how do i load weights for layers which the two layers have different names？

[曹浩宇]

I want to use caffe to implement two network fusion， it means each network receive a different image，and via a set of networks（such as vgg），so the layer name of two network is same，how do i load two trained vgg model for initialize each network？