Skip to content

Instantly share code, notes, and snippets.

@microhello
Last active August 29, 2015 14:02
Show Gist options
  • Save microhello/971a68efa73066d450df to your computer and use it in GitHub Desktop.
Save microhello/971a68efa73066d450df to your computer and use it in GitHub Desktop.
Caffe Install and Config
一、Caffe安装过程:
0、安装Ubuntu 12.04 LTS_x64 虚拟机
1、根据http://caffe.berkeleyvision.org/installation.html此处的安装说明,
分别下载caffe-master.zip、cuda_5.5.22_linux_64.run、 l_mkl_online_11.1.0.080.sh;
2、安装cuda 5.5
sudo ./cuda_5.5.22_linux_64.run
sudo ldconfig /usr/local/cuda/lib64
3、 安装glog:
wget https://google-glog.googlecode.com/files/glog-0.3.3.tar.gz
tar zxvf glog-0.3.3.tar.gz
cd glog-0.3.3/
./configure
make & make install
ldconfig
4、安装MKL:
sudo sh l_mkl_online_11.1.0.080.sh
sudo ldconfig /opt/intel/mkl/lib/intel64
5、安装boost、opencv、leveldb、protobuf
sudo apt-get install libprotobuf-dev libleveldb-dev libsnappy-dev libopencv-dev libboost-all-dev libhdf5-serial-dev
6、安装caffe,配置环境变量
cd caffe-master
cp Makefile.config.example Makefile.config
make
make test
make runtest
export CAFFE_ROOT=/home/zxy/Downloads/caffe-master/
二、Caffe自带lenet训练
1、下载并转换数据集为leveldb格式
cd $CAFFE_ROOT/examples/lenet/
cd $CAFFE_ROOT/data/mnist
./get_mnist.sh
cd $CAFFE_ROOT/examples/lenet
./create_mnist.sh
会生成两个文件夹mnist-test-leveldb、mnist-train-leveldb
2、使用CPU模式、减少迭代次数
修改 $CAFFE_ROOT/examples/lenet/lenet_solver.prototxt,
# The maximum number of iterations
max_iter: 2000
# solver mode: 0 for CPU and 1 for GPU
solver_mode: 0
3、训练
./train_lenet.sh
生成结果文件:lenet_iter_2000
4、测试
编写测试脚本test_lenet.sh,内容为:
#!/usr/bin/env sh TOOLS=../../build/tools GLOG_logtostderr=1 $TOOLS/test_net.bin lenet_test.prototxt lenet_iter_2000 2000 CPU
运行结果: Test accuracy:0.9859
三、基于Caffe的开发
1、使用Eclipse做为开发工具,Java做为开发语言
2、采用在Java中调用shell命令行(train_lenet.sh,test_lenet.sh)的方式
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment