England riots and automatic face recognition
English youth trash shops, and the police can do nothing but study video from street surveillance cameras (fortunately, the UK ranks first in the world in terms of density of surveillance cameras). The police began to upload photos of bullies on Flickr and asks to identify violators. Citizens themselves also take photos from the windows of their houses, so many pictures have already accumulated on Facebook, Flickr and Twitter.
And the work has begun. London Riots Facial Recognition has been formed on Google Groups , which discusses face recognition technologies for processing an array of photos from open sources. One developer offers assistance in developing a program using Face.API. It will succeed or not, but over time, such technologies will certainly be created.
Thus, citizens themselves create tools that can enter the standard arsenal of police bodies for automatic face recognition on street surveillance cameras.
The technology can work in much the same way as automatic video recording systems for traffic violations work now. Radars and cameras hang over the road and automatically detect various violations of the rules (speeding, crossing the solid marking line, parking in the wrong place). Then another camera records the state registration number on the car, it is recognized, makes its way to the base - and a photograph of the car and a receipt of the fine are sent to the owner’s home address. The system works almost completely automatically without human intervention. Accountants who monitor the acceptance of money from the public, and engineers who install and configure new video recording systems, remain in the traffic police. Ideally, cameras should cover the entire territory of the city (for example, only in Moscow they are going to put800 video recording complexes over the next three years ).
Of course, in the case of face recognition, automation will not reach such an extent, because the Criminal Code is much more complicated than traffic rules, and face recognition is more complicated than recognition of state registration numbers. However, all technical problems can be solved over time, so that in the future, the fixing of some violations of the Criminal Code and the Civil Code can also be automated. For example, smoking in a prohibited place, drinking alcohol on the street, obscene language, throwing garbage on the sidewalk, uncleaning feces after a dog, participating in an unauthorized rally, etc. All these minor violations can be recorded, automatically recognize the identity of the offender - and send a receipt for a fine at the place of residence. Thus, public order will be fully controlled by automatic systems, which is much more effective than the work of lazy living cops.
The worst thing about such systems is not even a violation of privacy, which most citizens don't care about, but inevitable mistakes. After all, face recognition does not work as accurately as fingerprints. A person suffering from a mistake will be forced to prove his innocence , which contradicts the principle of the presumption of innocence. As in the case of those photographs with hooligans in the English streets, a person who lives nearby and walked home or tried to escape from hooligans could accidentally get into the frame. Yes, and the face recognition system can also fail and take you for another person (although this problem is solved by automatically tracking a person by a chain of cameras to a place of residence - the same way the “Stream” system works for cars).
Everyone can become a victim of a mistake, but real hooligans will quickly learn how to wear hoods and baseball caps with infrared illumination of their faces to shoot down video cameras.