8 secrets to installing facial recognition cameras intelligently

8 secrets to installing facial recognition cameras intelligently

Face ID keeps cities safe and helps businesses identify and analyze their customers. But it is not enough to buy equipment; you need to install video surveillance with elements of artificial intelligence correctly.

How to create conditions for identification?

I want to note that even “smart” cameras need to be set up in order to unleash their potential. Details are below. 

8 секретов

So, let’s list the ideal conditions for high-quality recognition of human faces by IP video cameras:

  1. The passage corridor should be narrow, people should walk straight. This will help the camera to read faces better and not be sprayed over a wide-angle of view.
  2. Ideally, there should be no more than three faces in the frame at a time. This increases the percentage of identification.
  3. The camera must be installed so that it looks directly at the faces of people passing by, neither the side nor the back, namely forward.
  4. Do you remember stills from movies where the camera picks out a face from a crowd in seconds? Unfortunately, technology has not yet reached this level. Camcorders cannot read faces in a crowd. People should walk calmly and not congregate.
  5. The monitored object must be well and uniformly illuminated. In the dark, the face recognition system will not cope.
  6. The device should be installed approximately at one level with people’s faces. This will increase the percentage and quality of recognition.
  7. The device must shoot at a minimum of 5 frames per second.
  8. The human face in the camera must be read at least 160 pixels.

At first glance, it may seem that system configuration is very complex. But there is a simple solution from a Faceter service. It is highly customizable and has both hearing and intelligent vision. The identification percentage is 98%.

Law 

According to article 152.2 of the Civil Code of the Russian Federation, video surveillance in public places is allowed, and the consent of people is not required if the shooting is carried out in the state or public interest.

If the cameras are mounted in private facilities, you must hang an information plate next to the equipment.

You cannot install the camcorder only in the following cases:

  • installation in the toilet / bathroom / shower room, changing rooms, changing rooms, and other similar rooms is prohibited;
  • if the surveillance is carried out covertly;
  • if there is no notification of the shooting in progress near the equipment;
  • video surveillance is aimed at spying on people’s private homes and their real estate.

Read more about the nuances of installing cameras in our article “Ban on video surveillance. What is it?“.

How to sabotage the system? 

The best facial recognition systems have a very low error rate. But the pandemic and the need to wear masks have greatly affected the technology’s reputation.

In fact, the system needs 50-60% of a person’s open face to identify one’s identity. But for a more precise result, other conditions must also be met:

  • absence of unnecessary details on the face. For example, dark glasses, long bangs, a hood, or a scarf can trick the system;
  • the location of the human head. The person’s gaze should be roughly focused on the camera. If the eyes are looking down alongside the mask on the face, then the camera can miss a person.

For more information on how Face ID works with masks on people’s faces, read this article, “Practices of face recognition in masks. Does it work or not?

Summing up

If you correctly approach the system’s installation, you can get a loyal employee working 24/7 at a minimal cost. You should not deny yourself video surveillance with elements of artificial intelligence if you know little about this technology. A lot of useful information is in our blog.

Александр Вебер

Alexander Weber

Specialist in video surveillance, video analytics, cloud storage systems. Consultant on the integration of video surveillance systems and tools in various business sectors. Over 10 years of industry experience.

See also