How facial recognition technology works
To begin with here is a little bit of entertaining statistics to understand the question. Modern computer technology is capable of recognizing a person from a photograph or video recording with an accuracy of 97.5 to 99.6%.
A facial recognition system in a major cities of about 15 million inhabitants processes a billion images from tens of thousands of video cameras in just half a second. Banks are collecting biometric data from customers with the prospect of soon switching to Facepay technology – ATM services without using a plastic card – identification through facial scanning.
According to the developers, neither a change of hair style, nor a medical mask, nor even sunglasses will interfere with the recognition of the client’s face. Even actor’s makeup will not be able to mislead the system. Progress was spurred by the sharp increase in the power and speed of computer processors. And social networks became a testing ground for the development and improvement of recognition algorithms.
The role of social media in biometrics
Nowadays users post billions of photos and videos of themselves and their loved ones on Facebook and especially on Instagram. But for social media the process of developing facial recognition would take decades.
Facebook alone has provided an endless test database for biometrics:
- First, the system must select (fix) an object similar to a face within the general field of perception. To solve the problem, the system uses the characteristic features inherent to a human face. The more features are detected in a picture, the higher the probability of identifying the target as a subject.
- Second step is fixation of anthropological points on the face that can be used to identify the individual. These are distance between the pupils of the eyes, size of the ears, lips, nose, and everything else.
- In the third step, the algorithm restrict and crops the face area, eliminating the influence of irrelevant parameters and factors.
- Now the system can start comparing the collected measurements with the data of people who have previously undergone the biometrics procedure.
The rest is just a matter of technique. Finding a specific person using the available list of parameters is no more difficult than finding the answer in a Google keyword search. Only in this case, instead of key phrases, we use graphical data. As you can see, facial recognition technology in terms of algorithms is built quite simply.
Practical implementation of the facial recognition technology
Fans of conspiracy theories scare ordinary people that the face identification technology was invented for total surveillance and the establishment of digital slavery. In reality, neither of these makes any practical or economic sense.
The media is already capable of controlling people’s consciousness at a sufficient level. Of course, the task of finding criminals and terrorists is extremely important. But efforts to develop facial recognition systems are first and foremost of economic importance.
The introduction of such systems will save millions of dollars spent on bank cards and badges at businesses. No identification documents will be needed with the widespread introduction of biometric identification. In any institution a visitor just needs to smile into a video camera and it will 100% replace the act of presenting a document.
In terms of reliability, it is difficult to imagine how criminals could imitate the thousands of parameters by which an individual’s face is recognized. Computerized facial recognition systems would help to eliminate fraud if not entirely, then at least reduce cheating to a reasonable minimum.