Take a moment to imagine what the future could hold: It is 2:00 a.m. in Modesto, California. Among hundreds of video feeds, filtering into the Modesto Police Department’s real-time crime center, one camera detects activity that is deemed abnormal by the software analytics. Not a human is in sight, but the intersection-mounted camera observes a vehicle without its headlights on enter the parking lot adjacent to an electronics store and park. As the burglar, dressed in dark clothing, approaches the side door of the business, the software sends an anomaly alert to the analyst staffing the real-time crime center. The system-generated alert was accompanied by a very short video clip of the events leading to the alert. Within 30 seconds, the analyst concludes that he or she is watching an in-progress burglary and dispatches police units. Within minutes, officers surround the store and eventually arrest the man who had been emboldened by a series of successful burglaries in the past. Even the occasional early morning motorist wouldn’t think to take notice of the events preceding the attempted burglary, but the video analytics software did.
Almost daily, social media or traditional media highlights a serious crime for which the suspects are still on the loose. Those determined to victimize the helpless and vulnerable used to hide in the shadows and play the odds, which too often landed in their favor. Some would say the stupid criminals get caught, but the reality is that law enforcement is increasingly outnumbered in the quest for law and order. However, policing now has, at its fingertips, an array of innovative, high-tech crime-fighting tools to tilt the odds in its favor. These tools could potentially reduce crime significantly and make communities safer. One of these emerging tools is intelligent video analytics, which can be used to assist in the investigation of crimes captured on video.
The Rise of Intelligent Video Analytics
Digital video is everywhere. High-quality video cameras are mounted on government buildings, intersections, schools, private businesses, and homes. These traditional sources of video are now supplemented with wireless and smartphone cameras capable of high-resolution live-streaming capabilities. Various non-law enforcement video feeds currently supplement police in-car video and, more recently, body-worn camera video. Though many police agencies are using video daily to investigate crimes after they occur, the exploration of current and future intelligent video analytics technology in the prevention of crime is only just emerging.
Several agencies, though, such as the Phoenix, Arizona, Police Department, are working to establish partnerships with their communities to identify and map private surveillance camera networks to allow for more efficient and timely law enforcement access to archived video footage that they believe could contain crime-relevant footage.1
The most forward thinking of law enforcement agencies have taken it a step further. In the Paterson, New Jersey, Police Department, law enforcement leaders have started the process of linking private video feeds to allow their personnel to view live feeds of connected cameras. Such public-private partnerships require significant trust between the police agencies and the residents they protect, and the benefits of real-time video feeds throughout the city allows for more timely and better-informed investigations.2 Paterson police have a valuable crime-fighting asset at their fingertips and are an example of the ways intelligent video analytics could bolster crime-fighting efforts by implementing intelligent video analytics software to allow for nearly instantaneous evidential review.
Harnessing this valuable evidence and efficiently analyzing data are where video analytics enters the equation. Vendors are touting sophisticated software that can detect anomalous behavior based upon computer-based artificially intelligent analytics of previously learned behaviors.3 In simple terms, the software will be able to determine what is out of place or abnormal, based upon watching and learning; a process known as machine learning. The analytics technology is complex and growing in sophistication, with at least one provider claiming that its software can monitor thousands of video feeds and learn what is “normal” and what is not, thus allowing personnel to be alerted in real-time to events or behaviors that are out of the ordinary.4
Catching Crime As It Occurs—Or Earlier!
Video evidence, though valuable, can be cumbersome and time consuming to review after the commission of a crime. Fortunately, tools that allow personnel to more efficiently review large quantities of video footage by looking or by user-designated criteria, such as the type of car or the color of the car, are already a reality. By eliminating footage that doesn’t contain desired criteria, the task becomes more manageable, but it still requires human attention.5 Intelligent video analytics can continuously learn; therefore, even when its human counterparts are completing other tasks, taking vacations, or simply looking the other way, the anomalous incident alert automation could trigger the response of personnel to problematic situations or behaviors.6
In a 2017 interview, Research Director Nick Ingelbrecht of the technology consultancy firm Gartner told Government Technology, “We project that by 2020, 95 percent of video or image content will never be viewed by humans, but instead will be vetted by machines that provide some degree of automated analysis.”7 As access to diverse video sources increases, a significant opportunity exists if machine learning intelligent analytics software lives up to all of the hype. For example, a local school district could link its schools’ high-definition camera networks to the police agency. Instead of waiting for an alarm activation to be routed through a third-party monitoring company, as is the current practice in most areas, the simple presence of a person near a locked door at 1:00 a.m. could trigger an alert directly to the police, before entry was even attempted. In this future equation, the criminals stand to lose their competitive advantage (time) and will realize that they could be detected, in real time, as they embark upon their illegal deeds.
