Finger Vein Verification on Different Datasets Based on Deep Learning with Triplet Loss

Table 3

The modified ResNet18 configuration was used in the research.

	Number of filter	Size of feature map	Size of kernel	Number of stride	Number of padding
Image input layer
Conv1	64
Maxpool1	1
Conv2	64
Conv3	64
Conv4	64
Conv5	64
Conv6	128
Conv7	128
Conv8	128
Conv9	128
Conv10	256
Conv11	256
Conv12	256
Conv13	256
Conv14	512
Conv15	512
Conv16	512
Conv17	512
AdaptiveAvgPool2d
Linear (512, 512)