First place in the LivDet2021 challenge 'Liveness Detection in Action' for the team composed by Eng. Domenico Mattiello, Michela Gravina and Antonio Galli, coordinated by Eng. Stefano Marrone and directed by Prof. Carlo Sansone, members of the PICUS group of the Department of Electrical Engineering and Information Technology - DIETI of the University Federico II.
LivDet is a biennial international competition, open to companies and universities, focused on the detection of counterfeit fingerprints. This year was the edition with more teams participating (23), coming from Italy, Germany, Czech Republic, USA, South Korea and China.
The challenge won by the federiciano team concerned the development of an algorithm for the detection of forged fingerprints (e.g. obtained through casts and/or latent prints of a target subject) within an authentication system based on biometrics.
The proposed solution is based on artificial intelligence techniques, in particular on a combination of Deep Learning techniques with the use of the so-called 'Adversarial Perturbations'. The result is an algorithm that has proved capable of detecting counterfeit prints when the sensor (scanner) and the materials used to make the fakes vary, surpassing by more than 5 percentage points the solution that came in second place, proposed by a Chinese team.
The solution was presented to the scientific community on August 7 during the International Joint Conference on Biometrics (IJCB), the most important international conference in the field of biometric systems, jointly sponsored by the IAPR (International Association for Pattern Recognition) and the IEEE Biometrics Council.
LivDet: https://livdet.diee.unica.it/
La conferenza IJCB2021: http://ijcb2021.iapr-tc4.org/