ACLU Fooled by Vendor PropagandaAuthor: John Honovich, Published on Feb 07, 2011
The ACLU's surveillance concerns are based on technology that does not work in the real world. Ironically, the ACLU is indirectly promoting technology that surveillance professionals know is little more than smoke and mirrors. The ACLU would be more accurate and helpful attacking the ineffectiveness of this technology and the wasted money spent on it.
Specifically, the ACLU's report charges that 3 technologies used in public surveillance are invasive:
- PTZ Cameras
- Facial Recognition
- Automatic Tracking of Objects Across Cameras
While PTZ cameras are mature technology used by everyone, everywhere, the other two technologies do not work in public surveillance. Indeed, vendors would love that it did but even they know it does not.
Update: The City of Chicago responded, denying they use facial recognition or automatic tracking of people. This is not surprising as, even if they did, the technology would work terribly in their environment.
In pilots and small scale controlled deployments, facial recognition and automatic tracking using public surveillance cameras can be done with moderate accuracy. However, once you start dealing with lots of objects, lots of cameras and various lighting scenes - critical in public surveillance such as Chicago - accuracy problems go through the roof.
For facial recognition, in public surveillance, you need to identify a single person out of millions with an image that is routinely poor (with sharp angles of incident from the camera and lighting issues impacting quality - bright sunlight during the day, too dark at night). Even facial recognition for verification in controlled environments is still questionable and that's much much easier to do than public surveillance facial identification.
Tracking people and cars across cameras suffers from similar challenges. In a public environment, you can have dozens of people and cars moving simultaneously who 'look' very similar. They can be obstructed by larger objects like poles and buildings. The camera's view may be impacted by bright sunlight or shadows that further obscure people or cars. Sizable gaps frequently exist between camera coverage where objects can easily be lost. It's just not possible for computers to do this with any degree of accuracy at scale.
Any video surveillance professional signing on to deliver such systems would need a death wish. Indeed, it's well accepted within the industry that video analytics is still struggling to provide basic detection and cannot deliver these magical solutions.
Worse, since 9/11, governments have spent hundred of millions on video analytic systems that have not worked.
Could this technology be invasive? Sure, if they worked but that's not the problem we are faced with today in 2011.
Related Reports on Facial Recognition
Most Recent Industry Reports
The world's leading video surveillance information source, IPVM provides the best reporting, testing and training for 10,000+ members globally. Dedicated to independent and objective information, we uniquely refuse any and all advertisements, sponsorship and consulting from manufacturers.