The study evaluates the impact of synthetic images on Morphing Attack Detection (MAD) using a Siamese network. Results show that training MAD with EfficientNetB0 from FERET, FRGCv2, and FRLL databases reduces error rates compared to state-of-the-art. However, training solely with synthetic images leads to worse performance. A mixed approach combining synthetic and digital images may enhance MAD accuracy. The research highlights the need to include synthetic images in training processes for improved detection capabilities.
In un'altra lingua
dal contenuto originale
arxiv.org
Approfondimenti chiave tratti da
by Juan Tapia,C... alle arxiv.org 03-15-2024
https://arxiv.org/pdf/2403.09380.pdfDomande più approfondite