Facebook photos serve to fool facial recognition

Authentication mechanisms are evolving. After the passwords, are increasingly emerging systems using the own user to authenticate . But these systems are still very rudimentary, a new study showed that , using simple Facebook photos , can be deceived.


It was only a theoretical , but researchers at the University of North Carolina were able to prove , in the Usenix security conference that facial recognition systems can be easily circumvented and deceived. Using the simple image of Facebook, available publicly on this network and have created 5 profiles for each of the volunteers, it was possible to create three-dimensional images that were used to fool some of the facial recognition systems , accessible in some applications for mobile devices.
Created these five models , four were able to circumvent the system and access the data that was protected. These systems are in use in which any user can find applications in both the Apple App Store and Play Store Google . Success rates were between 55 % and 85 %, which demonstrates that even though not a completely reliable way, since allows very high success rates. To create the models were used 20 volunteers, which were sought images on Facebook. These were between 3 and 27 Images available and that were used to create 3D models used to validate the authentication system by face recognition.
Despite being a very high value , the researchers warned that it was still early for panic. His idea was just to show the vulnerabilities of these systems and to initiate strengthening and improving the algorithms that validate users , so they are more robust and can avoid these simple techniques . It is still a new area and requires many improvements , and the users must trust still so reserved in the security guarantees they offer.


Thank you for being here
Like always see you next time
Gaming&Tech