Demo - False Fingerprint Detection

Using images from the testing datasets

  • Each time the sensor/dataset is changed a different image is selected.
Classifier trained using LivDet2013 - Swipe dataset
Fake fingerprint CORRECTLY classified as Fake in 0.131295 seconds

Fake Fingerprint

Using images from an URL:

Prediction from classifier trained using dataset LivDet2009-Identix: True fingerprint
Prediction from classifier trained using dataset LivDet2011-Digital: False fingerprint
Prediction from classifier trained using dataset LivDet2011-Sagem: True fingerprint
Prediction from classifier trained using dataset LivDet2013-Swipe: False fingerprint

Image from URL

The paper based on this work can be found here.

The results and best hyper-parameters and architectures used in each dataset can be found here.