Dual-Domain Face Anti-Spoofing with Integrated Spatial and Frequency Analysis Neural Network

In this paper, we introduce a novel dual-domain approach for face anti-spoofing, aimed at enhancing the robustness and accuracy of spoofing detection mechanisms. Our model uniquely combines spatial domain analysis with frequency domain insights, employing a two-branch and dual loss neural network architecture that leverages augmented images and their Fourier transforms during the training stage. T…