Using a printed adversarial patch to fool an AI system

By Bob Yirka , Tech Xplore

Left: The person without a patch is successfully detected. Right: The person holding the patch is ignored. Credit: arXiv:1904.08653 [cs.CV]

A trio of researchers at the University of KU Leuven in Belgium has found that it is possible to confuse an AI system by printing a certain picture and holding it against their body as the AI system tries to identify them as a human being. Simen Thys, Wiebe Van Ranst and Toon Goedemé have written a paper describing their efforts and have uploaded it to the arXiv preprint server. They have also posted a video on YouTube showing what they accomplished.

Read more at Tech Xplore