The New Reality
Examining whether good enough will actually do the job
- By Andy Newbom
- Apr 01, 2018
Technology marches ever forward,
lockstep with ‘Moore’s
Law.’ However, the influx of
technological change affects
different industries in different
ways. Some industries try to balance on
the bleeding edge, seeking out competitive
advantage and greater profits. Other industries
find that good enough beats best if good
enough gets the job done just fine. Until the
job changes.
Then, what is “good enough” changes
into something new. The corrections/detention
industry is undergoing this change. For
decades, the constraints on security and
surveillance in correctional facilities was essentially
how many camera feeds can you fit
into one screen and how many people does
it take to watch all of the feeds. Based partly
on technological constraints (analog CCTV,
low resolution, fixed cameras and high cost)
and partly on “how it’s always been done”
video surveillance was primarily a human
scale endeavor.
Those days are ending. Video surveillance
technology is leaping forward with real time
frames per second (15 to 30 fps), HD resolutions,
auto-follow cameras, motion detection
and tracking, and Artificial Intelligence (AI)
based machine learning, among many other
intriguing technologies. This technological
tidal wave is multiplied by increasing federal
and state regulatory requirements impacting
surveillance coverage, retention times, video
analysis and even legal challenges.
Technology x Regulations =
New Realities
The challenges of security and surveillance
in the correctional industry are daunting. Instead
of dozens of simple fixed cameras that
can easily be monitored by a crew of security
guards, correctional facilities now have many
hundreds, or thousands, of high resolution,
motion detecting, automated cameras and
drones feeding millions of frames per minute.
Those millions of frames must not only be recorded
and analyzed, they also have to be retained
for a year or more with zero frame loss.
These rapidly changing technologies combined
with the weight of the regulatory environment
are presenting not only new challenges,
but more importantly new opportunities.
These opportunities bring an entirely new level
of intelligence, service and safety to security
and surveillance in corrections facilities.
“Good enough” video surveillance is no
longer enough. Implementing an intelligent
video surveillance system that meets regulatory
requirements as well as delivers advanced
functionality for today and builds in
scalability and improvement for the future is
critical. What is needed is an intelligently designed
and implemented system that solves
problems efficiently and effectively, while
leaving room for growth and improvements,
and an empowered and well trained staff is
essential to successfully take advantage of
the tremendous opportunities an intelligent
surveillance system delivers.
Here are four keys to successfully implementing
intelligent surveillance in the new
security reality.
Human Analytics
Assume that all of your video surveillance will
move to human analyzable levels (20-plus fps,
HD resolution and full spectrum audio).
Humans are amazing pattern-recognition
machines and are remarkably good at discerning
patterns from complex visual and
auditory input. They have the ability to
recognize a large array of different types of
patterns and then transform these “recursive
probabilistic fractals” into concrete, actionable
steps based on what they learn.
As video surveillance in correctional
facilities moves from watching for major
events, towards ensuring security and safety,
the requirements for human analyzable video
surveillance is driving a massive movement
towards higher frame rates (20-plus fps),
and HD or higher resolutions. To fully take
advantage of human analytics in video surveillance
requires that the incoming video is
tuned to human movement and perception.
“It is no longer good enough to see that
someone is walking down the hallway carrying
something,” said a security professional
in a large correctional facility. “Real security
demands that we can quickly identify who is
walking and what they are carrying. Whether
it is happening right now, or during last
week’s cafeteria service. We need to know,
not just see.”
That’s just one of many reasons why the
demands of video surveillance are exploding.
Challenges. The challenges of humans
analyzing video surveillance are higher frame
rates, resolutions and camera counts result in
exponential growth in storage and analysis
requirements, human security officers cannot
watch all the video streams equally and more
than 90 percent of all surveillance video is
never watched or analyzed.
Opportunities. There are multiple opportunities,
however, which include using
motion detection, alerts and AI/machine
learning to augment staff to respond more
quickly and effectively, technology and system
design to get maximum video quality
and data while keeping requirements within
reason and more staff to monitor, analyze
and investigate with video surveillance can
improve safety and job satisfaction.
Acknowledge that advances in AI will continue to eliminate the
need for human perception.
Video Analytics
All your video belongs to the AI. Based on decades of investment
and research, AI based, machine learning video analytic engines are
evolving into fully deployable products, with user-friendly interfaces
and scenario-focused solutions.
According to artificial intelligence pioneer Ray Kurzweil, “All
learning results from massive, hierarchical and recursive processes
taking place in the brain.” AI and video analytics will continue to
bring substantial benefits to video surveillance.
“A lot of R&D now goes into making features in the VMS to help
human operators understand what is happening and what has already
happened,” said Tim Palmquist, vice president American, Milestone
Systems. “AI will likely be the ultimate delivery on the traditional
video analytics promise. The innovations underway in this area are
very significant and ultimately game changing. Smarter software will
ultimately render a lot of present-day actions obsolete.”
