Surveillance As A Service?

Surveillance as a service, pitfalls and caveats


Axis offers the AVHS solution, allowing users to add their camera to an offsite NVR-thingy. The concept, at a glance, appears solid; pop in camera, connect to the AVHS server and off you go. Simple and easy.

There are some caveats i suppose. Most consumer ADSL lines are asymmetric – receiving is much faster than sending (by order of magnitude usually), and while you can get symmetrical lines these are usually pretty expensive. I guess a large percentage of clients who’d be tempted to outsource their video surveillance, are also the ones who do not want to spend a fortune on the system, especially if this is a recurring cost.

But surely, a 1 MBit connection and the magic of h.264 will allow people to stream 16 channels at CIF/10 fps to a remote server?

Yes, and … well… perhaps.. it all depends. If the scene is static, and you only want to record when people enter the restricted zone, then I think this would work great, but it would present some challenges; What if the motion sensitivity is too high and we saturate the bandwidth? What if the compression is too great to use the video for identification? And what happens when the external network goes down (which tend to happen at the most inconvenient time)?

If the camera itself is “intelligent” enough to only send video, or perhaps send video in low resolution/high compression when nothing “interesting” is going on, then could I see this taking off in a big way, but we still have some ground to cover.

My concern is that “we” (as an industry) have a tendency to oversell. Often the caveats are not revealed until the client has crossed the point of no return.

  • “Ah.. yes, megapixel is only one camera per installation”
  • “oh, you can only record on motion”
  • “and by the way – motion can never be more than 20% of the time”
  • “don’t forget, and only 4 cameras can have motion at the same time”
  • “only works with camera a, b and c that cost 40% more”
  • “does not work for loitering detection”
  • “does not work for left object detection”

The way the AVHS is the way things should work, I can’t wait to take a closer look.

Keep Your Mouth Shut!

Microsoft presents Office on Windows Mobile 7

….until the product is ready for launch.

A poor Apple employee lost a phone*, and the world lost their collective mind. Microsoft does not have that problem, because they will yank out every single experiment and present it as “almost ready”. In some cases the “product” demonstration is staged, and is simply a video (same thing happened with the PS3 by the way.

The video below shows Windows 7 Mobile – but it is not the real thing. I am fairly confident that the real experience will be vastly different (choppy wipes, plenty of hesitation and inaccurate control), so I wonder why Microsoft insists on setting themselves up for dissapointment once the real devices are available in the shops.

If Microsoft can pull this off then my next phone would be sporting a Microsoft OS (otherwise a nice Android phone will be prominently displayed on my desktop)

*I don’t believe it was a PR stunt – contrary to popular belief, even Apple employees can make mistakes.

Good Article on Productivity

Most ideas are abandoned at what I’ve come to call the “project plateau”: the point when creative excitement wanes and the pain of deadlines and project management becomes burdensome.

Yes, I’ve had a million ideas – two or three actually became modestly successful, but what happened to the remaining 999.998?

Without realizing it, most of us have gradually adopted a “reactionary workflow.” We are constantly bombarded with incoming communication: email, text messages, tweets, Facebook posts, phone calls, instant messages, etc. Rather than be proactive with our energy, we spend all of our energy reacting, enslaved to the last incoming item.

Reactivity and The Project Plateau is spot on.

Five Tips For Making Ideas Happen | Smashing Magazine

Is IP Video Surveillance Still In Its “Infancy”?

About 10 years I wrote my first piece IP video surveillance code. The camera was an Axis 200+, it was outrageously expensive, extremely slow, rather low resolution and no-one in their right mind would ever acquire such a device. Not to mention that there was not software to store the video (more accurately, a slow stream of single images), so I guess that was why I was frantically hammering the keys on my Celeron 300 MHz based PC.

Over the years I have worked in a number of fields were standards defined (TWAIN, OPC, DICOM to name a few). Yet, in none of them interconnectivity was trivial. What was sold as “plug and play” and “open” almost always required a lot of glue, spit and experimentation to get part A to work with part B. Usually vendor A had one interpretation of the spec, and vendor B had another. Several profitable companies doing nothing but converting from A to B emerged to capitalize on the difference in interpretation (plus there are often strategic reasons to prevent interoperability).

Today we have bits and pieces for function, but the interconnectivity between systems is just not there yet. We see attempts from companies to attain a sort of de-facto “standard”, usually there are claims of “Open Architecture” (I am still not sure what that means exactly), but my impression is that a client will be locked in – rather tightly – with whatever vendor they chose. Once they pick an NVR from vendor A, chances are that everything else bought after that initial installation will have to come from the same vendor (or someone with a tightly knit relation).

A clothing store in the NY area has the motto “an educated customer is our best customer” (Syms). I happen to like that motto. If the customer knows what they want, they can make a better choice. If they need a wrench, and I am selling hammers, then I will have no problem telling them where the “wrench store” is. It allows me to focus on building a better hammer. If – on the other hand – we try to trick the client into using a hammer instead of a wrench, then the client will not be happy, I will spend resources making a half-assed do-it-all tool.

It is my impression that many NVRs are “do it all tools”. There seem to be a fear of specialization right now, we all want to be “excellent at everything” which is obviously not possible. They become these mutant applications that are access control, forensics and CAD drawing all fused into one monolithic application.

I think part of maturing is to allow interoperability across all the elements – let a good forensic tool from vendor A access the NVR’s database from vendor B.

Perhaps we’ll see some of that in the future.