Video Content Analysis: The Last Mile is Key

Statistics indicated that roughly US$50 million was spent on PC- and server-based video analytics back in 2008, whereas market forecasts for this year and 2012 are estimated at $120 to $140 million. Turning raw video into actionable information is invaluable not only for security purposes, but also for a wide variety of applications to improve operational efficiency. Despite a steady uptake for the past few years, mainstream adoption remains relatively challenging, and this feature explores why.

Having the system understand, to a certain extent, what is going on in the scene truly creates value. The human mind is not optimized for tedious, repetitive tasks; a guard can only stare at monitors and concentrate for so long before his mind begins to drift. Video content analysis (VCA) is here to help alert relevant personnel of any critical incident, and that is exactly what users should expect from it.

A scalable, intelligent detection system can be built through utilizing VCA, said Gerard Otterspeer, CCTV Product Marketing Manager, Bosch Security Systems. “You make your system intelligent, and it will tell you when it needs your attention. It should be positioned as a ‘guard assistance' system, where the human is notified of and acts upon alerts and alarms. This is what users are beginning to realize, and this helps them understand how intelligence fits into their security system.”

The key word was assistance, not replacement, as VCA can do many things, but it is not as magical as some make it sound. Video analytics is a science — one with clearly defined rules that must be followed for optimal results, Otterspeer said. “VCA provides additional benefits in modern surveillance systems and its overall value further justifies the investment in an advanced surveillance system.”

“A survey we did of 1,200 customers this year has shown that good video analytics was the most important component of their security system,” said Ivy Li, cofounder and MD of iOmniscient. “For more sophisticated users, the value of having a camera diminishes when it does not have video analytics, and they question why they need the camera at all.”

Smoke and Mirrors
Video analytics has been around for quite a while, but has not seen wider adoption over the ye ars. One reason may be unrealistic expectations from end users. These expectations were built up in the early days by overaggressive marketing, and later fueled through movies and TV shows, only to have end users disappointed in what could actually be delivered.

Over selling features that clearly need more R&D before reaching commercial viability can be problematic, said Thejaswi Bharadwaj, Head of Civilian Technologies at Delopt. “The overselling of premature algorithms over several years has unfortunately worked, and customers expect human-level accuracies. Efforts from VCA solution providers are required to communicate with consultants who draw up project specifications.”

Educating the market is, thus, critical for VCA to be successful, Otterspeer echoed. “Hollywood productions generate a ‘CSI effect' that sticks in people's minds. They have amazing special effects for intelligent video to ‘wow' people, but reality is a bit different.”

When users' high hopes were beaten down by a myriad of false alarms, some customers actually turned off analytic capabilities, said Steve Gorski, GM for the Americas, Mobotix. “The industry has learned much more about market needs and the technology to offer fully robust solutions that provide significant value to the end customer today.”

The focus is now on the analytics that provide the most operational value, such as people counting, loitering and directional detection.

“Many users have had very poor experiences with poor-quality products and are hesitant to spend money again. I believe this has affected overall growth in the market,” Li said. “But note that the market still exists. Most of our projects involve thousands of cameras. The largest one that we are working on now has well over 100,000 cameras.”

Another reason for some VCA companies today involve commitment issues, Li continued. Many vendors have significant venture capital(VC) backing . “VCs usually have an expectation of an exit strategy within two to three years. Their focus is on having the minimum acceptable technology, and they then attempt to get a few sales so that they can position themselves for an exit. In virtually all these companies, there is very little focus on improving their technology. With poor technology generating lots of false alarms, it is very difficult for such companies to grow sales. When the VCs grow impatient, these companies shut down and leave a bad name for VCA. Users have been scarred and become wary of similar technology,” Li said.

Empty Pockets
Another factor is cost. "There have been cost barriers to analytics being installed more widely — not just the cost of purchasing but also time required from staff to install, fine-tune, manage and maintain systems with analytics," said Christian Bohn, VP of Product Management and Marketing at Milestone Systems. “Furthermore, they are processing-heavy on the system and require more power in the hardware, so the investment is not just in buying the software. Typically, after installing analytics, fewer cameras can be operated perserver. This adds to the overall cost and complexity of an installation.”