On a larger scale, strategically placed cameras throughout a city could provide instant alerts for a wide array of circumstances the software deems out of place based on its continuous learning of what is normal and what is not. By analyzing each frame of video and converting it to metadata, the software is able to conduct real-time image analysis.8 While some of the most intriguing applications come in the form of detecting crime as or before it occurs, intelligent video analytics, when coupled with existing and future camera networks, could also provide instantaneous alerts for non-criminal acts such as traffic collisions and fire. The possibilities are likely limited only by the resolution and coverage of the cameras.
Another vital element of making this technological advance possible will be the integration of video analytics into real-time crime centers. The St. Louis, Missouri, Police Department has established such facilities, and the Modesto Police Department is investing to bring all technological and statistical crime-fighting tools into one centralized location.9 These centers, chock full of flat-screen monitors showing actionable analytical data, will be the hubs for the video and analytics software technology, and the well-trained non-sworn personnel staffing these regional centers will provide the information needed to better direct officers. It is anticipated that the next several years will prove this analytics technology to be a tremendous force multiplier for law enforcement agencies struggling with staffing challenges.
In a personal interview with the author in November 2017, Modesto Police Chief Galen Carroll said, “Many agencies are finding it more and more difficult to fill their police officer ranks. Video analytics looks to be a valuable tool that can maximize and leverage the existing resources of our department.”10 To put it simply, the software can do the heavy lifting, alerting real-time crime center staff only when it detects something out of the ordinary. The present-day delays of cellphone 911 calls and the lag time needed to dispatch proper resources would be reduced. Other forms of suspicious anomalous behavior detected by machine learning analytics technology would boost response times to in-progress crimes and increase the chances of bringing the culprits to justice.
Facial recognition implementation and integration could be another important enhancement to intelligent video–enabled networks.11 Video networks comparing those within camera view to wanted person databases could increase those brought to justice for serious crimes. This concept would, by its nature, demand debate, and the extent of the technology’s usage would have to be carefully evaluated and discussed in advance to ensure adherence to a particular country’s laws and constitutional principles. Even utilizing this technology to identify only the worst-of-the-worst, such as murderers and rapists, would greatly enhance community safety and the efficiency of the justice systems. Likewise, it would instill fear in the minds of those who prey upon the innocent and perhaps deter them from committing further offenses.
Hurdles to Implementation
The cost of intelligent video analytics technology is likely to be substantial and could slow the implementation in some jurisdictions. A U.K. police agency implemented the technology, at a cost of £250,000 (approximately $354,500 or €286,900). The software allowed police there to comb through video footage to search for specific faces or clothing and even limit video to instances where a person was riding a bicycle.12 It is difficult to determine the overall effects of video analytics on crime rates, but the addition of technology might affect the decision-making processes (risk vs. profit) of some criminals, thus, affecting the overall crime rate in a jurisdiction. Benefits to the communities will come in the form of higher crime clearance rates and a greater overall feeling of safety. Lessons can also be learned from early adopters of intelligent video analytics technology, such as the Washington, DC, region’s Metro Transit Police Department, which uses the technology to fight crime on the city’s trains. Though crime continues, the arrests by the department are mounting—as of 2015, there were 500 adult arrests and an additional 300 cases that came to resolution due to video analytics.13
Privacy concerns could also be a hurdle to implementation, though the general acceptance of cameras rose to 78 percent in a survey of U.S. citizens taken in the weeks following the Boston Marathon bombing.14 It will be important to involve the community leaders, the public, and allied agencies in discussions surrounding the implementation of intelligent video analytics technology. In the end, the public will need to support the overall policing philosophy of their jurisdiction for it to succeed, but external factors such as regional, national, and international events will influence the overall acceptance of the technology.
