Masks that cheat on face recognition algorithms

Artist Sterling Crispin (Sterling Crispin) continues to replenish the collection of data masks on his website, created specifically to deceive face recognition algorithms in outdoor surveillance cameras.

Computer vision programs after training on large collections of photos produce specific features that distinguish one object from another. These signs can be calculated by combining the algorithm for generating random images with the pattern recognition system.

From an artistic point of view, the project symbolizes a protest against the system of total surveillance, which has recently become more and more effective. Outdoor video cameras gradually merge into a single system, tracking people throughout the city and “passing” them along the chain to each other. This is accompanied by a rapid increase in the number of video cameras. According to Sterling Crispin, the process has a negative impact on society.
')
Generating such data masks, the artist wants to visually show exactly how computers perceive people's faces.



Technically, the generation of masks takes place using genetic algorithms that transform the 3D form until it meets the requirements of the face recognition program.





“These masks are intended to be used in acts of protest, poetry, civil disobedience and shaman rituals by citizens of our global village, which is increasingly covered by the technosphere,” writes Crispin.

See also:
Camouflage against CCTV