Caffe Installation (9): Caffe Download and compile

Go to official github to download the Caffe zip file and unzip it

CD to Caffe-master folder, generate Makefile.config configuration file, execute:

$ CP Makefile.config.example Makefile.config

Configure Makefile.config file (only the modified parts are listed)

A. If you enable CUDNN, remove the "#" in front of it

USE_CUDNN: = 1

B. Configure some reference files (the additional part is mainly to solve the problem of the HDF5 path under the new version)

Include_dirs: = $ (python_include)/usr/local/include/usr/lib/x86_64-linux-gnu/hdf5/serial/include

Library_dirs: = $ (python_lib)/usr/local/lib/usr/lib/usr/lib/x86_64-linux-gnu/hdf5/serial

C. Enable Intel Parallel Studio XE (Intel MKL) or Atlas (two previously described)

BLAS: = Mkl

or Blas: =altas

D. Configuring the path for Caffe support for Python and MATLAB interfaces

Python_lib: =/usr/local/lib

If the use of Anaconda, etc., change to the corresponding path. The use of Anaconda is not recommended and is prone to problems.

Matlab_dir: =/usr/local/matlab/r2014a

C. If OpenCV 3.0 is enabled, remove the "#"

Opencv_version =3

I'm using opencv-2.4.9, so I'm not canceling the comment here.

D. Use_python_wrapper Uncomment

Compiling Caffe-master

"-j32" is compiled using multi-core CPU, which can greatly accelerate the speed of compiling, it is recommended.

$ make All–j32

$ make Test–j32

$ make Runtest–j32

If make fails to be re-make, first

$ make clean

Compile the caffe files used by Python and matlab

$ make Pycaffe–j32

$ make Matcaffe–j32

Caffe Installation (9): Caffe Download and compile