The implications of analytics

March 10, 2017
A conversation with Hitachi’s Justin Bean

The growth numbers are staggering. According to several research surveys, the overall public safety and global security market size is estimated to see growth of $247.53 billion this past year to more than $456.56 billion by 2021; at a healthy Compound Annual Growth Rate (CAGR) of 13.0 percent. By contrast, the video surveillance market is expected to increase to $40.8 billion by 2020 at a CAGR of 15.86 percent. However, the challenges to this growth windfall are even more compelling when one considers concerns surrounding intelligent data consumption and privacy.

Most video technology vendors realize that just capturing the images is not enough. The push to add actionable intelligence and forensics to video generated by this army of surveillance cameras has reached critical mass as almost every vertical market from municipalities with expanding video systems to college campuses, hospitals, airports and retail centers are insisting on reliable analytics their security and risk personnel can use.

Last week, Hitachi unveiled  new advancements to its Smart City and Public Safety portfolio that applies high-end analytics to video data to deliver a range of features designed to unleash its full potential, including activity and traffic analysis, people counting, queue detection, license plate, facial and object recognition, video enhancement and stabilization, camera health alerts and privacy protection.

I reached out to Justin Bean of Hitachi this week to have him share some of his company’s vision and its technology roadmap involved in the launch of its new offerings in their video management suite. Bean is the Director of Hitachi’s Smart City Solutions Marketing group.

Lasky: Can you tell us what the solution concepts are with your video analytics addition?

Bean: This [video analytics solution] complements Hitachi's Video Management Platform (VMP), which is a converged storage and management solution for video data. The Visualization Suite is now bringing together video, IoT, and analytics on a single pane of glass. With the video analytics options, we are now adding a lot more value to the data being offered. We’re able to pull into that system and offer a lot more data and alerts to serve a broader market.

For a long time, we have been able to see video as a source of images and eyes on the street if you will. It allows us to see what is happening. But with the advent of things like computer vision and machine learning, we are turning this video data into insightful and actionable data.  From the public safety and security side, you have intrusion alerts and object detection added to business intelligence where we are counting people and analyzing flow and ques, and also transportation and traffic where we are able to not only count cars and trucks but actually discern the difference between these various types of vehicles. There are a variety of different use cases we can pull out of this solution.

Today, with a lot of IoT sensors, you get only one type of data for each sensor deployed versus that of a video camera where you can pull out a wealth of different insights from the same sensor. With that, I think it is inevitable we will see a lot more video providing mission critical data and bringing a high degree of value to the users and municipalities that have access. I recently read there were more than 240 million video surveillance cameras deployed around the world, so if we were able to convert that visual data into insights, alerts and IoT data like we’ve been talking about here, we would have such a broader base of knowledge to pull from, enabling us to keep people safer, make smarter business decisions and shorten the time to emergencies and prevent crime.

Lasky: What defines this market and makes your approach a little more unique?

Bean: The goal for us is having a complete solution for our varied customers; a robust portfolio of video analytics that can provide business intelligence, operational intelligence for security and the public safety side of it as well. We are going about this in a bit of a different way than a lot of companies out there by offering a very strong privacy option to mitigate concerns that come with deploying these type of sophisticated video systems. 

This is a rapidly expanding global market that spans a wide variety of industries, with close to a 20 percent CAGR globally and a market that is expecting substantial growth in the next four years. We have a wide array of customers today where analytics can be an added solution for them, and of course, our sales force is reaching out to new clients generating new projects. So the expanding market and our new solutions offer a lot of opportunities for Hitachi. The things that are driving this industry are basically the innovations in thinking about how to use analytics. It has primarily been used for public safety and security and the challenges it faces. But with the business intelligence capabilities, you are seeing an expanded market beyond these two traditional markets.

Lasky: What separates the Hitachi analytics technology solution from the competition?

Bean: A lot of companies are out there are doing pixel analysis. With some other solution, you will have a few pixels move which creates a trip wire effect that can set off an alert. This leads to a lot of false positives. Instead, we are doing object detection where we detect particles, not pixels. We can see where these pixels come together then we can group them with other particles to see where they’re moving. The intelligence behind the vision determines that this is one object moving on the screen. This is a huge step forward in being able to improve accuracy and reduce all those false-positives that have plagued the industry in the past.

