Description
The MSU SpoofGAN Dataset, was created to synthesize highly realistic, synthetic fingerprint spoof impressions with multiple impressions per generated identitiy. Currently, it consists of 1,500 unique, syntheticly generated fingerprint identities with 3 impressions of each spoof material type (6 spoof types plus live), for a total of 31,500 images. Fig. 1 presents some example SpoofGAN fingerprint images. The database was produced at Michigan State University’s Pattern Recognition and Image Processing (PRIP) Lab, in East Lansing, MI, US. This dataset may be augmented later with additional synthetic fingerprint identities and more spoof types.
Figure 1: Example images of multiple impressions of the same finger generated by SpoofGAN. (a) and (b) show three impressions each of two fingers rendered in a bona fide style, whereas (c) and (d) show three impressions each of the same two fingers in a PA style (ecoflex and body double, respectively)..