Bookmark and Share

Looking Up - Where Next For Cloud Based Video Surveillance?

By William Rhodes
IMS Research

At the end of 2010, IMS Research forecast the increased traction of cloud-based video surveillance, also known as Video Surveillance as a Service (VSaaS). The market certainly picked up in 2011, with notable sales growth at some of the vendors in this space. Overall, the market grew by around 20-30%. However, the market is still emerging, developing and evolving. IMS Research believes that there are a number of potential avenues for suppliers of VSaaS to explore in the coming year.

One application that could augment the functionality of VSaaS is the addition of cloud-based video content analysis. By running analytics in the cloud, users can take advantage of the large virtualized processing power available to them. This use of the cloud, named Infrastructure as a Service (IaaS), is common in other industries. Examples include the Apple’s voice control software, Siri, which utilises IaaS to process complex voice commands and Amazon’s Silk Internet browser that harnesses IaaS to accelerate page loading times. This application opens up the potential for a different type of business model for analytics vendors. Rather than charging on a per channel basis, vendors can charge for services aligned to analytics, such as report generation. For example, a retailer could pay a given amount each year to receive monthly reports on store metrics, such as the daily customer count, average queue lengths, etc.

A majority of the security cameras used in small and medium sized enterprises are solely used for security reasons. However, some of the more technology savvy users are beginning to leverage and monetize video from their existing systems. Rather than simply leaving video dormant on a hard disk drive, business owners could post or stream video to the Internet, allowing customers to see real-time footage of the place they intend to visit. Video could be integrated with Groupon, TripAdvisor, Yelp or even Google Street View. Prior investment in security equipment could be transformed from an expense to a revenue generating tool; for example, by analyzing the effects of a Groupon or Yelp promotion. IMS Research estimates nearly 65% of the DVRs sold in 2011 were Ethernet enabled, demonstrating the potential market for video to be distributed to the cloud and used for non-security related solutions.

Another potential application for VSaaS is video analysis by people. Video from security cameras can be distributed to the cloud for monitoring purposes. This monitoring process could be privately conducted, where operators (either trained or publicly crowd-sourced) view video to identify shoplifters or ensure the correct practices are taking place in factories, for example. This type of monitoring can be conducted anywhere in the world making the solution potentially more accurate and cost-effective than some software-based analytics systems. With regards to crowd-sourced monitoring, this involves the general public viewing video from public spaces. Individuals are tasked with identifying unusual or illegal behaviour. Whilst there are privacy issues with this application in some countries, a small number of vendors have started to offer solutions. One example is Internet Eyes, a UK company, which offers an online monitoring solution that allows registered members to view live camera feeds from businesses, and notify them when a crime is observed.

Whichever of these applications for VSaaS takes-off, one thing is for certain; for VSaaS to continue to gain traction, solution providers need to look at new applications for video surveillance. VSaaS is not an inexpensive option, so providing solutions that offer “real” benefits (e.g. monetizing otherwise unused video) will fuel growth for this emerging market.

krish108
Offline
Last seen: 48 weeks 6 days ago
Level 1: Prestidigitator
Joined: 2012-06-20
Points: 1

@newman: true that as computational price/performance improved, the 'smarts' moved back into the individual boxes (the thin clients example below). But does seem like Moore's law extension to processor performance has lost steam. Since 2004, performance has started flattenning out (2.5GHz has been around for quite a while now). Yes, chips have been getting cheaper and smaller (Moore's law for technology shrink continues to hold). But throwing in more processors makes programming tough -  you get a 1.3x boost with 2 processors, not 2x but maybe 20-30%. So, it is not going to be easy to put in "lots and lots of analysis" into embedded systems. But yes, there is focus on it, with specialized architechtures, FPGAs, DSPs, GPUs, all trying hard. So, yes, we have room for improvement here, 'medium-term'. But if moore has truly stopped helping performance, the long-term does mean centralized. Why? It is more efficient. But the real benefits of aggregation (cloud) comes not from cost. Yes, it is cost. It is also upgradability to more features. Upgradability to more processing capacity (better and higher level algorithms using the same device), making the vision sensor live a longer useful life. It is in being able to "see" history...(is that the same guy that walked into the store last Saturday?). It is in not having to maintain updates. It is in being able to view from anywhere (even if the device is down, you can view history). Take a Siri - because it can connect to Cloud, the server can leverage training from millions of voices across the world. Yes, it does increase bandwidth usage, but why have all these fiber to the home been layed for? Now, how about making multiple devices talk to each other - because the "cloud" knows what's happenning here, it can ask the other camera to "look for that". Yet another aspect is the ability for sourcing the apps. An Apple app-store enables so many apps on millions of a ipads. Now a cloud analytics can enable millions of apps for the millions of devices out there that can connect up. So, seems like there is a lot lot more to it than a simple thin-client to multiple-PCs analogy. But, for embedded processors, there is a lot of fun waiting the next few years (the 'medium term'). And they will stay in there, maybe still performing a lot of "lower level" processing (measuring th length of a smile :). And the cloud grows more and more on the "higher level processing" (is that the same smile :-))

p.r.newman
Offline
Last seen: 3 years 5 weeks ago
Level 1: Prestidigitator
Joined: 2012-01-25
Points: 3

Cloud based services are fine when the ratio of computational demand to server supply is workable. In this view, IaaS is just the good old-fashioned "thin client" approach rebranded. But once demand exceeds supply, you hit computational and bandwidth bottlenecks and have to move the "smarts" back closer to the problem once more.

My guess is VSaaS will go that way once everyone wants all their cameras monitored all the time. Embedded vision analysis is the way forward. Likewise Siri and voice recognition.  There are an awful lot of cameras out there, and an awful lot of voices.  I note that Siri already "doubles iPhone data usage" (see www.tmcnet.com/topics/articles/252303-apples-siri-doubles-iphone-data-usage.htm) and all that data has to get into a server warehouse somewhere.