Video Analytic's Greatest Value - Storage Reduction

While video analytic promotion focuses on stopping and solving crime, the greatest value for video analytics historically and for the immediate future is reducing the cost and amount of storage. This report analyzes the top 4 potential impacts of video analytics and explains why storage reduction is the most valueable.

Many disagree with the value in storage reduction.  For instance, Milestone recently claimed about analytics: ""It's a myth . . . to think that you're going to reduce what you send back over the Ethernet. You're going to need to have the full recording." Yesterday, Bosch agreed claiming the Milestone statement was "spot on."

Here are my central claims:

• To date, video analytics has created the most value by enabling motion based video recording, saving hundreds of millions of dollars for security customers.
• Video analytics for alerting is 'sexy' but very hard to deliver in production.
• Video analytics for searching has limited customer value (certainly far smaller than alerting)
• More advanced video analytics could reduce storage costs by more than 75%, savings hundreds of millions more for security customers.
• The rise of megapixel cameras and the expansion of managed video as a service will drive the adoption of video analytics for greater storage reduction.

Motion based video recording is such an essential part of video management that most people, even myself, do not include it as a 'video analytic.' Usually the term is applied to more exciting uses like perimeter violation, abandoned object detection, loitering, facial recognition, etc.  Nonetheless, motion based video recording does analyze video to determine when to record.

Today's motion based video recording reduces storage consumption dramatically.  While it depends on level of activity in the monitored area, the reduction can easily by 50% to 80% per site which translates into $1,000 to $2,000 in savings for an individual office or store. This is especially valuable in most facilities where activity is generally modest during the day and non-existent during the night (office building, most stores, hospitals, etc). Multiply this by the hundreds of thousands of video recorders deployed and the economic impact is tremendous.  Most people use or should be using motion based recording, unless they are constrained by regulations or are a casino where slight of hand and incidents are basically continuous.

What's especially important is that motion based video recording can be implemented fairly easily and with minimal risk of problems.  Even the simple algorithms used today err on the side of caution and record even at the slightest hint of motion.  'False matches' in motion based video recording simply reduce storage savings. In my experience, you rarely if ever lose a valid incident with most based recording but you save significantly in storage costs.

Contrast this to real time alerting where 'false matches' create a major usability issue. So while the potential value of alerting is high, the value that can be achieved in production is questionable.  By contrast, with motion based video recording, you can almost always guarantee that as soon as you turn it on you are saving thousands of dollars per recorder.

Now, let's examine using video analytics for searching.  Dr. Bob at Bosch claims that he "personally find this feature to be invaluable". The value proposition is that a searcher can instantly scan through months of video to find exactly what they want.

In practice, though, using video analytics for search offers limited value.

1. In security, the key is stopping or preventing incidents. Solving cases is necessary but far less valuable than eliminating an event. This is the opposite of information retrieval (e.g., Google searches) where most people want the best information regardless of time.
2. In most first world countries, the number of investigations per camera is very low (crime is simply not that common relative to the number of cameras deployed). This limits the potential value because there just are not that many cases.
3. Most cases can be solved by a simple time search or basic motion search (e.g., super simple, the burglar alarm went off at 2:22pm or still moderately simple, someone broke in last night between 11pm and 7 am).  Very few cases need specialized analytics to be solved.
4. This results in very few cases that cannot be solved otherwise.  And of those cases, video analytic based search may be only able to solve a fraction of them.

This is not to say that video analytic driven search can never create value.  However, it is difficult and you should carefully assess such claims.

Conclusion - How Video Analytics Can Reduce Storage Costs Further

As valuable as motion based recording has been, 3 major opportunies exist for video analytics to reduce storage costs even futher:

1. Today's motion based analytics still record a lot of non-motion activity.  More advanced motion based analytics could reduce storage costs even further. For instance, 3VR is already using analytics to selectively record only faces. Apply this same method to recording people, cars, license plates, etc. The result can be near infinite storage of the most valuable evidence at almost no cost.
   
2. Megapixel cameras storage demands are exploding the cost of storage. H.264, as it matures will certainly help.  Motion based analytics can have an equal impact in making megapixel cameras affordable to mainstream users.
3. Removing local storage offers significant economic benefits for customers.  To do so, either WAN bandwidth needs to cheapen significantly or video analytics need to reduce the amount of bandwidth needed.  Video analytics is the more likely choice and a key driver in making manged video a reality.

At the end of the day, storage is simply a tax on video surveillance users.  Security managers do not feel better by having more storage.  They would gladly do with less storage as long as they could get the same results.  This pattern is demonstrated by an almost constant desire by security managers to better optimize and reduce storage costs through numerous means.

The big DVR appliance companies certainly want to sell more storage (it's one of their key profit centers).  But the IP video surveillance software companies (ironically, companies like Milestone) should be motivated to cut storage costs.  All business should commoditize their compliments because it helps sell more of their own products.

I am hoping and watching to see how storage costs can be reduced further. It definitely can be done and security managers will invariably benefit from this.