VideoIQ's Flawed NYC Anti-Terrorist 'Plan'By: John Honovich, Published on May 20, 2010
Trying to get attention from the May 2010 NY terrorist attempt, video analytic vendor VideoIQ offers an unrealistic and logistically flawed solution. VideoIQ recommends:
- "A video analytics system would have quickly analyzed that there was a parked car in the area that shouldn’t have been there, and would have immediately notified authorities."
- "A video analytics system also has intelligent search capabilities, which would have allowed operators to automatically retrieve all images containing the suspect and/or his car across all the available cameras. This would have allowed them to recreate the path of the car-and-driver in minutes -- without any manual search -- reducing the forensic investigation from hours or days to just minutes."
Here are the issues we find:
- Lots of Nuisance notifications: Many many people park in restricted areas throughout the day in NYC. About 1 in 100,000 of them are terrorists. Are you going to check each car? Is it worth it to check each car?
- Avoid by parking in grey or approved areas: Many locations in NY have loading zones that allow vehicles to temporarily park. This would allow terrorists to successfully drop off bombs without the video analytic system triggering.
- Small camera coverage area: NYC streets are crowded and have lots of obstructions. Each camera would like cover a small segment of one side of a street. Covering a single side street could cost $20,000+ for the basic equipment.
- Intelligent search is a proprietary feature: You cannot mix and match intelligent search from 3 VideoIQ cameras with 2 Bosch encoders and 1 3VR recorder. This forces you to use a single vendor (good for VideoIQ, bad and unrealistic for users where numerous products are already in use).
- The retrieval claim is highly misleading: Search is very dependent on light levels, angles of cameras, width of FoV, etc. Plus, crowded areas like NYC are especially challenging for search because there are so many objects that 'look' similar. Even if you replace or add VideoIQ to all your existing cameras, you would still have these important issues.
- Search not the most basic problem: You cannot recreate the path of the driver in minutes because the cameras are controlled by dozens of different systems owned by numerous businesses that cannot be accessed 'instantly' by the police.
While poor security design, vendors jump on terrorist events because it is good (short term) business. The VideoIQ 'plan' is likely to waste significant money and have deep operational flaws. However, governments subsidize knee jerk reactions which can be deeply profitable for security vendors.