verification task and classification task based on siamese network

[kan.wa...@gmail.com]

thanks for your time in advance,here is my problem

     the  input of my network is a pair of images, each of them have a identification label(1-1000), and the pair has another verification label(1 means that the two images belong to the same person and 0 otherwise).

     the output os network consists of two loss, classification loss corresponse to the identification label  and a verificaiton loss.

     my problem is :

     how to construct the data layer with two kinds of label?

thank U