Who's Up For Analytics?
- By Ralph C. Jensen
- Aug 01, 2010
Frankly, I’ve always been a little skeptical of analytics. I’m not a naysayer, but I’m someone who has to see, touch and smell something before I’m a true believer.
To become a true believer, I decided a little research was in order.
Earlier this year, Aimetis, based in Ontario, Canada, released their people counter, which tracks the bidirectional flow of objects as they pass a boundary defined by the end user. This technology is ideal for retail environments, schools, banks, recreational facilities and airports. Its purpose is to gain intelligent customer and operations insight despite constraints on budgets, bandwidth and IT infrastructure support.
According to Marc Holtenhoff, CEO of Aimetis Corp., the solution is ideal for both small and large camera deployments, and it provides an accurate stand-alone solution for leveraging video data for operational intelligence.
“The AXIS camera application platform was launched so our application development partners can continue to leverage their expertise and develop video analytics needed by the security market,” said Fredrik Nilsson, general manager of the Americas for Axis Communications Inc. “The people counter does this by providing real-time intelligence and making it possible to use the camera for both surveillance and business intelligence -- making the deployment of video analytics more scalable and cost effective for end users.”
All educational facilities are moving toward better safety and security of students, facilities and staff. Schools are actively shopping for a system that will proactively monitor hallways and school grounds for possible intruders, and inappropriate student behavior. This enables school staff to better respond to emergency situations and provide video evidence for post-event investigations.
Princeton Regional Schools in New Jersey incorporated the software into its security system, which includes four elementary schools, one middle school and one regional high school. More than 3,375 students are protected by the security system.
Part of the challenge at Princeton schools was implementing an overall safety and security program, including an intelligent, centralized security management system for video, access control and alarms. It also was necessary that the system’s user interface be highly intuitive and user friendly because many people using the system would have minimal security training. Finally, the system must be able to work with users throughout the school and other district buildings so video and alarms could be viewed from multiple locations.
School campuses, whether large or small, are always in need of improved security, but any new project takes needed funds from other school priorities. Making the most of available money is paramount.
Princeton schools purchased software they felt provided the most cost-effective, scalable solution for incorporating video analytics. They looked specifically for a solution that had the ability to alert both security and non-security staff of key events in real time.
“The software we selected provides us with peace of mind, knowing that we have a powerful, yet easy-to-use video system that leaves us best prepared to prevent and respond to any threat to the safety and security of our facilities, students, teachers and staff,” said Peter Thompson, manager of IT at Princeton Regional Schools.
There are several factors that influence video analytic performance, which has helped convince me that the solution is here to stay. The algorithms work best under these strict environmental conditions.
Camera angle. The angle of the camera can influence perspective, occlusion and segmentation of objects.
Distance to object. Most video analytics require a minimum pixel size; however, if the pixel sizes of the objects are too large, the performance of the analytics can be distorted.
Lighting level. In order to detect objects, there needs to be minimum light. Abrupt changes in lighting can cause false conclusions.
Degree of activity. An environment’s activity will influence performance. The higher the level of activity, the greater the chance false conclusions will be drawn by the analytics algorithm.
Weather. Variance of weather, including sun, rain, snow and wind, can cause false conclusions, especially in outdoor environments. Weather also can impact video analytics in an indoor environment, especially if large glass windows and doors are in a camera’s view.
Background. The degree of change in the background of a camera’s view will impact the performance of video analytics. For example, if the view includes a constantly moving escalator, this could result in false conclusions and should be taken into account when developing or installing the solution.
It’s no secret that the world is experiencing a video data explosion, and intelligent video management will be part of that growth. Deployment of surveillance cameras shows no signs of declining. In fact, double-digit growth will be expected for the foreseeable future. Security threats worldwide necessitate technology improvements and a migration toward the use of surveillance cameras for non-security applications, including business intelligence.
The need to make sense of video data is ever-increasing, even though more than 90 percent of that data is unstructured content. It is nearly impossible for security monitoring staff to notice all relevant or potentially threatening activity. When post-event investigation is required, reviewing video content is laborious, at best. End users are demanding smarter video, and the ability to dissect the video data into reports, charts and graphs becomes critical for making business decisions.
Embedding analytics on edge-of-the-network devices allows video-feed processing to occur at the edge. End users can choose only to send significant data to the service, meaning less bandwidth and storage is required. The bottom line is cost savings. Embedding analytics on the edge reduces server hardware requirements, which reduces acquisition costs. Generally, this means hardware and maintenance requirements are minimal, which also lowers the costs of total ownership.
Officials at the Port of Jacksonville, Fla., recently realized they needed a comprehensive system for port facility security, including perimeter security and infrastructure protection. What they truly wanted was the ability to respond to potential threats more proactively and better coordination with Customs and Border Protection staff.
By implementing an innovative wireless command center -- with a map-based interface and flexible alert output -- a firstresponder system, which stitched together an open architecture that provides port officials a common operational picture, increased situational awareness and the ability to respond to threats in real time.
Analytics has been over promised and under delivered, but the fact is, it is still in its infancy. Motion detection is simple, and it is the most basic form of analytics. The optimum place to have analytics is at the camera because it is the only true scalable solution.
Like with all good things, mature analytics will take time. Analytics is coming; baby steps first.
This article originally appeared in the August 2010 issue of Network-Centric Security.
About the Author
Ralph C. Jensen is editor-in-chief of Security Products magazine.