Tutorial: Install Caffe on CentOS 7

Caffe is one of the most powerful framework to train deep neural networks. This tutorial will show you how to install Caffe on CentOS 7 step by step.

I suggest you to apply this tutorial on a computer with a new and clean installation of CentOS 7.

Step 1: Install Basics

Update yum to get last version

sudo yum update

Install gcc compiler:

sudo yum install gcc gcc-c++

Install git, vim, python dev and pip:

sudo yum install git vim python-devel python-pip

Step 2: Install Caffe Dependencies

Install required libraries

sudo yum install protobuf-devel leveldb-devel openblas-devel snappy-devel opencv-devel boost-devel hdf5-devel gflags-devel glog-devel lmdb-devel

Install CUDA

sudo wget http://developer.download.nvidia.com/compute/cuda/repos/rhel6/x86_64/cuda-repo-rhel6-6.5-14.x86_64.rpm
sudo rpm --install cuda-repo-rhel6-6.5-14.x86_64.rpm
sudo yum clean expire-cache
sudo yum install cuda

Step 2b: GPU Support (Optional: ONLY if computer has CUDA compatible GPU)

Note: CUDA only support NVIDIA graphic cards. ATI Radeon GPU can’t be used, you will have to stick with CPU only mode.

Download and Install last NVIDIA Driver for your device

Go to NVIDIA website, download your graphic card last driver and run the driver file to install it

Download and Install CUDNNv3 (You need to register to NVIDIA website to get last version, or just use the mirror provided below)

wget ...
sudo tar -xvf cudnn-7-0.tgz
sudo cp cuda/include/cudnn.h /usr/local/cuda/include/
sudo cp cuda/lib64/libcudnn* /usr/local/cuda/lib64/

Step 3: Get Caffe

Git clone Caffe repository

git clone https://github.com/BVLC/caffe

Step 4: Install Python Dependencies

Caffe has a Python interface for easy scripting, I suggest you to install it.

for req in $(cat caffe/python/requirements.txt); do sudo pip install $req; done

Step 5: Compile Caffe

Move to Caffe folder and copy/paste Make configuration file

cd caffe
cp Makefile.config.example Makefile.config

Edit Makefile.conf

vim Makefile.config

Edit the line

BLAS := atlas

Change ‘atlas’ to ‘open’

BLAS := open

And add a new line under it:

BLAS_INCLUDE := /usr/include/openblas

Then edit line: (located under “PYTHON_INCLUDE := /usr/include/python2.7 \”)

		/usr/lib/python2.7/dist-packages/numpy/core/include

Change python directory to ‘/usr/lib64/python2.7/site-packages/’, the line should then looks like that

		/usr/lib64/python2.7/site-packages/numpy/core/include

Then edit according to your device GPU capability:

Without GPU support:
Edit the line

# CPU_ONLY := 1

Remove the number sign “#”. The line will then looks like this:

CPU_ONLY := 1

Save and Close file

— OR —

With GPU support:

# USE_CUDNN := 1

Remove the number sign “#”. The line will then looks like this:

USE_CUDNN := 1

Start to compile Caffe

sudo make all
sudo make runtest
sudo make pycaffe
sudo make distribute

Done!

Now you can start to try Caffe examples and tutorials.

If you have any question, you can leave a comment.