VALUE OF DATA

SEEING
WHAT
MATTERS
A New Paradigm for
Public Safety Powered by
Responsible AI
A survey administered across 44 global cities
in 2015 and 2017, found that residents of 23 cities
(52 percent) felt less safe over that period of time.1
45% of Americans also worry about being a victim
of a mass shooting.2

The unpredictability of crime is increasing as more

of it occurs in areas typically deemed safe: from
schools to nightclubs to concerts to churches.3
Should we allow this to be the new normal?
It doesn’t have to be. Imagine a future where the entire city is monitored by responsible AI,
providing law enforcement with the tools and intelligence to stop atrocities and virtually all
crime in real time. Cameras could detect people running, falling, or fighting; unattended
items; other hazards (e.g., flame, smoke, vibration, non-standard noise)4 and potentially
suspicious activities such as a person carrying suitcases repeatedly or groups of masked
people continually gathering at night. The system alerts police who react in real-time—for
instance, recognizing an offender’s face before any crime is committed and automatically
activating alarm systems or contacting emergency services.5

These cameras of the future are an extension of what is used today—the pervasive closed-
circuit television (CCTV) cameras (see box), which are one of the most popular tools for
stemming crime. An analysis of 13 European and North American cities and towns that use
CCTVs found that these cameras reduced crime (i.e. burglary, theft, homicides) by 1% to 75%.6

Closed-circuit television (CCTV) monitors transmit video camera footage and often
have recording capabilities. CCTVs are used for security and most do not include analytics
features today.7

The vast difference in effectiveness is due to the public’s perception of whether or not the
CCTVs are properly monitored. Take, for instance, an Urban Institute study,8 which looked
at the deployment of public safety cameras in two Chicago neighborhoods: Humboldt Park
and West Garfield Park. Humboldt Park’s crime rate dropped 20 percent the month after
the cameras were installed and remained low; in contrast, West Garfield Park’s crime rates
did not budge. Why the difference? Researchers found that West Garfield Park residents
did not believe police regularly monitored the cameras, a perception that made the
cameras far less effective as deterrents. Ultimately, by implementing these systems, the
city realized net savings of more than $600,000 per month from criminal justice costs and
victims’ financial and emotional costs.9 However, the city could have saved much more if
West Garfield Park residents believed real time monitoring occurred.

Given the video surveillance industry’s expected 2.5X growth, less than
1% of video footage is likely to be analyzed by 2020.
Integrating artificial intelligence and analytics with public safety’s current
closed-circuit televisions can transform the ecosystem from reactive to
proactive, reducing US crimes by about 81.7 million by 2040 and shielding
approximately 76,000 mass shooting victims by 2040.
AN ANALOG SURVEILLANCE INDUSTRY
What is interesting is that West Garfield Park residents’ belief was not far from the truth.
According to Accenture analysis, most video footage from public safety cameras today
are only used for forensic or evidential purposes after criminal activity has occured. For
instance, Beijing is the world surveillance leader with 470,000 CCTV cameras, followed
by London with 420,000, and Washington, D.C. with 30,000.10 Yet, on average only 2% of
video footage is seen11 and even less is analyzed in real time today. This means that 98%
of security video footage is “dark” data (see box).

Gartner defines dark data as information collected, processed, and stored

during regular business activity, but not meaningfully utilized or analyzed.12 IDC
has stated that up to 90% of big data is dark data.13 Given 2% of video footage
is seen today, 98% is “dark.” However, the surveillance industry is expected to
grow 2.5X, which means less than 1% of footage will be analyzed by 2020.
Source: https://www.gartner.com/it-glossary/dark-data/ https://www.techopedia.com/definition/29373/dark-data

And the surveillance industry is only growing, expected to reach approximately $76
billion in revenue in 202014—up from $30 billion in 2016 (16.4% CAGR)—with China
alone expected to install 450 million new CCTVs by 2020.15 This means that even more
video will need to be monitored—mostly over the internet. In fact, traffic from video
surveillance transmitted over the internet is expected to increase sevenfold between
2016 and 2021.16 A big part of this increase will come from the availability of high-quality,
affordable cloud-based do-it-yourself video solutions that households are embracing to
obstruct crime. For example, 35 million Americans currently own a home security system,
which represents a 6 percent growth from 2008 to 2017.17

