According to Biometric Update, facial images on T-shirts can be subject to counterfeiting. T-shirts have become a threat to facial recognition, but a new study shows how to prevent it!
Discussions about biometric attacks typically focus on financial fraud attempts, but the increasing use of facial recognition in public places has prompted researchers to develop ways to trick the technology into bypassing security or surveillance. One method that has proven effective in controlled experiments involves images of attackers on T-shirts. The latest development in this field involves replacing images with face-presentation attacks: a T-shirt with a printed human face is presented to the camera, fooling the facial recognition system into believing it’s seeing a real, three-dimensional face.
A new paper by a group of researchers at the University of Applied Sciences in Germany presents a way to prevent these very facial attacks.
The researchers tested 3 widely used open-source face detection algorithms: RetinaFace, MTCNN, and dlib, against the T-shirt Face Presentation Attack (TFPA) database. The database contains over 1,600 images from 100 different T-shirts, each with a face printed on it.
8 people wore T-shirts with their faces printed on them in various poses, and their images were captured using a RealSense D435 camera, capable of capturing depth information in 3D images.
In almost all cases, the facial detection algorithms detected the face on the T-shirt. The average estimated detection rate of the three algorithms exceeded 99% for all eight poses, according to the results. The study also found that if the attacker concealed the face by covering it with hands, wearing a face mask, or tilting the head, the facial biometric system would likely return a match to the T-shirt—meaning the attack would be successful.
The success of this presentation attack is concerning because T-shirts are easy to create. They can also be concealed under a jacket, making them easier to use under surveillance than something conspicuous, such as a paper mask. T-shirt attacks have already been identified by border authorities as a potential threat.
To address this issue, the researchers expanded the database to include 152 authentic presentations and proposed a new detection method. According to the research results, the proposed algorithm can be easily combined with traditional presentation attack detection algorithms.
Text based on https://www.biometricupdate.com/202605/t-shirts-have-become-a-facial-recognition-threat-a-new-study-shows-how-to-stop-it
