Intro

In order to extract joints from images, we want to use the code from project deep cut . And the first step is to meet the prerequisites mentioned in its github page. During my configuration process, I met a lot of problems. So I want to write down how I overcome them. 

Steps

1. AWS

Using account asked from Ali, you can log in to AWS account and the dashboard looks like this:

Click on "Launch Instance" you can configure your own instance step by step:

My configuration step by step:

1. Choose ubuntu server 16.04

2. Choose g2.2xlarge

3. Did nothing

4. Change storage size to 30g

5. Lauch

And remember to follow the instruction to get your key file (.pem). Without this file you cannot access your instance. 

2. Connect to AWS

Here I use xshell to connect to my AWS instance. I prefer xshell than putty and you can download free version for personal or school use here. After installation,

1. Open xshell and open Tools->User Key Manager, and import your .pem key

2. Add new session and fill up the IP, port

3. Click "Authentication" and set username to "ubuntu" and select your key file

Now you can login to AWS instance.

3. Transfer file between AWS and your local PC

Here I use FileZilla to transfer files between AWS and my PC. Download and install FileZilla client. 

Here is a tutorial video:

Connect to Amazon EC2 file directory using FileZilla and SFTP, Video Tutorial

Summary of above video tutorial:

Edit (Preferences) > Settings > Connection > SFTP, Click "Add key file”

Browse to the location of your .pem file and select it.

A message box will appear asking your permission to convert the file into ppk format. Click Yes, then give the file a name and store it somewhere.

If the new file is shown in the list of Keyfiles, then continue to the next step. If not, then click "Add keyfile..." and select the converted file.

File > Site Manager Add a new site with the following parameters:

Host: Your public dns name of ec2 instance, or the public ip address of the server

Protocol: SFTP

Logon Type: Normal

User: From the docs: "For Amazon Linux, the default user name isec2-user. For RHEL5, the user name is often root but might be ec2-user. For Ubuntu, the user name isubuntu. For SUSE Linux, the user name isroot. For Debian, the user name isadmin. Otherwise, check with your AMI provider."

Press Connect Button - If saving of passwords has been disabled, you will be prompted that the logon type will be changed to 'Ask for password'. Say 'OK' and when connecting, at the password prompt push 'OK' without entering a password to proceed past the dialog.

Note:FileZilla automatically figures out which key to use. You do not need to specify the key after importing it as described above.

4. Prerequisites

HDF5

http://micro.stanford.edu/wiki/Install_HDF5#Ubuntu

sudo apt-get install libhdf5-serial-dev

cmake

http://askubuntu.com/questions/610291/how-to-install-cmake-3-2-on-ubuntu-14-04

1. Installation by a PPA (Upgrade to 3.2)

sudo apt-get install software-properties-common

sudo add-apt-repository ppa:george-edison55/cmake-3.x

sudo apt-get update

2.When cmake is not yet installed:

sudo apt-get install cmake

When cmake is already installed:

sudo apt-get upgrade

g++

http://charette.no-ip.com:81/programming/2011-12-24_GCCv47/

$ sudo apt-get install build-essential

CUDA

Register and download cuda toolkit first

https://developer.nvidia.com/cuda-downloads

Choose installer type and follow the instructions to install it

Here is official tutorial.

Matlab

From caffe official site we see that

Caffe’s MATLAB interface works with versions 2015a, 2014a/b, 2013a/b, and 2012b.

So here I choose 2014b version.

1. Download matlab iso file. 

2. Upload it to AWS

3. Mount the iso file

mount -o loop,rw /home/R2014b_glnxa64.iso /mnt

Here /mnt is the directory you want to mount this ISO.

Copy and change the file activate.ini  based on your need. Here are some important settings. 

isSilent=true

activateCommand=activateOffline

licenseFile=/home/ubuntu/matlab/MATLAB_R2014B_MAC_LINUX_crack/license.lic

Install matlab

./install -mode silent -fileInstallationKey 29797-39064-48306-32452 -agreeToLicense yes -activationPropertiesFile /home/activate.ini -destinationFolder

Caffe

This is the official guide for installation.

First clone caffe code

mkdir workspace

cd workspace

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

Change makefile and compile

cp Makefile.config.example Makefile.config

# Adjust Makefile.config (for example, if using Anaconda Python, or if cuDNN is desired)

make all

make test

make runtest

When I run " make all ", I met several problems. 

If you meet some problem about HDF5.h, you can try

You can also refer to this post if you meet other problems.