Yet, more cameras do not equate to lower crime. Today, most police officers’ time is
dedicated to safeguarding streets while the rest is reserved for administrative work.
Not many precincts have dedicated video analytics professionals. As a result, most
video footage is left unwatched. That needs to change. Society needs to think about
better allocating safety and security investment dollars—and emerging technologies
can help. Integrating artificial intelligence (AI) and analytics with public safety’s current
CCTVs can transform the security industry from reactive to proactive—thus, allowing law
enforcement to more effectively combat crime and mass shootings.
HOW TECHNOLOGY
ENABLERS CAN HELP
FIGHT CRIME
AI is an ideal solution to augment current public
safety systems. AI enables machines to interact
naturally with people, data, and the environment,
creating more intuitive interactions and extending
what either humans or machines can do on their own.

AI technologies, such as machine learning and deep learning,

can handle computational-heavy and monotonous monitoring
that no human can possibly do in an equivalent amount of time.18
Effectively they become smart eyes and ears that can alert, in
Artificial Intelligence is
real time, security guards and police to take action that can
defined as applications that
prevent crime.
can sense, comprehend, act
and learn leveraging machine
In a public safety setting, AI capabilities—programmed by
learning, natural language
humans and trained over time—could analyze thousands of video
processing and computer
feeds to track and alert authorities of anomalies. Current video
vision to automate and
feeds AI could analyze include those from CCTV, do-it-yourself
augment human activities.
video cameras, security services, badges, and smart locks.
In the future, these could be joined by new data sources such as
personal devices (e.g., cell phone and iPad cameras, dash cams)
connected cars, drones, and robots. If enabled, cities could
crowdsource commercial and residential security system data,
thus breaking down the silos across those systems and analyzing
the data in an aggregated way to get a real-time picture of
potential criminal activity.

5 | SEEING WHAT MATTERS

Riding the prevailing technology wave, some public safety
industry players – like China, Singapore, and startups globally – are
already upgrading their capabilities. They’re leveraging robotics,
computer vision, big data, and analytics to anticipate and prevent
crimes in real time. As startups and established technology
companies begin partnering, they can tap into each other’s
expertise and continue to develop these technologies faster and
more affordably. And when these technologies come together,
they can transform cameras from an illusion of protection to a real-
time vigilant protector, thereby, accomplishing the unthinkable:
detecting crime in just one second.19

PHYSICAL ROBOTS DO THE LEGWORK

One of the biggest challenges for police and security forces is
manpower: there are only 800,000 police officers and 1.1 million
security guards in the US20 to cover a population of 327 million.21
In other words, there’s simply not enough police and security
forces to be everywhere they are needed. Enter robots, such as
those from Silicon Valley’s Knightscope, which track blind spots
and provide location-based data. Knightscope offers a fleet of
autonomous, mobile robots that patrol offices, parking lots, malls,
and streets to serve as security forces’ smart eyes and ears. As they
gather data, the robots spot anomalies, in real time – in any type of
environmental condition – and alert security officials. The robots
can detect concealed weapons, and have helped security guards
apprehend thieves as well as stop vandals. To date, Knightscope’s
robots have collectively covered more than 100,000 miles in
180,000 hours of operation. Yet, having too many robots in public
spaces can intimidate the public. Hence, robots may serve as blind
spots detectors, while AI serves as the real-time field of vision
public safety counterpart.

