University of Toronto Engineers Disable Facial Recognition

University of Toronto engineering researchers led by Parham Aarabi and Avishek Bose recently developed an adversarial neural-network filter that disrupts automated facial recognition. Tested on the 300-W dataset of over 600 faces, the method reduced detectable faces from 100% to 0.5%, altering pixels imperceptibly to humans while foiling detectors. The team plans to make the privacy filter available as a mobile or web app.