Another industry issue has been the reuse of code. Some providers will use the same analytics they employ to count people and also use it count vehicles. It is not really the same thing, so that again leads to more false-positives and lower accuracy.  Instead, we are using scenario-based code that is written specifically for that application, which improves accuracy. Snow and rain can also set off alerts and disrupt detection, so until we can control the weather, we provide video enhancement that can mask that snow and rain out of the output video. This allows the human beings looking at the pictures to have a much cleaner view, and allows the analytics that are running on top of it to have clean feed to help with accuracy.

Lasky: You mentioned that privacy is a critical concern for many of your market clients, so just how are you managing those concerns?

Bean: One of the major constraints with any video solution is the public’s concerns about privacy. When people see security cameras going up in their downtown or in a public space, it raises various questions. But we are helping to solve these with our privacy solutions. However, there are challenges in the market when it comes to technology like video analytics, where promises have been made in the past with regards to its functionality that didn’t quite live up to expectations.  We understand those challenges and have strived to bring to market a solution with a lot more accuracy and a reduction in the false/positives.

The predominant solutions out there now are simply detecting a face and then pixelating that image. But you can argue there are a lot of characteristics beyond the face that can identify a person – their clothing, a tattoo, anything like that which can present a real problem. Hitachi is offering a full privacy protection package that pixelates the entire body and then masks them with a color shield. The transparency on the backend is having to log in and have your activities tracked on the system.

So, as Hitachi keeps moving forward, we continue to add value for our video, security and public safety stack. As the data becomes more and more important, being able to layer in solutions with the video products and systems we have creates a strong partnership between our customers and our business goals. 

Lasky: Can you offer us an example of a case use for this technology in a public safety environment?

Bean: Let’s imagine we are responsible for the security of an airport. We have a suspicious object that has been detected and we don’t know if it has been planted or just left behind.  If we do determine there is a danger we can look back at the video feeds and see the suspect who dropped it off. We can then do live face matching by searching the video surveillance system and see where the person was most recently detected by the security system.

And once they are apprehended, we can take a look back in time to do a similar face search to find the first instance the system detected this person. Now we can see that person being dropped off in the parking lot by a car and we can do license plate recognition on the car. If we are connected to the city surveillance system, we can search those traffic cameras that are at the intersections and run license plate recognition there as well to see where this car was going and where it has been, making it easier to apprehend any other suspects. This is a strong force multiplier for the police but also is a great tool for proactive policing.

Lasky: In that same airport setting, are there other possible applications tied to this analytic solution?

Bean: The que detector is more of an operational and business intelligence use that can also help the client. We can see when a line gets too long enabling them to open up another counter, making sure the customer experience is a good one as they move through the airport. This is a very good environment to highlight the three most important value pillars of video analytics. Of course we want to keep people safe in the airport. The airlines want to get their passengers through the system as quickly as possible and then once they get into the terminal it becomes a retail experience – almost like a shopping mall. Analytics can offer a variety of business insights for retail in that terminal so the airport has the data intelligence to make smarter business decisions.

Lasky: So just how can these analytics transfer to the retail space?

Bean: From a business perspective, we have had a sophisticated approach when using analytics on our websites to see where people are coming from, what pages they land on, how long they stay on that page, which links they click on, then taking all that and analyzing the conversion rates. We’ve done a good job using this digitally generated information to make our website businesses more efficient and profitable. Now, with video analytics, we are able to bring these insights into the real world like a downtown retail area, a shopping mall – any of these areas where we can track the people, see where they spend the most time, how they flow through the stores, what products they interact with – all these things that were previously just limited to the digital world. But now we can bring this into our physical brick and mortar store for intelligence. That same system being used for loss prevention and security by most retailers can now have analytics sitting on top of it providing really valuable business insights.

Lasky: Can the implementation of analytics also provide any ROI benefits?

Bean: Let’s go back to public safety. Parking is generally the second or third largest revenue source for most cities, so there is a very strong ROI factor they can get out of using video analytics to not only help in parking traffic flow but also in parking enforcement.

When we start looking at things like this from a Smart City perspective, these are the types of projects cities need to consider in order to kick-start their Smart City initiatives. So if you can implement a solution that generates a clear ROI where it can pay for itself, or is at least prove to be revenue neutral, it solves the problem for the residents as well as the city from an operational standpoint.  And then if we have a common system where we are bringing all that data together like the Hitachi Visualization Suite, you can bill that out to other organizations around the city who can’t yet afford to buy their own IoT platform or system. This is a great place for public safety to start and then expand out to transportation and sustainability.

About the Author: Steve Lasky is a 30-year veteran of the security industry and is the Editorial Director for SouthComm Security Media and the editor-in-chief of Security Technology Executive magazine. You can reach him at [email protected].