Communications of the ACM
How to Spot Deepfakes? Look at Light Reflection in the Eyes
University at Buffalo computer scientists have developed a tool that automatically identifies deepfake photos by analyzing light reflections in the eyes.
Credit: www.thispersondoesnotexist.com, University at Buffalo.
A tool developed by University at Buffalo computer scientists can automatically identify deepfake photos of people by analyzing light reflections in their eyes for minute deviations.
The tool exploits the fact that most artificial intelligence (AI)-generated images cannot accurately or consistently reflect the image of what the pictured person is seeing, possibly because many photos are combined to create the fake image.
The tool first maps out each face, then analyzes the eyes, the eyeballs, and finally the light reflected in each eyeball.
The tool was 94% effective in spotting deepfakes among portrait-like photos taken from actual images in the Flickr Faces-HQ dataset, as well as fake AI-generated faces from the www.thispersondoesnotexist.com repository.
From UB News Center
View Full Article
Abstracts Copyright © 2021 SmithBucklin, Washington, DC, USA
No entries found