Analysis of the effectiveness of neural network architectures for Replay Attack detection in the facial biometrics system

The paper deals with the construction of an effective neural network architecture for recognizing spoofing attacks on the facial biometrics system based on the substitution of a real person’s face in the field of view of a video surveillance camera for a video image of another person’s face formed on the screen of a wearable device. A comparative analysis of approaches to the construction of neural network architectures is carried out. Estimates of quality metrics for each approach are obtained.