Caffe 入坑3 ： 定义网络

caffe学习笔记

1. 生成caffe网络的配置文件，prototxt：

def create_net():
    net = caffe.NetSpec()

    # LeNet structure
    # define data layer
    net.data, net.label = caffe.layers.Data(source="data.lmdb",
                                            backend=caffe.params.Data.LMDB,
                                            batch_size=32,
                                            ntop=2,
                                            transform_param=dict(crop_size=40,mirror=True)
                                            )

    # define conv layer
    net.conv1 = caffe.layers.Convolution(net.data,num_output=20,kernel_size=5,
                                         weight_filler={"type":"xavier"},
                                         bias_filler={"type":"xavier"})

    # define activate layer
    net.relu1 = caffe.layers.ReLU(net.conv1,in_place=True)

    # define pooling layer
    net.pool1 = caffe.layers.Pooling(net.relu1,pool=caffe.params.Pooling.MAX,
                                     kernel_size=3,stride=2)

    # define conv layer
    net.conv2 = caffe.layers.Convolution(net.pool1, num_output=32, kernel_size=3,
                                         pad=1,
                                         weight_filler={"type": "xavier"},
                                         bias_filler={"type": "xavier"})

    # define activate layer
    net.relu2 = caffe.layers.ReLU(net.conv2, in_place=True)

    # define pooling layer
    net.pool2 = caffe.layers.Pooling(net.relu2, pool=caffe.params.Pooling.MAX,
                                     kernel_size=3, stride=2)

    # define fc layer
    net.fc3 = caffe.layers.InnerProduct(net.pool2,num_output=1024,
                                        weight_filler=dict(type='xavier'))
    # define activate layer
    net.relu3 = caffe.layers.ReLU(net.fc3, in_place=True)

    # define dropout layer
    net.drop = caffe.layers.Dropout(net.relu3, dropout_param=dict(dropout_ratio=0.5))

    # define output layer
    net.fc4 = caffe.layers.InnerProduct(net.drop, num_output=10,
                                        weight_filler=dict(type='xavier'))

    # define loss
    net.loss = caffe.layers.SoftmaxWithLoss(net.fc4, net.label)

    with open("net/tt.prototxt",'w') as f:
        f.write(str(net.to_proto()))

2. 利用可视化工具查看生成网络的结构：

打开： http://ethereon.github.io/netscope/quickstart.html

点击： Editor /  Launch Editor , 进入页面后，将生成的 prototxt 文件复制到左侧， 按 shift+ enter , 即可查看网络结构。