This week the Metropolitan Transportation Authority (MTA) of New York City announced a $212 million contract to install surveillance cameras in subway stations and also to upgrade cell phone reception in those areas. I am more interested in the justification for the former since I have done research in this area. Simply put, the claims made by the MTA official regarding the cameras were not made with any understanding of the technology. One of the special features of the cameras touted by the MTA rep is the ability to detect abandoned packages. She also said, “We hope (this) will detect the terrorists before an incident happens...”

Technology has become our magic.

My issues with the reasoning here are as follows:

1. Detecting Terrorists
   Terrorists do not wear uniforms. They look just like everyone else. Yes, a suicide bomber might wear a jacket in summer to hide his vest, but that is not detectable through an automated visual system.
2. Response Time
   Let's say for argument's sake that the system could automatically detect a suicide bomber. What can be done about it? They only way to prevent him from blowing himself up in the station is to prevent him from gaining access to the station. The cameras are in the station.
3. Detecting Abandoned Packages
   This is a really hard problem. People are actively researching it. With controlled lighting, little pedestrian traffic, and high contrast between the object and the background, we can do a decent job of this with computer vision techniques. Any attack would probably focus on a busy station. Plus, you can defeat this by putting the bomb in a garbage can or any number of others ways.
4. False Alarms
   This has been the bugaboo of many automatic video surveillance systems whether using motion detection or face recognition. Many objects get left on platforms. It could be the lunch bag from McDonald's or the suitcase of the confused tourist. Other possible detections could be from a partially obscured person standing still for a long time or a poster falling on the floor. Every time there is a detection, someone has to investigate — most likely bomb teams at first. Eventually, this is so much of an annoyance that this part of the system gets turned off and you're left with a conventional video surveillance system that is dependent on operators.