caffe 编译

1. Update the source list of APT

  sudo add-apt-repository universe
  sudo add-apt-repository multiverse
  sudo apt-get update

2. Install CUDA-7.0 toolkit

  sudo dpkg -i cuda-repo-l4t-r23.1-7-0-local_7.0-71_armhf.deb
  sudo apt-get update
  sudo apt-get install cuda-toolkit-7-0
  export LD_LIBRARY_PATH=/usr/local/cuda-7.0/lib/

3. Install OpenCV for Tegra

  sudo dpkg -i libopencv4tegra-repo_2.4.12.2_armhf.deb
  sudo apt-get update
  sudo apt-get install libopencv4tegra libopencv4tegra-dev

4. Install cnDNN

  sudo tar xvf cudnn-7.0-linux-ARMv7-v4.0-prod.tgz
  sudo cp -a cuda/lib/libcudnn* /usr/local/cuda/lib/
  sudo cp cuda/include/cudnn.h /usr/local/cuda/include/

5. Setup Caffe environment

  a) Install related packages
     sudo apt-get install libboost-all-dev libprotobuf-dev libleveldb-dev libsnappy-dev
     sudo apt-get install libhdf5-serial-dev protobuf-compiler libgflags-dev libgoogle-glog-dev
     sudo apt-get install liblmdb-dev libblas-dev libatlas-base-dev

  b) Download Caffe source code
     Download caffe source package from https://github.com/BVLC/caffe/, click "download zip" to
     download caffe-master.zip, unzip the package in Jetson /home/ubuntu/Work/caffe/.
     sudo mkdir /home/ubuntu/Work
     sudo mkdir /home/ubuntu/Work/caffe
     sudo cp caffe-master.zip /home/ubuntu/Work/caffe/
     sudo cd /home/ubuntu/Work/caffe/
     sudo unzip caffe-master.zip

  c) Compile Caffe source code
     cd /home/ubuntu/Work/caffe/caffe-master
     Edit the file Makefile.config.example and un-comment the following line, then CAFFE can use
     cuDNN acceleration. Save the file as Makefile.config.
        USE_CUDNN := 1
     cp Makefile.config.example Makefile.config
     make -j4
     When it's compiled successfully, build/lib/libcaffe.so will be generated.