6 | SEEING WHAT MATTERS

AI’S COMPUTER VISION CAPABILITY
SUPERCHARGES RESPONSE TIME
Emerging AI solutions illustrate what future public safety could
look like—which, in many ways, is already here. Through facial and
Deep Learning is an AI
object recognition, AI’s computer vision moves cameras from an function that imitates the
illusion of protection to a vigilant partner in fighting crime. human brain in processing
data and creating patterns to
One computer vision startup working on facial recognition is
make decisions. Its algorithms
Alibaba-backed SenseTime.22 Its software matches crime scene
require massive data to
offenders to criminal databases in seconds to remove threats provide the best outcome.27
off streets. Thus far, the tool has identified 2,000 suspects
and “solved 100 cases” in Guangzhou, China.23 The company
is currently working on software that will “parse data from
thousands of live camera feeds” and be used by police to “track
everything from vice and accidents to suspects on blacklists.”24
Within facial recognition,
Meanwhile, One Smart Labs, is integrating AI into existing deep learning replicates
CCTV systems to identify subjects by matching faces with a images very closely to
database of millions of photos (including mug shots and driver’s originals—one startup claims
licenses).25 Understandably, such a tool could raise fears among to be 99.8% accurate—while
ordinary citizens of privacy invasion, which is why One Smart machine learning’s images
Labs and other companies with similar technologies continually are blurry.28
test their tools’ algorithms to identify false positives and matches Sources: https://www.investopedia.
and minimize bias. A subset of AI’s machine learning, called com/terms/d/deep-learning.asp
https://www.kairos.com/blog/the-
deep learning (see box) is at the heart of a third solution.
best-explanation-machine-learning-
Using deep learning, a system called SkyNet can accurately vs-deep-learning
identify people’s faces from different angles, motion speeds, and
lighting conditions. It claims to be up to 99.8 percent accurate.26

Likewise, deep-learning startup, Deep Science AI,29 is working

on object recognition within its analytics-powered platform.
It supports remote monitoring of facilities such as offices and
retail stores to detect threats including weapons, masks, intruders,
and fire. A human analyst monitors feeds from the facilities’
security cameras and immediately contacts the business and
authorities when a threat emerges. Deep learning detects threats
in fractions of a second, enabling an analyst to simultaneously
monitor hundreds of feeds instead of a handful. And if the system
generates a false positive, that data is kept to re-train the system.30
By 2022, video and image analysis will comprise 30% of enterprise
deep-learning deployments,31 meaning the technology will
increasingly become smarter over time as it is fed more data.

7 | SEEING WHAT MATTERS

By leveraging facial and object recognition with real-time action,
the impact of the most mature level of AI surveillance could be
monumental. The perpetrator of the deadliest mass shooting
Computer vision tracks and
in American history brought more than 10 suitcases to his hotel
identifies crime suspects
room filled with weapons.32 Sophisticated AI-enabled analytics based on data it is fed. It has
could have flagged the number of bags as anomalous behavior or identified 2,000 suspects
robotics could have identified the weapons hidden in his luggage. for one police district.

Increased public safety due to real-time monitoring has

profound effects on violent and non-violent crime as our analysis
demonstrates (“What’s the Value?” section below). However, the
mass public safety ecosystem could also help countries combat Deep learning processes
another major issue—the rising incarceration rate. massive data and creates
patterns for decision-making.
It will become smarter over
PUBLIC SAFETY ECOSYSTEM DUAL-PURPOSE: time: by 2022 video analysis
“PROPERTY CRIME” PAROLE PROGRAM will comprise 30% of enterprise
deep-learning deployments.
In 2017, the US had a nationwide incarceration rate of 860 inmates
for every 100,000 adults—which equates to 2.2 million people of
the country’s 327 million population behind bars.33 The cost of this
incarceration is more than $80 billion a year to support federal,
state, and local corrections agencies.34
Robots patrol public spaces,
More sophisticated public safety technology could reduce that detecting concealed weapons.
spend significantly. For instance, a large number of inmates are One startup’s robots have
covered 100,000 miles while
serving time for non-violent “property crimes” (i.e. burglary,
identifying thieves and vandals.
larceny-theft, or motor vehicle theft). While in prison, these inmates
are non-contributing members of society. However, many of them
could be eligible for parole if “supervised” by technology.
AI’s tools and solutions could track these individuals in real
time after their release and ensure they do not pose any threat
to society—and allow governments to reduce their corrections
footprint.35 An additional upside benefit: If released on parole, non-
violent inmates can get jobs, thereby contributing to GDP.36

8 | SEEING WHAT MATTERS

AN EAGLE EYE FOR SECURITY
Enabling the future vision of public safety depicted above,
the video surveillance industry is set to boom over the next
decade. With over 450 million CCTV cameras currently installed From 2018 to 2020, China’s
worldwide,37 the market is set to explode driven by the ambitions increase in CCTVs will
of the largest market – China.38 By 2020, China expects its installed generate 6.4 exabytes
base to reach 626 million cameras.39 With 29 percent of new of video per day. This is
shipments recording 4K resolution,40 the approximate 400 million equivalent to the data held
cameras that China aims to install in the next two years would in 1.4 billion DVDs. This
amount of DVDs stacked
generate roughly 6.4 exabytes of video per day,41 equivalent to
equates to two round trips
1.4 billion DVD’s. This amount of DVDs stacked equates to two
to the moon. Clearly, it is
round trips to the moon. Clearly, it is impossible to dissect all this
impossible to dissect all this
information without AI.
information without AI.

