The discussion of video analytics can be enough to send anyone’s mind into a complete tailspin: mixed reviews on the pros versus cons; suggestions that video analytics caused a major over-hype causing such mixed reviews; that the disadvantages outshine the benefits. All of these ideas carry some truth, yet the question still remains. Is video analytics really all that it is cracked up to be? Majority vote points to yes, however, there’s tons of information for the integrator to sift through to find out what best fits the customer and the application.
This is not meant to stray from the fact that the overall video surveillance market has seen major improvements in the past few years. DSP technology motivated the use of video content analysis to be embedded on such devices as DVRs and cameras. Developers of encoders and decoders continue to focus on the speed of data transfer over wireless links in addressing video distribution. Megapixel security cameras have seen major growth and advancement in resolution technology, including H.264. IMS Research recently issued a report predicting that the remote video monitoring and surveillance market could grow as large as three times the estimated $158 million 2008 revenue by 2013. One thing is clear–video analytics carries a multiplex of unknowns that developers and manufacturers of these solutions today need to address in order for widespread adoption to occur.
Facing the misconceptions
Misconceptions of video analytics today lead back to its early presence 10 years ago. The overall idea of its progression is that not many major developments have occurred in comparison to the changes in expectations of video analytics. Right now, defining what video analytics is and accomplishes adds to its overall slow growth in the market, also due in fact to the economic strain of the past few years. In a statement about the forecast for 2008, IMS Research reported that the market got off to a slow start in 2008 and claimed it unlikely that the market would grow as fast as it did the year before.
“The market has been very slow to evolve,” said Jack Cabasso, managing director, Aventura Technologies, Hauppauge, N.Y. “The word ‘video analytics’ is an issue. Customers will sometimes stress that they are looking for video analytics but don’t necessarily know what it is they want it to do.”
Part of the struggle in the market is due largely to the fact that expectations were not set properly on what analytics could accomplish.
“Going back three or four years ago, there was a lot of hype around the technology and its capabilities,” explained Peter Wilenius, vice president of Global Marketing, March Networks, Ottawa. “Certainly the promise was far in excess of the actual capabilities so there was a bit of an overpromise on the part of the industry, which has probably slowed things down and has created skepticism as well as reluctance in adoption by both end-users and integrators.”
Still other misconceptions revolve around the technologies that it does apply to.
“Motion detection is mistaken for being video analytics and it creates quite a few false alarms,” said John Whiteman, vice president and general manager, ioimage, Denton, Texas. “People think that they’re getting analytics when in reality they’re getting motion detection and they set it up only to receive all these alarms on inanimate objects that they didn’t want to detect.”
When video analytics first became available, it started out as advanced research and development initiatives meant to protect critical infrastructure initiated by the government, according to Nick Gagvani Ph.D., Chief Technology Officer (CTO), Cernium Corp., Reston, Va.
Targeting what video analytics provides is the differentiating factor among the increasing number of these providers in the security market.
“There are a lot of people who label things as video analytics and it’s questionable as to whether or not those are really video analytics or they’re some subset of what an analytics system can be,” said John Szczygiel, president, Mate-Intelligent Video, McLean, Va. “Consumers have to be careful and take a hard look at what the company is really offering. I’ve seen a number of manufacturers recently who will make a big splashy advertisement abut their video analytics but if you look a little deeper into it, the analytics that they’re describing are things like detecting video loss, camera tampering or perhaps a simple type of video motion detection. True, those are forms of video analytics, but I believe that when most consumers think about video analytics, they’re envisioning more robust capabilities.”
Identifying the goal
Video analytics applies to a number of applications. Video motion detection, which is the most basic, according to Fredrik Nilsson, general manager, Axis Communications, Chelmsford, Mass., does help to filter out a lot of the activity. Camera tampering and license plate recognition also receive decent commercial traction.
“Today analytics encompasses a lot of things to different people,” said Whiteman. “People count, facial recognition, tailgating. But intrusion detection remains the No. 1 analytic choice for the market and is the No. 1 revenue-generator for analytics globally.”
