Caffe Learning Series (i) Ubuntu16.04 build Caffe environment and run mnist example (CPU only)

Objective:

Body:

1. Install the necessary dependent packages:

sudo Install libprotobuf-dev libleveldb-dev libsnappy-dev libopencv-dev libhdf5-serial-dev protobuf-Compiler  sudoinstall --no-install-recommends libboost-all-devsudo  Install libatlas-base-devsudoinstall python-devsudo Install Libgflags-dev libgoogle-glog-dev Liblmdb-dev

Python requires version 2.7, which is already installed by the operating system itself. The input python2.7--version will display the specific version number instructions installed.

1. Download Caffe:

First install the Git tool:

sudo Install git

Then download the Caffe package:

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

Next go to the Caffe directory (I said Caffe downloaded to/home).

CD caffe/

Go to the Python folder inside and enter

 for  in $ (catdoesinstall done

If you don't have a pip, install it first: sudo apt install python-pip

2, installation Caffe

Enter the Caffe root directory and use the template to write a makefile.config. The specific is to copy the template first, and then change some content

CP Makefile.config.example Makefile.config

First, #CPU mode, in the cpu_only: = 1 before the # to be removed. (Because running Caffe only under the CPU)

Then, in which to find the following, two paths to change to this: (Add the following two hdf5 path, otherwise compile times HDF5 error)

# Whatever Else you find your need goes here.

Include_dirs: = $ (python_include)/usr/local/include/usr/include/hdf5/serial

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

Now you're ready to start installing Caffe:

Execute the following command in turn:

 Make Pycaffe  Make  All  Make Test  make Runtest

When I executed the Make pycaffe command, I made a mistake: fatal error:numpy/arrayobject.h didn't have that file or directory

In fact, NumPy is already installed, anaconda2 inside, Python import numpy also no problem, but is here error, solution:

sudo Install Python-numpy

然后再执行一遍 make pycaffe命令就行了。

Show all tests PASSED, instructions already installed

3, Training mnist set

Locating the terminal to the Caffe root directory
Then execute the script file, download the Mnist database, and unzip

./data/mnist/get_mnist. SH

I do not receive HTTP due to network problems, so according to the code in the script, according to the Web site to download the compressed package, CP to the Mnist folder, using the decompression commands in the script to extract.

Then, convert it to the LMDB database format

./examples/mnist/create_mnist. SH

Then train the network

./examples/mnist/train_lenet. SH

When you are training, you can see the loss and accuracy values

You can see that the final training accuracy is 0.9911.

Completed successfully.

Share only for Better records ~

by Still,

Caffe Learning Series (i) Ubuntu16.04 build Caffe environment and run mnist example (CPU only)