Will BRS Labs Revolutionize Video Analytics?By John Honovich, Published on Aug 09, 2008
Recently, BRS Labs announced its company to great fanfare. Indeed, BRS Labs has a very interesting offering but great questions remain about its use and its competitive viability. This report examines what BRS Labs is offering and what potential impact it may have in security operations. [Update: BRS Labs reports $12 M USD in new funding].
BRS Labs claims to alert on generally suspicious or abnormal situations without having to define specific rules or activites of interest (the way today's video analytics do). Here is a summary of the key points:
- Their offering is fundamentally different than any commercial product.
- It has the potential to solve significant security problems not possible with existing video analytics.
- BRS Labs needs to determine what applications they will focus on.
- If they focus on applications served by existing analytic providers, the probability is high they will lose.
- The highest potential is in reducing loads on actively monitored video surveillance systems.
- It remains to be seen what the performance levels and cost structure they can achieve – a huge factor.
BRS Labs' claims are fundamentally different. While all commercial available analytics today monitor specific pre-defined conditions (perimeter violation, abandon object, person loitering, etc), BRS Labs proposes to identify events that are out of place relative to their environment. The best explanation of this is offered by the embedded flash video on the BRS Labs' homepage (about 3 minutes in). The example is that BRS Labs will learn the patterns of people in a garage (what paths people walk, how they act, etc). When a person acts sufficiently different from 'normal' activity for that camera, BRS Labs triggers an alert.
Solve Significant Security Problems
By identifying the abnormal, BRS Labs opens up the potential to use analytics for a great variety of issues that constrained rules based analytics cannot do.
A lot of security incidents cannot be neatly divided into rules such as crossing a perimeter or leaving an object. This is fundamental reason people are so important to security - strange behavior that is difficult to clearly define is often an indicator of problems. Indeed, being observant of abnormal behavior is a key element of human security monitors. Most of the time the abnormal behavior is not threatening but security monitors know that this is where the greatest probability of valid incidents lie.
What Will They Focus On?
From a production and deployment standpoint, BRS Labs is still a young company and does not have set focus on markets or applications yet. This will certainly come with time. The major question is whether they will offer a better solution for existing analytic uses (perimeter violation, abandoned objects) or they will open up new uses for analytics.
Competing with Existing Providers
If BRS Labs decides to address analytic use cases that existing vendors supply (such as ioimage [link no longer available], ObjectVideo, Agent Vi, etc), this offers little value for customers and high risks for BRS Labs. Even if BRS Labs is as good as existing companies for these specific use cases, they will inevitably have difficulties matching the cost structure, system support and packaging that these fairly mature products already offer.
Actively Monitored Video Surveillance
The greatest potential for BRS Labs is in reducing the staffing needed to run large activiely monitored video surveillance operations. Malls, retailers, campuses, municipalities, all routinely have dedicated monitors for watching security cameras. Existing video analytics are of limited use in these scenarios as analytics can only identify a small subset of the types of problems that occur (unlike a military base which are usually locked down and tightly controlled).
For actively monitored video surveillance, the operators could use BRS Labs alerts as pointers to focus on abnormal situations rather than constantly scanning dozens of cameras. This would likely increase scale of cameras monitored as well as probability of detecting genuine incidents. Furthermore, because these operations already dedicate humans, nuisance alerts will be less of a problem. Moreover, the only company currently targeting this application is Cernium [link no longer available], making this a more open segment.
Performance and Cost
All of this being stated is simply a matter of potential performance and positioning. BRS Labs is still a young company. While they have numerous early stage deployments under way, they are still learning and refining. Their analytics are not yet deployed in cameras nor is there publicly released integration with major video management systems.
Some of this is a matter of time but part of this is an issue of the resouce demands necessary to make this work. One advantage of existing point analytics (like perimeter violation) is that companies can optimize for a very specific set of conditions. A more general approach like BRS Labs could require significantly more resources resulting in more processing power and architectural restrictions. I will not speculate further but potential partners and customers should check and understand any implications here.
BRS Labs certainly is offering something significantly different and potentially quite valuable. As they are new, it will definitely take time for them and the industry to better determine what applications they will focus on, how well it will work and how much it will cost. Large security operations and video surveillance companies should watch BRS Labs carefully.