The main goal of video analytics is to analyze video using software algorithms and it is important to define its realistic expectations.
• Video analytics is more of an aid then a replacement.
• Finding applications where false alarms are not a big issue and can be dealt
• Providing ease-of-use.
• To increase operator efficiency and security.
• Key goals of video analytics are automation and gathering intelligence about
Hardware versus software solutions
Many video analytics providers offer video analytics as a multiple integrated solution with video management software, while other companies follow the embedded-technology approach, offering the video analytics directly on the device.
“We provide hardware and software and tell people, ‘with a hardware-based system you’re going to be limited to these functions-here are the false alarms you’re going to have and why you’re going to have them,’” said Cabasso. In other words, software-based solutions allow for establishing parameters specific to the environment in which it will operate.
“Embedding the video analytics on the camera today means a tradeoff in performance,” said Szczygiel. “If a camera is an IP camera and it does encoding and video analytics, it’s typically either a really good encoder or video analytics platform.”
Other technologies that are improving the use of video analytics is that of encoders and decoders.
“Encoders and decoders are essential in video analytics solutions,” explained Gagvani. “I think it is an incorrect perception today that analytics requires an investment and replacement of the existing infrastructure. You can give existing analog infrastructure a new lead of life by employing analytics-enabled encoders.”
Video analytics takes a new stand
With endless discussions on video analytics, it is vital for present developers of this solution to remember what unmet expectations can mean.
“People have a strong desire for this technology because the promise of video analytics is that it will automate the most mundane part of active surveillance and enable guard-like protection to a much wider portion of the market at lower operating cost,” said Scott Schnell, president and CEO, VideoIQ, Bedford, Mass.
With the many video analytics providers as there are today, it is vital to know your target audience in providing the right solutions for the right end-user.
“The technology has had time to mature,” said Warren Brown, director of Product Management for Access Control and Video Systems (ACVS), Tyco International, Princeton, N.J. “Particularly server-side analytics, leading providers can now credibly state, ‘here is the clear value proposition of how this side of analytics generates return and value for surveillance and security professionals.’”
Today’s expectations of video analytics are definitely more realistic.
“For the video analytics portion, it is still for the most part a niche market although it is growing,” said Richard Caballero, vice president of Channel Sales, Aimetis Corp., Waterloo, Canada. “From a system maintenance standpoint, that is where a lot of our partners need a lot of support and potentially may be a threat if they don’t know what they are doing.”
With all that is said, it is safe to say that yes, video analytics does seem to have a bright future as companies and developers continue to reinvent solutions for end-users wanting this technology. But it will take time for these changes to really affect the growth of video analytics and push it back into the industry as a solution that really does offer all that it claims.
Developments in Video Analytics
• Improved processing power equals improved intelligence of algorithms
• Migration to business intelligence
• Increase of analytics-enabled video channels
• Retailers looking for solutions for increased theft will drive demand for
video intelligence and also tie it to sales and marketing
• Increased interest to integrated solutions of analytics
• More hybrid analytics solutions
Texas Instruments New Solution
Texas Instrument released their new VLIB software library designed to accelerate development and increase performance of video analytics applications by optimizing key kernels specific to the video surveillance market. TI’s new VLIB library is offered at no cost as a free aid to video analytics solutions providers.
“The focus of VLIB is one of acceleration,” explained Bruce Flinchbaugh, director of Video and Image Processing Laboratories, Texas Instruments. “It helps some of the very small functions that are key to some of these video analytics algorithms and the role of VLIB is to accelerate those key functions.”
For more information, visit www.ti.com
Benefits of analytics as software
• Software-based solutions generally deployed on IP video based solutions.
• Less limitations in terms of function.
• Bigger possibility of integration which benefits the end-user more.
• Better control of analytics from end-to-end.
Benefits of analytics as Hardware
• Able to deploy in highly distributed fashion.
• Hardware lets you bridge from analog to IP.
• Embedded technology on a camera equals an easier package to sell.
• Hardware better for fixed, non-moving parts such as DVRs.