A typical restaurant is staffed 95 hours per week. The countless weeks I was in my restaurant each of those hours certainly proved out in the P&L. Customers had an extraordinary experience, labor was kept in check, mistakes were at a minimum, waste was non-existent and gross profit was above average. However, these hours are not sustainable. If I stepped away, would the P&L prove sustainable?

I will focus several subsequent posts on how I reduced “employee theft” in my restaurant using managed video. Let’s be very clear – I’m not talking about a premeditated theft of $20 out of the register. Instead, I am referring to a more dangerous and habitual operational problem that bleeds the top line, bloats food and labor costs and ultimately erodes gross profit to unacceptable levels.

In the 1979 version of When a Stranger Calls, babysitter Jill Johnson (played to perfection by Carol Kane) is repeatedly called by a person whom she does not know. The caller, in a calm and deliberate manner, continues to ask her “Have you checked the children?” Obviously concerned by the harassment, Jill Johnson calls the police for help. Who wouldn’t?

For those of you who have not seen the film, you may view the climax of the scene below. Great seventies film-making!

What you may not know is where this post is heading. What does this have to do with Managed Video? Any guesses?

When a Stranger Calls

A key component of managed video is that it must work like it’s just “built-in” to your network.

Step by step, access to video, video acquisition, storage and management will become components of “the network”.

The form factors that these components will take will vary.  They may be separate hardware devices, or expansion cards for an existing devices or software added to some device.  But the end result is that these components will become increasingly integrated as part of “the network infrastructure”, just as email, websites, fileservers, routers and switches are all part of the network infrastructure.
Why?  To maximize the effectiveness of video and minimize the overhead.

Effectiveness is massively improved when the video can be linked to data.  This data resides on other systems in the network, but unless they are linked together, value is lost.  Effectiveness is also significantly improved when access to the video when many people can view the video.  Outside of internal security, there is external security, such as first responders.  Then there’s operations, marketing, maintenance, management, Human resources and the list goes on.  All of these people are on the network now and could take advantage of the appropriate, authorized access to video.

Even at a medium sized business thousands of components may be involved in a video system.  Installation and operation of these must be as plug and play as possible.  Users with basic computer skills need to be able to jump in and find what they are looking for quickly.  Component failures must be rare, easily detected and remedied.

By integrating components into the network infrastructure one can apply established methods for remotely operating, monitoring, repairing and controlling access to massive, distributed video systems.

If your video system isn’t linked to “real” information, it is providing only a mere fraction of the value it could be.

The future of video surveillance lies in how to manage all the data being collected and turn it into something useful. Our ability to collect ever more massive amounts of data is growing at a rapid pace. Video management systems are part of the reason we can more easily collect such data.

However, most surveillance video is not extremely valuable until it is accompanied by some kind of information that makes that video searchable. In fact, without information to give context to the video, it is not valuable at all. That information might come from a person sitting in front of a video who knows something about what is going on or it might come from a door access control system or a point of sale system. Given the sea of video out there, one must have some help from automated data systems.
This is the premise of Steve Hunt’s PSIM acronym which he talks about at his Security Dreamer blog.

I think Steve would agree that the value of turning video data into “information” goes well beyond Physical Security and is especially pertinent to the broader category of business intelligence. When it comes to business applications, Steve points out that security is really not the point; rather the ability to improve one’s business is a huge factor driving growth in video.

I spoke with Paul Bodell from IQeye recently.  Paul wrote an interesting article calling for an industry video quality standard in Video Technology & Applications of SDM Magazine.

Explaining video quality to customers who are non-technical is difficult.  How can you tell a customer what level of quality to expect from a particular installation?  IQEye’s approach has a lot of merit.  Paul is promoting creation of a industry standard for a quality level that is high enough detail to identify a face.  Paul calls this a “forensic” level of detail.  The technical spec to achieve this is “40 pixels per foot” for this forensic level of detail.  Take a quick look at this proposed specification.

To give a real world example, a customer wants to identify the face of a customer at a cash register and the denomination of bills going in and out of the register.   A cost effective to do way to do this is with one analog camera at 640×480 resolution.  With this, one can cover the register and counter next to it, covering the transaction with a customer and have a good chance of being able to identify the bills and customer faces.

Here’s how came up with that:  A 640×480 image is 640 pixels wide and 480 pixels tall.  640 pixels / 40 pixels-per-foot  = 16 feet.  So this camera can cover a 16 foot wide field of view and retain a forensic level of detail.  The highest end analog cameras support up to 704×576 resolution, giving a slight boost.

Some scenarios don’t require a forensic level of detail to be useful.  For example, the manager of a retail store can recognize her own employees with less than the forensic level of detail.  A single analog camera might cover a restaurant dining area that’s 32 feet across.

If one needs to be able to identify license plates across 8 parking spaces, that’s about a 64 foot wide field of view.  In this scenario, single multi-megapixel IP camera may present a cost competitive option compared to the installation of 4-5 analog cameras.

Krissi Danielsson provides a nice podcast interview with Paul Giurata of Catalyst Resources where they discuss the The Right and Wrong Way to Do SaaS.

One of the key characteristics they is the frequency of updates. Specifically, Paul suggests that “being able to update the software on probably a no more than 90 to 180 day cycle as opposed to one year or 18-month cycle” is important.

I wonder what other SaaS product teams think of this. At Envysion, our current goal is to release an update into production once a month. Is this too frequent? Tomorrow I’ll share with you what one of our customers thinks!

Businesses are beginning to leverage video in a number of different ways. Surveillance video is no longer being used solely as a security application.