As staggering as this number is, it involves only the growth in the

Chinese market and only a single technology – CCTV cameras.
Additional sources such as body-worn cameras, drones and
dashcams make China’s 1.4 billion DVD’s of content per day a
drop in an ocean of potential data sources which will need to be
proactively monitored to ensure safety.”

Clearly, as the amount of data generated from both existing

and new sources increases dramatically, so will demand for
places in which data can reside and be used. In other words,
law enforcement agencies will need surveillance-optimized data
storage management that allows them to capture, store, and
analyze exponentially more data to generate real-time insights.
Specialized surveillance equipment exists today and will continue
to advance in features. This equipment can record continuously
for years, works with deep learning, and is weather, vibration, and
shock-proof.

9 | SEEING WHAT MATTERS

A TIERED APPROACH
TO ADOPTION
While the preceding technologies and data sources
are just now emerging, broad adoption probably
will not happen quickly given public concerns about
privacy and the massive investment required.
Instead, global adoption will likely unfold across three tiers of maturity of mass public
safety to allow society to gradually accept and advocate for it. However, given the rapid
public safety momentum in places like Singapore and China, such locales may reach
these tiers much more quickly.

TIER 1
Tier 1 is the current mass public safety ecosystem, which primarily features CCTV
feeds used retroactively by security and police forces to understand “what happened.”
Data is typically in silos, making it difficult to create a broader and more comprehensive
picture of a particular situation. However, some cities – like those in China and
Singapore – and startups begin to leverage emerging technologies to manage
mundane public safety tasks.

TIER 2
Tier 2 takes the next step toward a mass real-time-oriented public safety ecosystem where
through AI, police can see the unseeable. This tier is expected to arrive by 2025, by which
smart cities become the norm42 and 88 municipalities are predicted to achieve “smart city”
status.43 In addition, by 2025, 70% of security surveillance cameras will ship with on-device
real-time monitoring and analytics functions within the camera, compared with less than 5%
in 2018.44 As a result, governments and law enforcement can crowdsource video data from
businesses and public institutions (such as schools and hospitals) to augment their current
CCTV feeds and add AI capabilities that enable them to track and analyze footage in real time
to identify anomalies and threats. Due to increased public safety coverage and a centralized
AI-enabled analytics system, response times are improved dramatically. Federal regulators
estimate that shaving a minute off response times could save as many as 10,000 lives a year.45

10 | SEEING WHAT MATTERS

At Tier 3, the most mature predictive-oriented public safety ecosystem emerges by
2035, the date by which all devices are connected via the internet.46 Data automatically
knows where to go to be processed, allowing governments and law enforcement to add
households to their crowdsourced data sources, focusing on those in high-crime areas.
Some residents voluntarily donate their feed data, while others are encouraged to do so
by tax-break incentives or nominal compensation (determined by the quality of camera
generating the feed, the uniqueness of the view provided, the camera’s location, and the
neighborhood’s crime rate). These transactions are facilitated through a data marketplace.

TIER 3
Tier 3 relies on a rapidly growing public safety ecosystem where data is pulled from
disparate databases such as social media, driver’s licenses, police databases, and dark
data. The more data, the better the systems perform as deep learning enables the system
to become more knowledgeable and, as a result, more accurate.47 With these new sources
added to the mix, public safety covers virtually every inch of a city.

Such predictive analytics can stop crime and atrocities in real time. Utilizing FBI data,
an analysis of mass shootings between 2009 to 2016, demonstrated that 42 percent of
shooters exhibited “red flags” like acts, attempted acts, or threats of violence towards
oneself or others; violations of protective orders; or evidence of ongoing substance abuse.
In addition, 34 percent of shooters were prohibited from possessing firearms.48 Within Tier
3, cities have full public safety access housed in a centralized AI-enabled analytics unit.
Within the unit, police can view such anomalies in real time and interrupt a crime before it
is committed such as by locking down facilities to respond to threats immediately.49

The mature predictive-oriented public safety ecosystem emerges

TIER 1 TIER 2 TIER 3

Current Public Safety 2025 Public Safety 2035 Public Safety
Ecosystem: CCTVs are used Ecosystem: Video data is Ecosystem: Video data is
retroactively to understand crowdsourced from the crowdsourced from residents,
“what happened.” Data is private sector, augmenting and data is supplied from
housed in siloes. CCTVs with AI capabilities disparate sources, to predict
and real-time analytics to crime in real time.
identify anomalies.

