02 Jun Researchers Develop Anti Facial Recognition Systems AI
Researchers from the University of Toronto announced this week that they were able to develop an algorithm that can counter facial recognition systems on the fly by using light transformations in images.
According to developers Avishek Bose and Parham Aarabi, spits out a real-time filter that targets individual pixels on an image. They added that it can be applied to any picture and is almost imperceptible to the human eye.
The researchers further revealed that they used adversarial training which is AI composed of two neural networks – a “generator” used to identify faces, and a “discriminator” that disrupts the facial recognition. Aarabi and Bose says they hope to make the neural networks available on an app or website in the near future.