caffe 07 win10 使用caffe python接口定义网络配置文件

最新推荐文章于 2022-11-13 11:50:10 发布

longji

最新推荐文章于 2022-11-13 11:50:10 发布

阅读量694

点赞数

分类专栏： caffe 文章标签： caffe

本文链接：https://blog.csdn.net/longji/article/details/62069302

版权

caffe 专栏收录该内容

21 篇文章 0 订阅

订阅专栏

本文参考：
http://blog.csdn.net/u011762313/article/details/48213421

在实际项目中，为了自动化调整网络定义，需要脚本化调整网络定义参数。可以使用caffe的python接口生成caffe的网络定义文件。
代码如下：

# D:\git\DeepLearning\caffe\test\write_net.py # 使用python代码生产网络定义文件
import sys
# 引入caffe的python接口
caffe_root='D:/git/DeepLearning/caffe/build/x64/install/'
sys.path.insert(0, caffe_root+'python')
import caffe

# 引入caffe的layers，别名L
from caffe import layers as L
# 引入caffe的params，别名P
from caffe import params as P

# 定义生成网络定义文件的函数create_net
def create_net(lmdb, batch_size):
  # 获取caffe的网络定义命名空间
  n = caffe.NetSpec()
  # 定义DataLayer
  n.data, n.label = L.Data(batch_size = batch_size, 
    backend = P.Data.LMDB, source = lmdb, 
    transform_param = dict(scale = 1./255), ntop = 2)
  # 定义卷积层
  n.conv1 = L.Convolution(n.data, kernel_size = 5, 
    num_output = 20, weight_filler = dict(type = 'xavier'))
  # 定义池化层
  n.pool1 = L.Pooling(n.conv1, kernel_size = 2, 
    stride = 2, pool = P.Pooling.MAX)
  n.conv2 = L.Convolution(n.pool1, kernel_size = 5, 
    num_output = 50, weight_filler = dict(type = 'xavier'))
  n.pool2 = L.Pooling(n.conv2, kernel_size = 2, 
    stride = 2, pool = P.Pooling.MAX)
  # 定义全连接层
  n.ip1 = L.InnerProduct(n.pool2, 
    num_output = 500, weight_filler = dict(type = 'xavier'))
  # 定义ReLU层
  n.relu1 = L.ReLU(n.ip1, in_place = True)
  n.ip2 = L.InnerProduct(n.relu1, num_output = 10, 
    weight_filler = dict(type = 'xavier'))
  # 定义SoftmaxWithLossLayer层
  n.loss = L.SoftmaxWithLoss(n.ip2, n.label)
  return n.to_proto()

# 调用定义函数，生成网络定义文件
# 生成的网络定义文件和本python文件write_net.py在同一个目录
# 生成训练网络write_net_train.prototxt
with open('write_net_train.prototxt', 'w') as f:
  f.write(str(create_net('examples/mnist/mnist_train_lmdb', 64)))

# 生成测试网络write_net_test.prototxt
with open('write_net_test.prototxt', 'w') as f:
  f.write(str(create_net('examples/mnist/mnist_test_lmdb', 32)))

# 在caffe根目录(D:\git\DeepLearning\caffe)下，
# 运行下面语句，生成网络图
# python python\draw_net.py test\write_net_train.prototxt build\x64\tmp\write_net_train_test01.png --rankdir=BT
# python python\draw_net.py test\write_net_test.prototxt build\x64\tmp\write_net_test_test01.png --rankdir=BT