11 | SEEING WHAT MATTERS

WHAT’S THE VALUE?
AI-enabled public safety can counter some of
the most common types of crime: property crime
(i.e. burglary, larceny-theft, motor vehicle theft,
and arson) and violent crime (i.e. murder, robbery,
and aggravated assault).50

Implementing a comprehensive public safety system

will clearly require significant investment, not to mention
a sharp, ongoing focus on protecting citizens’ privacy.
But the payoff could be worth it many times over in lives Number of Crimes AI Public
saved, mass shooting victims shielded, violent and property Safety May Prevent by 2040:
crimes reduced, and economic impact. 81.7 million

SAVING LIVES
Our analysis shows an increasing number of homicides occur
in public—which means they are typically visible to anyone Lives that May Be Saved by
or anything nearby. The problem is, the current retroactive- AI Public Safety by 2040:
oriented Tier 1 public safety industry may only prevent one
115,000
homicide per year.

Machine learning, deep learning, and cognitive analytics can

track homicides and mass shootings in real time—alerting
authorities and allowing them to respond more quickly and
Mass Shooting Victims AI
efficiently and, thus, saving lives. For instance, a Tier 2 public
Public Safety May Shield by
safety system may save under 3,700 lives in 2025 and over
2040: 76,000
8,000 in 2034. At Tier 3, homicides may be reduced by
approximately 8,800 in 2035. All three tiers combined may
result in nearly 115,000 lives being saved by 2040.

Economic Impact of “Property”

SHIELDING POTENTIAL VICTIMS
Crime Parole Program by
There’s similar potential to protect potential victims—those
2040: $191.8 billion
injured or killed—from mass shootings. Today’s public safety
approaches may shield from harm over 30 mass shooting
victims annually. At Tiers 2 and 3, the number jumps to
approximately 4,500 in 2025 and 4,500 in 2035, respectively.
All three tiers combined may result in about 76,000 mass
shooting victims shielded by 2040.

12 | SEEING WHAT MATTERS

REDUCING VIOLENT CRIMES AND PROPERTY CRIMES
When people are aware they are being watched, they are less likely to commit crimes
of any type.51 If people choose to proceed regardless, real-time AI-detection of their activity
would disrupt would-be criminals’ attempts to carry out a crime.

By progressively applying new technologies to their approaches, public safety agencies

could boost the number of violent and property crimes they prevent each year from
approximately 2,100 today (Tier 1), to 4.2 million in 2025 (Tier 2), to 5.6 million in 2035
(Tier 3)—or 81.7 million by 2040 (all three tiers combined).

BOOSTING ECONOMIC IMPACT THROUGH “PROPERTY CRIME”

PAROLE PROGRAM
As mentioned earlier, a mass public safety ecosystem could make possible the release,
and tracking via AI solutions, of non-violent prisoners serving time for “property” crimes—
allowing these individuals to become once again contributing members of society and
eliminating their associated incarceration costs. The accompanying economic impact is
dramatic. Because the mass public safety systems required to support this complex initiative
are not currently available, this impact would not be felt until Tier 2, which could be roughly
$21.5 billion generated from the release of about 176,000 eligible parolees. At Tier 3, the
number could rise to approximately $21.8 billion from the 136,000 parolees. Combined,
Tiers 2 and 3 could generate a massive $191.8 billion in economic impact by 2040.

When people are aware they are being watched, they are less likely to
commit crimes of any type. If people choose to proceed regardless, real-time
AI-detection of their activity would disrupt would-be criminals’ attempts
to carry out a crime.

13 | SEEING WHAT MATTERS

FORGING AHEAD TO A
MORE TRANSPARENT AND
RESPONSIBLE FUTURE
With so much data available today, it is understandable
people are wary of proposals that suggest even
greater data collection by government and law
enforcement—knowing the potential exists for misuse
of the data and, consequently, violations of privacy.

