I\'m looking at implementing a Caffe CNN which accepts two input images and a label (later perhaps other data) and was wondering if anyone was aware of the correct syntax in the
You may also consider using HDF5_DATA layer with multiple "top"s