301.519.9237 exdirector@nesaus.org

3.25.19 – SIW –

Milestone’s Jeremy Scott shares his insights on how predictive analytics has emerged as a key part of any risk mitigation strategy

The fact that predictive analytics has been a technology practiced by many organizations for years doesn’t necessarily mean it is has had strong roots in the security industry. But as more and more enterprise organizations turn to predictive analytics to help them solve problems and become proactive in everything from cybersecurity to video surveillance, security solutions providers are vigorously looking for ways to implement it.

Security Technology Executive magazine’s editorial director Steve Lasky posed several questions to Jeremy Scott, Strategic Alliances Program Manager of the Americas at Milestone Systems related to evolving world of predictive analytics and how it is being deployed in the security industry.

STE: Please define “predictive analytics” as it relates to video security and surveillance?

Scott: To define predictive analytics as it relates to video security is almost situational because it’s a very early science and is a subset of Artificial Intelligence technology. There’s also a real misunderstanding of Artificial Intelligence as it relates to security. Often when people think of AI, they think of a machine doing the thinking, and that’s true to a point. But there are two thoughts behind Artificial Intelligence: creating intelligent software algorithms for analyzing data, and then there are uses where the machine is actually doing the “thinking” and decision-making. Predictive analytics is somewhere between these two approaches.

A simple example of a basic predictive analytic would be in analyzing an object’s direction of travel. Once software detects a moving object, its direction of motion and speed of travel, it can predict where that object is going to end up. This level of AI can be applied to predictions of traffic congestion, maritime surveillance, suspects fleeing a scene, etc. When we start thinking about the potential, predictive tools can become quite varied and very useful. Computers notice patterns and can process large amounts of data much faster than we can.

STE: Are there Milestone partners currently in this space? Emerging?

Scott: Milestone has many partners in video analytics. Snap, for example, is self-learning software that learns the cameras on the network and their associations to one another, enabling users to easily track subjects from one camera scene to the next. Ipsotek’s Visual Intelligence Suite can analyze recorded video feeds in real time and serve as an intelligent first line of response by notifying an operator when certain behaviors are observed. There is a wide range of video analytics behaviors available now and even more under development. As these technologies advance, we’ll be able to take many video analytic behaviors and morph them into predictive analytics.

STE: Where might predictive analytics best serve the security executives and integrators and/or video applications beyond security?

Predictive analytics uses can range from city surveillance and infrastructure monitoring to smart cities and large campuses and facilities where vehicle and pedestrian traffic patterns, crowd size and movements, and other events are worth noticing and monitoring.

Milestone Solution Partner BriefCam currently offers real-time analysis, for example, in a retail environment to track customer flow and browsing patterns. There’s a chain of grocery stores that use video to monitor customers as they enter the store. They use the analytics to see if customers most often visit as a single person grabbing a shopping cart, or if a group of people walk in together, and if it’s a group do they use more than one shopping cart? Store managers use this information to predict customer checkout flow, so they can predict how many cash registers need to be open 30 minutes from any given point during the day. There’s no reason that software can’t make these predictions. This strategy could be used for all types of automatic, predictive queuing management.

STE: The pace at which video surveillance technology has advanced over the last 3-5 years is significant and shows no signs of slowing. What is the current landscape with predictive analytics and where do you see advancement opportunities over the next 3-5 years?

Scott: The advancements in the last three to five years have been significant. A good example is in how we’re now using GPU processing over CPU processing for very fast, tandem processing of mathematical equations, analytics, and algorithms at a much higher rate. As GPU’s become more prevalent, more and more analytics will base their processing requirements on GPU processing, which dramatically increases bandwidth and speed, leading to effective, predictive analytics.

Another example along the same line is with how Milestone has been developing our software to process metadata. Metadata is the data that’s derived from a device’s output. For example, there might be temperature data, data regarding people movement, or a wealth of other data that are being derived from the original information from a camera or other device. To date, the industry has not been doing much with this data because we couldn’t process it efficiently. These types of advances are multiplying the usefulness of video data exponentially.

STE: In a city surveillance situation, how might technology advances lead to but also prevent privacy issues and profiling and things of that nature?

Scott: Privacy is a top priority for most developers and users. Think about analytics within a healthcare environment. There are HIPAA regulations in healthcare where people aren’t allowed any patient medical information unless they have explicit permission. And video data is full of personal identifying information. How can advanced video analytics systems have access to the data needed, while at the same time securely respect and protect patient information? The same issues come up with use in schools where minors are the primary population.

Those are very real concerns, and yes, there’s technology that exists to mitigate these issues. A good example is that we have a feature in our software called Screen Recorder.

In a city surveillance situation where there are cameras all around a city and we have strict rules on surveillance usage, we can deploy our Screen Recorder on workstations along with a camera that views and records the person who’s using the investigative computer. This way we have a record of exactly who is operating the system and what they’re doing. Screen Recorder captures everything, and the evidence can’t be altered in any way. So, now, we’re using technology to monitor the people who are using the cameras and making sure they are in complete compliance. We see surveillance legislation, technologies and best practices advancing together.

STE: How can an open platform community facilitate increased collaboration between private and public partnerships? What are some of the key advantages that can result from these partnerships?

Scott: An open platform community is based on the premise that it’s not just a few individuals at a company that comes up with solutions. An open community builds upon the collective good ideas of the industry, so it can grow at a faster rate than any one company can sustain alone. Active collaboration between technology developers and users, as well as private and public organizations, build a comprehensive community.

If you look at the city of Las Vegas, for instance, there are more cameras in a two-mile stretch of Las Vegas Boulevard than probably any other place in the United States, most of which are installed on private property. With so much tourism in Las Vegas, there needs to be solid collaboration between public and private concerns to ensure public safety and security. There is computer dispatch software available today that integrates with our open-platform VMS to enable better response times for EMS, fire, and other law enforcement agencies — an excellent example of open-platform technology enabling public and private entities to work together.

STE: What are important criteria or prerequisites that security executives and other business decision makers should know when considering these types of solutions?

Scott: In such a connected, networked environment, technology is no longer a unilateral decision, especially when it comes to security. Video systems today involve and benefit all aspects of an organization, from Operations and Maintenance to Finance and Human Resources — especially for organizations within GDPR-impacted countries — there are huge implications for the handling of private information.

It’s critical that security executives work with a range of internal and external partners who can provide insights from all points of view. When we talk about the community, we’re referring to hardware and software developers as well as the channel partners who design the solutions, perform the installations, and provide the user training and support the end users.

There are people who stand behind the technology all along the design and deployment path, and their experience and knowledge are very valuable. It’s important to work with consultants, integrators, and subject matter experts to deploy secure and effective systems that meet the organization’s goals.