A team of researchers from Mu'tah University has created a means of identifying masked terrorists through their hand gestures using an artificial intelligence (AI) algorithm in combination with biometrics. In particular, this new method uses the "V" signs that terrorists make while posing for photos or videos in order to reveal the identities of subjects.
"Sometimes the only evidence to identify a person is their hand geometry, and not the whole hand -- only two fingers (the index and the middle fingers) while showing the victory sign, as seen in many terrorists videos," the researchers wrote.
Biometrics - the use of human characteristics as a measurement - is currently used in many technologies, including voice recognition and fingerprint scanning, as well as facial recognition on Facebook. For the current research, the team focused on the geometry of the human hand and how these unique measurements can aid in the identification of people.
Using a sample of 50 men and women, the team photographed each subject's hand numerous times and created a database of 500 images. The photos were then fed to an AI algorithm, which analyzed each hand, with particular focus on the endpoints of the two fingers in the "V" shape and the lowest point of the valley in between these fingers and two points on the palm.
Although the data set used in the current study is small and the technique will likely need to be used in combination with other information in order to be truly effective, the team achieved an accuracy of over 90 percent in some cases, a number that could be improved with further research and the addition of more metrics.
The "V" sign commonly used in photo and videos by terrorists in desert or army fatigues stands for victory, and it is typically flashed while the perpetrator - whose face is covered with a scarf or hood - stands over a decapitated body. The increased prevalence of this sign in photos and videos, as well as the lack of other identifying information on the culprits, are pushing law enforcement agencies to look for other means of pinpointing their identities.
The findings were released on Feb. 26 on the airXiv server.
