Do You Know Who I Am?

Shutterstock / ImageFlow
Shutterstock / ImageFlow

Facial recognition has been a biometric technology under development for years. In the past, confidence in the technology waned as it presented risk of inaccuracy and even racial bias.

The National Institutes of Standards and Technology (NIST) tested facial recognition algorithms applied to nearly 27 million images to analyze their ability to identify people. NIST researchers tested artificial intelligence algorithms of 26.6 million reasonably well-controlled, live portrait photos of 12.3 million individuals. Three smaller datasets containing more unconstrained photos were also used: 3.2 million webcam images, 2.5 million photojournalism and amateur photographer photos, and 90,000 faces cropped from surveillance-style video clips.

NIST researchers found, with good-quality portrait photos, the most accurate algorithms will find matching entries in galleries containing 12 million individuals, with error rates below 0.2%.

While there’s still some controversy over its bias and misuse, A.I. and facial recognition have come a long way in the past decade. If the utility is well-captured, the use of facial recognition in everyday life is something any electrical contractor should keep an ear to the
ground about.

Is it viable in today’s world?

“Automated face recognition is already a viable technology comparable to human-level competence in most situations,” said Brian Burns, senior computer scientist at SRI International , Menlo Park, Calif. “It is dependent on the training set of images, which if not correctly balanced, can have biases with respect to subject race, age, etc.

“At SRI, we have successfully used it in our media forensic work to detect dubbing by matching the IDs of talking faces in videos to the IDs of speakers detected in the audio track. One issue can be speed and computational requirements. Newer, more lightweight networks exist and can probably be used to do frame-rate face recognition on conventional computer hardware.”

Tim Lynch is the president of Psychsoftpc, Quincy, Mass., and a psychologist who studies how computer interaction affects personality and how to make computer interfaces more user-friendly.

“While strides have been made in facial recognition and artificial intelligence, many problems remain since modern machine-learning-based A.I. is a black box; we cannot be sure of what the A.I. is thinking,” Lynch said. “This [allows] prejudice and bias to creep into the program, causing misidentification and false positives. It is an unfortunate fact of life now and possibly in the future, if we don’t act on it. It threatens our privacy; is subject to errors and misidentification, especially in regard to minorities; and has the potential to subject people to false accusations, detention and false denial of entry due to misidentification. 

"For electrical contractors, this could mean denial of work, detainment and denial of entry into workplaces because of misidentification or built-in program bias. It also means that everything they do can and will be tracked by the A.I. on the job and off. Be aware that you can be targeted, misidentified, falsely accused of misconduct or prior bad acts, and they will be closely monitored by the A.I. every second.”

Globally identifiable images

Cade Metz wrote in a July 2019 New York Times article, “Facial Recognition Tech Is Growing Stronger, Thanks to Your Face,” that a glut of photos and images is now available and traceable to an individual. Each individual face, like a fingerprint, is unique and identifiable and such identification is becoming even more possible thanks to the amount of data. The greater number of photos of you on the internet enables the technology to “machine learn” the data in a portrait photo and link it to a particular identity.

“Companies and labs have gathered facial images for more than a decade, and the databases are merely one layer to building facial-recognition technology. But people often have no idea that their faces are in [such databases]. And while names are typically not attached to the photos, individuals can be recognized because each face is unique to a person,” Metz stated.

“Regulations are being attempted at various government levels in the United States," he said, adding that, "On the international front, people are protesting the use of official recognition for citizens oppression in China and other countries. U.S. companies are the subject of protests to stop them from providing the technology to such states many times by their own employees. Be aware of these and follow them."

About the Author

Jim Romeo

Freelance Writer

Jim Romeo is a freelance writer based in Chesapeake, Va. He focuses on business and technology topics. Find him at www.JimRomeo.net.

Stay Informed Join our Newsletter

Having trouble finding time to sit down with the latest issue of
ELECTRICAL CONTRACTOR? Don't worry, we'll come to you.