Analytics - To Lead or Not to LeadBy: IPVM, Published on Sep 19, 2008
I'm not much for Shakespeare, but this is a good question. Over the past few years, many security companies have been leading into an Enterprise IP Video project with analytics as the main driving force. In a presentation, they BEGIN by trying to show the exciting stuff of analytics (bags left behind at the airport, tripwires at water treatment plants, facial recognition at a soccer game, license plate recognition at the toll booth, etc.), and leaving out the entire video management platform upon which those analytics would run..
The truth is, there are several underlying considerations that must first be discussed before a project should consider even the NEED for analytics. Unfortunately, this is not often the case.
Here's the basics in my humble opinion:
Legacy Analog Video - If the project needs to incorporate existing analog camera infrastructure as it moves forward into the IP video world, they need to be considering encoders into the project to ensure the can simply get the video onto the network.
Network Infrastructure - Do they even have a network resilient enough to transmit the video to and fro. Or will they be building a private network just for the IP Video?
Recording Architecture - Consider how to get that video OFF the network and onto HDDs (or other storage, like EMC, SAN, NAS, etc.). A Commercial Off The Shelf, server-based recording platform is often better, but in some cases actual NVRs or even DVRs have a better architecture.
Centralization - Now we start to consider the "brains" of the operation. Consider if the system will have a database for all the configurations (framerate, resolutions, bitrates, alarm configuration, integrations, etc.). By centralizing this, it allows the end user to build many different configuration and push them out to individual recorders, encoders, IP cams, etc.
Workstation Application - Next is the software application for reviewing video. Most Enterprise Video Platform providers have their own ready made version of a video retrieval software. Other rely more heavily on integration software applications (like Orsus or Proximex) to give a central retrieval app for all users.
IP cameras - Now that the legacy cameras are taken care of by the encoders above, the end user can consider the plethora of IP cameras in the market place today (Axis, Sony, Verint, IQInvision, Arecont to name a few). Some serious thought needs to go into this section as much of the market today is going toward Multi Mega Pixel Cameras for replacement of multiple Standard Def cameras.
Video Walls (IP and Analog) and other display techniques - Next the End User may have a need to build a Security Ops Center with an existing Analog Video Wall, or even a newer IP based video wall (like Barco or Christie). Knowing how to incorporate a truly digital IP based video wall or even a simple guard shack environment becomes a significant consideration. Will that older guard who's used to using that Pelco Keyboard really be excitied about now having to use a workstation and mouse to watch video? Perhaps not!
FINALLY, we get to Analytics... Loitering, Tripwires, People Counting, Object Tracking, Objects left behind, Objects taken away, License Plate Recognition, POS Trending, Slip and Fall and Endcap Fullness. So many options, and even within each category there are even more SUB-categories. Day and Night time views can change the configuration depending on the location of the sun. Objects blocking the scene can ruin a view at any time, if this is not considered up front. The main point here is, as many reading this article already know, Analytics can be extremely complex to understand and plan for...
So, I guess I look at it this way, BEFORE one even introduces Analytics to the project, there should FIRST be a discussion of the video platform, IP Cameras and more, upon which the Analytics will reside. Then take that customer confidently into the exciting realm of analytics.
Just food for thought...
Matt Marshall runs a thoughtful blog, Security Caffeine.