Dataset

Training Dataset: Datasets from Iris LivDet previous competitions will made available for training algorithms. In addition any publicly available dataset can be used to train the algorithm.

Testing Datset: One dataset will be offered as part of Iris LivDet 2020. It has a total of 10573 samples (live: 4418, spoof: 6155) with resolution of 640 x 480. The dataset has 5 major spoof categories as detailed below. A small subset of samples from this dataset will be shared with all participants as an overview of the dataset.

ClassAttack TypesSample CountSensorMiscellaneous Information
Live4418
Spoof Printed Eyes 1049 Iris ID iCAM7000 Printers Used:
Epson Stylus Pro 9900
HP M652
Xerox C60
OKI MB-471
Cannon Super G3
Paper Quality:
Matte
Gloss
SpoofTextured Contact Lens 3,395 LG 4000 AD 100,
Iris ID iCAM7000
SpoofElectronic Display
81
Iris ID iCAM7000 Display Type:
Kindle
Spoof Fake/Prosthetic Eyes and printed with add-ons 541 Iris ID iCAM7000
SpoofCadaver Iris1089 IriTech IriShield

If you would like to receive the training datasets, please Register. If you have already registered once, please do not re-submit the registration form. We will contact you shortly. If you have any questions please feel free to email us at datasets@livdet.org