Plus, many people are inherently uncomfortable with the thought 01 Public input and
of constantly being monitored as they make their way through oversight of public safety
their days. But the tradeoffs in the form of crime prevention— systems is necessary
saving lives and property—and positive economic impacts argue
02 Data is used for only
in favor of more expansive, and more intelligent, public safety.
for the purpose for which
it is gathered, is not sold, is
Thus, going forward, a balance needs to be struck between
anonymized, and is secured
data privacy and data responsibility. Public input and oversight
are necessary to ensure that video public safety systems are 03 GDPR serves as
designed to prevent misuse and abuse. Members of communities governance model for future
considering such a system should participate in the decision- of public safety ecosystem
making process to build trust and tailor public safety processes to and is secured
the community’s needs and circumstances. Such public forums
also give officials a venue in which they can explain how the
system’s costs are far outweighed by the its immense benefits.52

Indeed, some civilians aren’t opposed to being tracked as long as

the process is transparent and parameters are in place for the use
of data53—e.g., data is used solely for the purpose gathered, not
sold, secured, and anonymized. Annual or bi-annual public safety
audits could be completed by key public-sector entities to ensure
parameters remain in place and are having the intended effect.

14 | SEEING WHAT MATTERS

The 2018 GDPR legislation in Europe could serve as an initial
model for governing the public safety ecosystem.43 GDPR rules
specify that individuals “can contest ‘legal or similarly significant’
decisions made by algorithms and appeal for human intervention,”
thereby getting an explanation of how an algorithm generated a
particular outcome.54 This could easily be applied to how public
safety data is handled.

There’s no doubt people overwhelmingly want to feel safe and

enjoy a high quality of life. Cities and countries can better equip
themselves to fulfill those wishes by using more public safety data,
and more-powerful tools to make sense of that data, to prevent
crimes of all types before they happen.

15 | SEEING WHAT MATTERS

APPENDIX A:
THE VALUE OF AI IN PUBLIC SAFETY

ASSUMPTIONS

Violent Crime Seen on Based on “circumstances” analysis of crime, we estimate

Camera Percentage that currently 1% of crime is seen on camera, 41% with
AI in 2025, and 74% with AI in 2035.

Crime Identified by Utilized startups algorithm accuracy rate. For current,

AI Percentage assumed 2% of crime is identified given 2% of video
footage is seen.

Incidents Where For given driver, excluded incidents above average

Response Time Greater response time as such incidents are currently accounted
than Average Percentage for with today’s public safety ecosystem.

Additional Impact Quantified impact of AI by ecosystem’s ability to reduce

by AI Percentage response time and thereby reduce crime

Loss Value to Society The summation of direct incarceration cost per inmate
by Inmates Imprisoned and loss in GDP per inmate

16 | SEEING WHAT MATTERS

APPENDIX A:
THE VALUE OF AI IN PUBLIC SAFETY

VALUATION APPROACH

Lives That May Be Saved   Number of students eligible for graduation

with AI Surveillance x % of victims murdered in public
x % of violent crimes identified by AI
x % of incidents where response time > average
x Expected % of additional lives saved by AI
  Lives that may be saved by AI-surveillance

Victims Shielded from   Mass shooting victims

Mass Shootings x % of victims murdered in public
x % of violent crimes identified by AI
x % of incidents where response time > average
x Expected % of additional victims AI may shield
  Victims that may be shielded by AI-surveillance

Reduction in Violent Crimes   Number of violent and property crime victims

from AI-Surveillance x % of violent/property crime in public
x % of violent/property crimes identified by AI
x % of incidents where response time > average
x Expected % of additional victims prevented by AI
  Decreased victims by AI-surveillance

Economic Impact from   Prisoners serving for “Property” crimes

AI in Law Enforcement x Time served in prison
x Loss Values to Society by Inmates Imprisoned
  Economic Impact of Inmates Getting Released to Parole

