News, Announcements and Events

Testing the impact of cloaked images on face recognition results

Cognitec repeatedly tests software that is supposed to “cloak” images, with changes barely visible to the human eye confusing deep learning methods for face recognition. See two test results below, showing that both cloaking methods caused a slight deviation from a perfect match score of 1.0, but did not change the high probability that the images show the same person.

Face recognition tests with cloaked images

Go back