我有一个如下所示的prototxt文件:

layer {  name: "data"  type: "HDF5Data"  top: "data1"  top: "data2"  top: "label"  include {    phase: TRAIN   }  hdf5_data_param {    source: "./source_list.txt"    batch_size: 2    shuffle: true   }}layer {  name: "concat"  type: "Concat"   bottom: "data1"  bottom: "data2"  top: "data"  concat_param {    concat_dim:1  }}

我想使用caffe的NetSpec在Python中生成上述prototxt文件。然而，我的尝试出现了错误。这是我的代码。请帮我修复它。谢谢

from caffe import layers as L...n = caffe.NetSpec()n.data, n.label = L.HDF5Data(top=["data1", "data2"], batch_size=2,                             source="./source_list.txt", ntop=2,shuffle= True,                             include={'phase': caffe.TRAIN})n.concat = L.Concat(n.data["data1"], n.data["data2"])

回答：

你需要有两个data输出

n.data1, n.data2, n.label = L.HDF5Data(ntop=3, name="data",                                       hdf5_data_param={'source': "./source_list.txt",                                                          'shuffle': True,                                                         'batch_size': 2}                                       include={'phase': caffe.TRAIN})n.data = L.Concat(n.data1, n.data2, name="concat", concat_param={'concat_dim':1})