17 | SEEING WHAT MATTERS

AUTHORS

WESTERN DIGITAL CORPORATION ACCENTURE STRATEGY

David Tang Esther Colwill
SVP, Corporate Marketing Managing Director, Communications,
[email protected] Media, & Technology
[email protected]
Judy Fujii-Hwang
Corporate Brand and Thought Leadership Shobit Arora
[email protected] Senior Manager, CFO and Enterprise Value
[email protected]
Kris Rangarajan
Senior Manager, Corporate Marketing Alvaro Mendizabal
and Social Media Senior Strategy Consultant,
[email protected] Communications, Media & Technology
[email protected]

Alireza Safi
Senior Strategy Consultant, Advanced
Customer Strategy
[email protected]

Lincoln Lam
Strategy Consultant, Competitiveness
Center of Excellence
[email protected]

Fabio Jaime
Strategy Analyst
[email protected]

18 | SEEING WHAT MATTERS

REFERENCES
1 The Economist: http://safecities.economist.com/ 19 The Guardian: https://www.theguardian.com/
safe-cities-index-2017 world/2017/sep/01/facial-recognition-china-beer-
festival
2 Quinnipiac University: https://poll.qu.edu/
national/release-detail?ReleaseID=2521 20 Forbes: https://www.forbes.com/sites/
niallmccarthy/2017/08/31/private-security-
3 The New York Times: https://www.nytimes.
outnumbers-the-police-in-most-countries-
com/2018/02/15/learning/what-is-your-reaction-to-
worldwide-infographic/#41918644210f
the-deadly-shooting-at-a-florida-high-school.html
21 United States Census Bureau: https://www.
4 Faceter: https://tokensale.faceter.io/Faceter_
census.gov/popclock
Whitepaper_en.pdf
22 The Verge: https://www.theverge.
5 The Next Web: https://thenextweb.com/
com/2018/4/11/17223504/ai-startup-sensetime-
cryptocurrency/2018/03/26/this-blockchain-based-
china-most-valuable-facial-recognition-surveillance
surveillance-startup-detects-crime-in-real-time/
23 IBID
6 Justice Quarterly: https://www.tandfonline.com/
doi/abs/10.1080/07418820802506206 24 IBID
7 Acme Security Systems: http://www.acmesecurity. 25 The Wall Street Journal: https://www.wsj.
com/closed-circuit-television-cctv.asp com/articles/artificial-intelligence-could-
soon-enhance-real-time-police-surveillance-
8 Homeland Security News Wire: http://www.
1522761813?mod=searchresults&page=1&pos=2
homelandsecuritynewswire.com/study-shows-
surveillance-cameras-reduce-crime-some-cases 26 Global Times: http://www.globaltimes.cn/
content/1095176.shtml
9 IBID
27 Investopedia: https://www.investopedia.com/
10 The Brookings Institution: https://www.brookings.
terms/d/deep-learning.asp
edu/wp-content/uploads/2017/10/safe-city-
innovation_final.pdf 28 Kairos: https://www.kairos.com/blog/the-best-
explanation-machine-learning-vs-deep-learning
11 Accenture: https://www.accenture.com/
t20150523T055755__w__/us-en/_acnmedia/ 29 DeepScience AI: http://deepscience.ai/index.html
Accenture/Conversion-Assets/DotCom/
30 TechCrunch: https://techcrunch.com/2017/05/16/
Documents/Global/PDF/Technology_2/Accenture-
deep-science-ai-monitors-security-feeds-for-
Video-Analytics-Operational-Marketing-and-
masks-and-guns-to-quicken-response-times
Security-Insights-from-CCTV.pdf
31 Gartner: https://www.gartner.com/
12 Gartner: https://www.gartner.com/it-glossary/
document/3873805/meter/charge
dark-data
32 Internet of Things Institute: http://www.ioti.com/
13 Techopedia: https://www.techopedia.com/
security/how-smart-building-technology-could-
definition/29373/dark-data
address-mass-shootings
14 Faceter: https://tokensale.faceter.io/Faceter_
33 Pew Research Center: http://www.pewresearch.
Whitepaper_en.pdf
org/fact-tank/2018/05/02/americas-
15 The Wall Street Journal: https://www.wsj.com/ incarceration-rate-is-at-a-two-decade-low
articles/the-all-seeing-surveillance-state-feared-
34 US Committee for a Responsible Federal Budget:
in-the-west-is-a-reality-in-china-1498493020
http://www.crfb.org/blogs/us-spends-80-billion-
16 Cisco: https://www.cisco.com/c/en/us/solutions/ year-incarceration
collateral/service-provider/visual-networking-
35 ABC News: http://www.abc.net.au/news/2017-08-
index-vni/vni-hyperconnectivity-wp.html
14/how-ai-could-put-an-end-to-prisons-as-we-
17 Statista: https://www.statista.com/ know-them/8794910
statistics/228880/people-living-in-households-
36 Cornell University: https://www.cs.cornell.edu/
that-own-a-home-security-system-usa
home/kleinber/w23180.pdf
18 Knightscope: https://www.knightscope.com

