Simon Hall examines the problems of retrospectiveness and poor observation inherent in CCTV systems with manned control rooms. Artificial intelligence (AI), he contends, could be the answer.
We are living in a time in human history where the words ‘security' and ‘prevention' are becoming synonymous. There exist a new wave of criminals, expert in new forms of weaponry and destructive technologies and more insidious and more ingenious in their designs. This frightening reality has created frustration with what many see as a widespread tendency to use CCTV as a source of retrospective video evidence.
Both the NSW Crime Prevention Division in Australia and the United Kingdom Home Office have recently cited academic reports to this effect. Global demand exists for a new level of intelligence in CCTV, one that can play a reliable and preventative role
CCTV used in applications such as street surveillance and critical infrastructure protection, require constant video monitoring by human operators - and therein lies a problem - deploying preventative action is only possibly if the operator sees the incident.
In a study issued by the US National Institute of Justice it was reported that after only 20 minutes of watching video monitors the viewing attention of operators quickly degenerated to a level well below acceptable standards. 
This and other studies highlight the fact that monitoring video screens has a dulling effect on the senses, which leads to distraction.
To counteract this problem, some manufacturers have developed auditing systems which run in the background, mining data such as - mouse clicks, voice commands and software interaction - to critique a video operator's effectiveness. Companies using these systems have found them costly to setup, time consuming to manage and don't address the limitations inherent with using human operators.
So what is the solution?
Imagine a control room operator that can continually scan video from a 100 cameras, simultaneously - immediately alerting to important events, with over ninety percent accuracy. The operator never gets distracted - doesn't sleep, have lunch or break for coffee.
Impossible? For a human, yes! But not for an ‘artificially intelligent operator'.
Recent improvements in computer processing power have enabled AI developers to release a commercially viable ‘artificially intelligent' system commonly referred to as video analytics. Also know as - content analysis software- this exciting new platform uses powerful algorithms to replicate the logical thinking process a human uses to recognise and alert to a particular situation or object.
Video analytics systems can perform superhuman feats above and beyond the capabilities of a human operator. In a recent media demonstration - the software picked out a black briefcase that had been left on a black marble floor in the foyer of a critical site. Looking at the live video feed, the briefcase appeared totally invisible on the black background - however the software identified the case and immediately created an alarm. In this commercial application the software notifies security staff wirelessly using PDA technology.
AI researchers such as ObjectVideo, Clarity and iOmniscient are delivering video analytic and content analysis solutions to organisations who are seeking to add another layer of intelligence to their existing security camera systems. Airports, power stations, sporting venues and others are seeking specialised integrators to assist them in creating preventative security using the power of AI found within video analytics software.
Simon Hall
Technical writer and security journalist
