An effective method for detecting deepfakes is found
A group of scientists has developed a method for recognizing deepfakes by analyzing the shape of a pupillary with a photo or video according to the Screenrant portal.
Experts from the University of Albany, the University of Buffalo and Keya Medical note in their research work that the human pupillary has a round shape. In this case, the artificially created pupillary has an ambiguous distorted geometry.
The researchers note that irregular shape of the pupillary is common even in high-quality deepfakes and are easily distinguishable with the naked eye.
They also created an automated system that analyzes fake images and videos. However, the technology has several disadvantages. So, it does not take into account that some diseases and infections can change the shape of the pupillary.