Terrorism and mass casualty incidents may play the largest role. Surveys taken after the 9/11 terror attacks have illustrated the increase in acceptance of routine video surveillance. Even without terrorism, mass casualty incidents are tragically increasing. The rate nearly tripled from an average of 6.4 mass casualty shootings between 2000 and 2006, to 16.4 between 2007 and 2013.15 San Bernardino, California; Sandy Hook, Connecticut; Las Vegas, Nevada; Aurora, Colorado; Orlando, Florida—all now known for mass casualty incidents of senseless violence. In the United Kingdom, public surveillance once was a controversial topic, but terrorism has played an integral role in reducing debate on the privacy concerns of that country’s very robust CCTV camera network.
Public surveillance was just one security measure that grew after several notable attacks, including the 2004 Madrid, Spain, and 2005 London, England, train bombings.16 Intelligent video analytics advancements might not be the panacea for averting these tragedies, but they can certainly help during the analysis of the growing number of video feeds available to police agencies. Today’s world is incredibly unpredictable, and it is commonly acknowledged that domestic terrorist attacks and other mass casualty incidents are unfortunately no longer an “if” proposition, but a “when.” The higher risk environment caused by these tragedies might increase the public’s acceptance of increased video surveillance efforts by local, state, and federal jurisdictions.
Conclusion
Intelligent video analytics in police work may be a key tool in the future of law enforcement. Using high-tech tools to fight crime is becoming more prevalent, but agencies struggle with staffing and budget constraints while trying to keep up with the technology. Machine learning video analytics, powered by artificial intelligence, could be an important development in law enforcement in the coming years.
The change is happening now. Agencies such as the Metro Transit Police Department have successfully implemented intelligent video analytics to protect the traveling public in the U.S. capital region, and other departments are actively pursuing public-private video feed sharing. The key over the next few years will be putting all of the pieces together: cooperative and shared video feeds, intelligent video analytics software, and real-time crime centers to centralize the technology. Advancements in machine learning video analytics and the rapid increase of available video feeds seem to indicate that this technology will dramatically change the landscape of law enforcement by the middle of the next decade. Crime will always exist, but criminals will quickly learn that their actions are increasingly likely to be detected, and law enforcement will have a new, proactive way to prevent crimes by those who are not deterred.
Notes:
1 Caroline Liddle, “Phoenix Police Hope to Gain Extra Eyes on Crime through Private CCTV Security Cameras,” AZ Central.com, February 16, 2017.
2 Matt Fagan, “Police Learn to Use Private Security Cameras as Their Eyes and Ears,” NorthJersey.com, July 22, 2017.
3 Bosch Security Systems, Focus Your Attention: Bosch Intelligent Video Analysis, 2008.
4 Bosch Security Systems, Focus Your Attention.
5 Wylie Wong, “Louisville Deploys Cutting-Edge Data Analytics and Video Surveillance Technology,” StateTech, January 20, 2018.
6 Rick Delgado, “AI Is Tapping Into Intelligent Video Analytics and Is Transforming Security,” Tripwire, September 18, 2016.
7 Eyragon Eidem, “Video Analytics Is Ripe for Automation, But Is Government Ready?” Government Technology, March 2017.
8 Bosch Security Systems, Focus Your Attention.
9 “St. Louis Police Launch Real-Time Crime Center,” Government Technology, April 17, 2015.
10 Galen Carroll (chief, Modesto Police Department), personal interview, November 2017.
11 Delgado, “AI Is Tapping Into Intelligent Video Analytics and Is Transforming Security.”
12 James Temperton, “One Nation under CCTV: The Future of Automated Surveillance,” Wired, August 17, 2015.
13 Max Smith, “WMATA Video Center Gets Thousands of Requests, Leads to Hundreds of Arrests,” WTOP, December 26, 2015.
14 Mark Landler and Dalia Sussman, “Poll Finds Strong Acceptance for Public Surveillance,” New York Times, April 30, 2013.
15 Bosch Security Systems, Focus Your Attention.; Michael S. Schmidt, “F.B.I. Confirms a Sharp Rise in Mass Shootings Since 2000,” New York Times, September 24, 2014.
16 Jon Lee Anderson, “After Manchester, The U.K. Weighs Security and Freedoms,” The New Yorker, May 30, 2017.
