Just like with most technology, facial recognition systems have its pros and cons. The pros is that it can act as a security measure, while some of the cons is that it could be used by companies and software to identify people in photos and videos, something that not all of us feel comfortable with.
However it seems that researchers from the Max Planck Institute for Informatics have come up with what appears to be a pretty effective way of beating such systems. In their paper uploaded to arXiv, the method proposed by the researchers basically tricks facial recognition systems into not being able to identify the person in the photo, while at the same time allowing humans to still recognize who it is.
How this is done is that it recreates the person’s face/head by “painting” it, which fools facial recognition systems, but at the same time the end result is still realistic and accurate enough where you could point at it and say you recognize that person. At the moment one of the more popular ways of obscuring one’s identity is by blurring or adding a mosaic to your face in photos and videos, which the researchers claim is ineffective against more state-of-the-art systems.
Filed in . Read more about Privacy, Security and Social Hit.