- Facial Recognition
Facial recognition is to human faces what ANPR is to vehicle number plates: it enables a CCTV system to match a person’s face with an internal database record, telling you *who* just passed before the camera.
Facial recognition is a very powerful CCTV feature, which is used in many settings where security or customer service are of utmost importance. It can be used, for example, to spot and track VIP customers in hotels or casinos — or, equally well, to recognize people who are no longer allowed to be on a business’ premises, or suspected criminals.
CCTV-based facial recognition is a very complex problem, though, and this has two importance consequences.
The first one is price. Facial recognition is a premium feature; manufacturers may not necessarily charge extra for it, but the hardware that supports is on the pricey side of the budget divide.
The second one is accuracy. Recognizing human faces is such a complex task that even biological systems sometimes fail.
Even we humans will sometimes mistake a stranger for an acquaintance, for example. Computers are even worse at it. They will get things right most of the time, but will hit both false positives (confuse a stranger with someone known to them) and false negatives (failing to recognize someone that is known to them).
CCTV facial recognition is not yet at the level when you can rely on it entirely. It can make your staff’s work considerably easier, but it cannot replace it.