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

Table 4

The modified VGGNet16 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
Conv2	64
Maxpool1	1
Conv3	128
Conv4	128
Maxpool2	1
Conv5	256
Conv6	256
Conv7	256
Maxpool3	1
Conv8	512
Conv9	512
Conv10	512
Maxpool4	1
Conv11	512
Conv12	512
conv13	512
Maxpool5	1
Linear (25088, 512)		512