams for example. The first time Keller Williams deployed Zenus’ facial recognition technology at a conference, 45% of attendees opted in, Mooutafis said. By the fifth event, 68% opted in.

“People are more willing to opt in than event planners think,” said Moutafis, “as long as they feel comfortable about how the data is being handled.”

Zenus doesn’t carry any personally definable data on conference attendees. Instead, they assign each image with a conference ID number and match the facial recognition scan to that number. It then trashes that data after seven days. The event organizers typically hold data like names, addresses, and email addresses. 

A Technology in its Infancy

Facial recognition technology doesn’t always work. Researchers at the University of Toronto claim to be able to disrupt facial recognition systems by creating very slight disturbances to images that are untraceable to the human eye—like making the corners of the eyes just a bit less noticeable. They say they can reduce a system’s power from 100 percent recognition to 0.5 percent. Meanwhile the ACLU tested Amazon’s facial recognition technology on members of Congress, finding that it incorrectly matched 28 of them to criminal mugshots—and the majority of those false positives were minorities.

“Overall, facial recognition technology is in its infancy,” said Jason Porter, vice president at private security firm Pinkerton. “It’s very, very young technology. As history has shown, any new technology is often subject to vulnerabilities.”

Nevertheless, facial recognition (along with computer vision and artificial intelligence) are primed to transform the meetings and events industry over the next few years. Conference check-ins are just the beginning. Moutafis argues that conferences will soon have monitors around the venue offering personalized announcements like “your next session is at 12:30” or “your lunch is being offered upstairs.” He also believes biometric scanners will read body language to show attendees which sessions they liked best or tell organizers if attendee energy is lagging. He even thinks there’ll be a way to optimize your elevator pitch so you don’t have to repeat it over and over again as you meet new people.

“Every single touchpoint will have some form of computer vision or artificial intelligence,” he said. “But at the core of it, you have to focus on privacy.”

Four Steps for Successfully Deploying Facial Recognition at Conferences

1. Let attendees opt-in. Facial recognition technology is legal for use at events in every state besides Texas and Illinois, but organizers “must collect explicit consent during the image collection process and should explain to participants what is happening with the data,” said David Reischer, attorney and CEO of LegalAdvice.com.

2. Look for discrimination or bias and take steps to mitigate it. If you notice that your system works better on one racial group than another, you have to take action. Either find another software vendor or work with the one you’re using to train the system to become more accurate over time. In the meantime, have simple fallback procedures in place (like QR codes) that can get people through the line quickly.

3. Ensure your data-retention policy is consistent. Porter recommends having a consistent data retention policy that answers basic questions like: Who has access to this data? How long are they storing it? And are they selling it to anyone else or are they safely deleting it? 

4. Know what to do when isn’t recognized by your system. Event staff needs to be trained to respond if your facial recognition system fails to identify an attendee. Should they really be excluded from the event? “Or did we get a false positive on them because they decided to wear their glasses instead of their contacts?” said Porter.