19 | SEEING WHAT MATTERS

37 Squire Patton Boggs: https://www. pdf
securityprivacybytes.com/2018/02/the-gdprs-
53 The Huffington Post: https://www.huffingtonpost.
impact-on-cctv-and-workplace-surveillance/
co.uk/entry/surveillance-cities_uk_5a2db64fe4b0
38 IDC: https://www.idc.com/getdoc. 22ec613b83dd?guccounter=2
jsp?containerId=US42002217
54 The Brookings Institution: https://www.brookings.
39 Technode: https://technode.com/2017/11/22/ edu/research/how-artificial-intelligence-is-
china-to-have-626-million-surveillance-cameras- transforming-the-world/
within-3-years/
55 Western Digital: http://datamakespossible.com/
40 IHS Markit: https://www.lto.org/wp-content/
56 Accenture: https://www.accenture.com/us-en/
uploads/2014/06/IHS-Markit-Whitepaper-Caught-
insight-cctv-public-safety
on-Tape-Keep-it-Secure-highres.pdf
57 Accenture: http://voicesfrompublicservice.
41 IHS Markit: https://business.panasonic.com.
accenture.com/unitedkingdom/ai-opens-up-
au/security-solutions/sites/default/pssap_files/
ways-to-see-the-unseeable-in-public-safety-
security-solutions/node_file_uploads/Video%20
provided-its-used-responsibly?lang=en_GB
Surveillance%20Trends%20for%202018%20-%20
IHS%20Markit.pdf
42 IT Pro: http://www.itpro.co.uk/big-data/30541/
smart-cities-to-become-the-norm-by-2025
43 HIS Markit: http://news.ihsmarkit.com/press-
release/design-supply-chain-media/smart-cities-
rise-fourfold-number-2013-2025
44 Gartner: https://www.gartner.com/
document/3871931/meter/charge
45 The Wall Street Journal: https://www.wsj.com/
articles/apple-to-push-fix-for-locating-911-
callers-1529314200
46 Infineon: https://www.infineon.com/cms/en/
discoveries/internet-of-things-2030/
47 The Wall Street Journal: https://www.wsj.com/
articles/the-all-seeing-surveillance-state-feared-
in-the-west-is-a-reality-in-china-1498493020
48 Everytown Research: https://everytownresearch.
org/reports/mass-shootings-analysis/
49 Genetec: https://www.genetec.com/solutions/
industries/cities
50 US Federal Bureau of Investigation (FBI): https://
ucr.fbi.gov/crime-in-the-u.s/2016/crime-in-the-
u.s.-2016/topic-pages/offenses-known-to-law-
enforcement
51 Blueline Security Services: http://www.
bluelinesecurityservices.com/how-video-
surveillance-cameras-can-prevent-crime-before-
it-happens/
52 The Constitution Project: https://www.law.
berkeley.edu/files/Video_surveillance_guidelines.

20 | SEEING WHAT MATTERS

LEARN MORE ABOUT ACCENTURE
To learn more about the Value of Data, visit: Accenture is a leading global professional services company,
Data Makes Possible 55
providing a broad range of services and solutions in strategy,
consulting, digital, technology and operations. Combining
To learn more about disruptive applications of VR/AR, visit: unmatched experience and specialized skills across more
Accenture Strategy’s Latest Thinking56 than 40 industries and all business functions—underpinned
by the world’s largest delivery network—Accenture works at
To learn more about how to value your data, visit the intersection of business and technology to help clients
Accenture Strategy57 improve their performance and create sustainable value
for their stakeholders. With approximately 442,000 people
serving clients in more than 120 countries, Accenture drives
innovation to improve the way the world works and lives.
Visit us at www.accenture.com.

Copyright ©2018 Western Digital and Accenture. All rights reserved.   181055