Schools are Investing in Face Recognition. Will It Make Campuses Safer?
Schools are investing in security with advanced technology such as facial recognition
- By Peter Trepp
- August 01, 2018
As Dr. Larry Barton soberly noted at a security conference
in Las Vegas this past February, mass shootings
are “the new normal.” For many years, the University of
Central Florida professor and FBI instructor tracked
the increasing frequency of mass shootings with dots
on a timeline. But in recent years, mass shootings became so frequent
that the dots on Dr. Barton’s graph formed a single solid line.
Education Week’s log of school shootings reveals at least 13 incidents
in 2018 so far, with 97 people killed or injured. How do we stop
this unending cycle of violence? The political debate surrounding this
issue has become more fervent since the shooting that took 17 lives at
Marjory Stoneman Douglas High School earlier this year. While proposals
range from banning all guns to arming teachers on campus,
almost every serious plan is somehow divisive, with major opposition
from public, corporate and political factions.
But a new layer of security that is substantially more pragmatic is
now gaining traction: face recognition.
Consider how much was known about the suspect in the Parkland
incident before an Uber dropped him off in front of Marjory Stoneman
Douglas High School, armed with an AR-15. Parkland’s sheriff, Scott
Israel, revealed that the department had received 23 calls about the
suspect over the past decade. The calls mentioned a “mentally ill person,”
a “domestic disturbance” and more. An unidentified peer counselor
alerted the high school that the suspect had inflicted self-harm
and “wished to purchase a gun.” The sheriff ’s office received a tip that
he had been collecting guns and knives and “could be a school shooter
in the making.” According to a math teacher at the high school,
“administration had sent out an email warning teachers that [he] had
made threats against other teenagers.”
By the time that the suspect was expelled from school, there was
enough data to conclude that he posed a very real risk to school safety.
Given all the valuable information available, how did an expelled student
with a history of making threats and violence end up taking 17
innocent lives? The fact is that data itself is only valuable if it is delivered
in an actionable context. By the time the shooter was recognized
by a faculty member, walking purposefully toward a school building, it
was already too late. Law enforcement officials could only react to a
horrible crime in progress.
If the suspect's photograph and details had
been enrolled in a school face recognition
database, could lives have been saved?
Technology companies are typically contacted
by enterprises with expensive problems
that need solving again and again. For
example, retailers need to identify shoplifters
and other criminals on a daily basis across
many locations, and face recognition offers
large chains massive ROI as well as safer
stores. Increasingly, however, the number of
concerned administrators, security personnel,
campus police and even parents looking
to use technology and face recognition to
supplement school security has grown.
HOW FACE RECOGNITION
COULD PROTECT SCHOOLS
Face recognition technology has been around,
in one form or another, since the 1960s. Starting
around 2011, it reached the point where it
was fast enough and accurate enough to provide
actionable intelligence about persons of
interest in real time.
Although many situations play out like the
one at Marjory Stoneman Douglas, not all do.
What is certain is that in a mass shooting or
any other violent crime situation, every second
is vital. The sooner that armed security or
law enforcement officials are notified, the better
chance there is to save lives.
In a campus situation, entrances can be
monitored by inexpensive HD cameras posted
at distances up to 100 feet. Artificial intelligence
automatically scores the facial images
captured from live video, selecting the
exact angle, facial expression and lighting
before matching it against a database at a
rate of 25 million images per second. Speed
is essential, as match alerts will be automatically
routed to campus security or other personnel
Depending on the type of person matched,
face recognition systems can also automatically
determine whether to alert others, such
as teachers, professors or even law enforcement.
Match confidence scores can also
determine who receives alerts. For example,
some organizations only want to be notified
on matches with at least a 95 percent confidence
If such a system had existed in Parkland, it
might have included disgruntled former
employees, expelled students and even banned
parents. Based on everything we know now, it
would have also certainly included the shooter.
The moment he appeared on a face recognition-
equipped surveillance camera, the system
would have then alerted an armed school
police officer or security guard.
For such a system to work, school personnel
would be trained on how to interpret
mobile alerts, how to approach matched persons
depending on various scenarios, and
more. The training is actually the easy part.
Based on experience with retailers and public
safety officials, training is easily completed
within a half day.
While K-12 schools are increasingly looking
to face recognition surveillance as a safety
solution, facial recognition also presents colleges
and universities with a means of protecting
campuses and school events. As San
Francisco University has found, face recognition
can be used to ensure that only authorized
students and personnel are admitted to
dorm rooms or school buildings. The technology
can also be used to instantly recognize
when a known sex offender or violent criminal
enters campus grounds.
Facial recognition can also be used to
secure school sporting events. Unfortunately,
it’s not unusual for fans to exhibit violent or
dangerous behavior at games. As just one
example, at one popular tailgating spot for
Nebraska football, law enforcement had to
clear around 3,000 people off the property
due to unruly behavior prior to the start of a
game against University of Miami. Close to
two dozen people were arrested. Even if the
University banned those individuals from
returning, it’s nearly impossible to enforce
that ban. But face recognition can typically
recognize banned fans, despite hats, glasses,
facial hair or other changes in appearance.
FACE RECOGNITION AND THE
FALLIBILITY OF HUMAN MEMORY
To understand why face recognition is necessary
for protecting public spaces, look no further
than human memory. The human brain
was simply not designed to memorize names
and details about a large list of strangers.
Robin Dunbar, an esteemed evolutionary
anthropologist at Oxford University, posits
that the upper limit of faces that human
beings can match with names is 1,500.
FaceFirst recently conducted a survey that
asked whether those tasked with guarding
airports and public attractions would be able
to remember the names and faces of individuals
that posed a threat. Seventy-seven percent
said no, and it’s easy to see why. Even a truly
gifted security professional can’t possibly be
expected to instantly recognize every individual
that may pose a public safety threat,
including all relevant details and the best
course of action to take. This is certainly true
at K-12 schools, but doubly true of guards
tasked with protecting large university campuses
and college sporting events.
While it remains to be seen whether political
solutions will help decrease school shootings,
one thing is for sure: progress in the political
realm will be a long and slow
process. Fortunately, we
have the technology to start
saving lives right now.
This article originally appeared in the August 2018 issue of CSLS.