Artificial intelligence-based video analytics is poised to drive entirely
new security and surveillance solutions that deliver impressive
results at lower costs.
Challenges. The challenges of AI and machine learning are leading
edge technologies that require deep understanding and skill to
implement successfully. To truly leverage the power of AI, video surveillance
systems must have the higher FPS and HD resolutions that
also make them human readable.
Opportunities. While AI and video analytics are remarkably good
at recognizing objects, events and patterns in video they require
skilled analysts and investigators to interpret those results. Build your
security and surveillance systems with both long term storage and
long term analytics in mind. Design for the future.
Accept that new capture points like drones, robots, body cams
and other systems will add complexity.
IoT Devices
The proliferation of edge capture devices is expanding at breakneck
speed. Ground-based drones, flying drones, body cams and in-car
cameras for prisoner transport are driving new security processes
and adding complexity. Without a proactive plan, these new devices
can overwhelm a video surveillance system. Add in the explosion of
personal smart phones and myriad Internet of Things (IoT) devices
and a 500-camera system can easily grow by hundreds of inputs. Be
aware of both the physical security and cyber security implications
of these new devices.
“The issue of cybersecurity and protecting network surveillance
systems will be at the center of every discussion with vendors and
existing and new end-user customers to assess and combat the vulnerabilities
of IoT devices,” said Jennifer Hackenburg, senior product
marketing manager at Dahua Technology USA.
Challenges. As if managing video from hundreds of fixed IP
cameras wasn’t hard enough, now the cameras fly, walk, drive and
run around. All IP connected devices present inherent security risks,
newer IoT drives even more so, cybersecurity risks are exploding.
Opportunities. IoT does present some opportunities such as empowering
the staff to develop new skills utilizing these new mobile
and smart security devices while also improving officer and inmate
safety and security. Work with IT to implement facility wide BYOB
and IoT cybersecurity polices that serve and protect in balance.
Assert your plan for continuous growth in storage and analytics
requirements.
Video Surveillance
By 2020, video surveillance is expected to generate more than 859
Petabytes of new video every day. A 1080P HD resolution camera
in a typical corrections setting generates up to 10 GB of video every
day. With the number of cameras increasing rapidly and mandated
retention times stretching into years, these baseline requirements can
quickly consume Petabytes of storage. It’s not a question of whether
your storage and data protection needs will increase, it is only a matter
of how much will they grow: 2x, 4x or 10x over the next two to
five years?
To successfully solve these challenges and take advantage of the
opportunities, an intelligent storage system is essential. Modern video
surveillance has insatiable read/write performance demands and
long-term retention and analysis requirements. Desktop level solutions
are no longer good enough.
To meet these requirements, ensure your storage server vendor
is using Enterprise class SAS drives with MTBF’s of no less than
2M hours and optimized for 4k block sizes. To accommodate the
higher camera counts and resolutions, a solution that decouples the
video storage and protection from the video management is crucial
to enable ‘pay as you grow’ scalability and open standards, COTS
(Common off the Shelf) flexibility. Stick with a system that delivers
double drive failure protection and if needed on and off-site data
protection to ensure archived video is available to meet investigations
and regulations requirements.
Be proactive with your security and surveillance system. Actively
plan with the certainty that legally mandated retention times will increase,
resolutions and frame rates will continue to go up and that
video analytics and other AI functions will deliver even more value
from your video surveillance data.
Challenges. Increasing analytics requirements are driving exponential
growth in storage and intelligent systems. The algebra of increasing
FPS by resolution and cameras equals exponential storage
growth. Scalability is crucial, but budget constraints demand pay as
you scale approaches.
Opportunities. Assume requirements will increase dramatically
and build in scalability and open standards flexibility. Leverage newer
technologies such as Software Defined Storage and intelligent appliances
to provide storage over iSCSI/Ethernet and file-sharing NAS.
Use technology advances to bring new opportunities to staff for advanced
learning and skills development.
Intelligence + Technology + Skilled Staff
= New Security Reality
The future of video surveillance in corrections is thousands of cameras
running at megapixel resolutions and 20 fps and higher. Increased
human and machine intelligence and analytics both at the edge and
in the core will drive new, exciting opportunities and solutions. Conventional
IT-centric problems such as disaster recovery, backup, cybersecurity,
and on and offsite data protection are moving into video
surveillance systems in corrections and beyond.
The two most critical pieces of any video surveillance system are
the video management system and the data storage platform.
Smart organizations are building in intelligence and scalability
into both of these crucial systems. Storage and processing demands
will continue exponential growth. Instead of trying to bet against the
tsunami of technological advances, pro-actively leverage these advances
to deliver new capabilities, improved safety and security and
lower operational costs. Take a layered approach
when optimizing your security and surveillance
system by leveraging both human and video analytics.
That is a truly intelligent surveillance system
backed by intelligent people.
This article originally appeared in the April 2018 issue of Security Today.