Skip to main content
. 2017 Jun 6;17(6):1297. doi: 10.3390/s17061297

Table 5.

Comparison of the error rates of finger-vein recognition between the original vein image (case 1 in Table 3) and the Gabor filtered image (case 3) in VGG Net-16.

Method Name Input Image EER (%)
Good-Quality Database Mid-Quality Database Low-Quality Database
B (of Table 4) (VGG Net-16 (no fine-tuning)) Gabor filtered image 1.078 4.016 7.905
Original image 1.481 4.928 7.278
B-1 (of Table 4) (VGG Net-16 (fine-tuning)) Gabor filtered image 0.830 3.412 7.437
Original image 0.804 2.967 6.115