CGPraktikum

Repositories for the course Computer Graphics I at the Eberhard Karls University Tübingen.

Setup

(Does not fully work yet)

ssh username@cgpool120[0-7].informatik.uni-tuebingen.de # (or cgcontact first if not in wsi network)

pip install virtualenv --user
pip install virtualenvwrapper --user

nano ~/.profile

Add the following lines to the file:

export PATH="/graphics/projects/cuda/toolkit_deeplearning/cuda/bin:$HOME/.local/bin:$PATH"
export LD_LIBRARY_PATH="/graphics/projects/cuda/toolkit_deeplearning/cuda/lib64:$LD_LIBRARY_PATH"
export PYTHONPATH="$HOME/.local/lib/python2.7/site-packages:$PYTHONPATH"
export WORKON_HOME=$HOME/.virtualenvs
export PROJECT_HOME=$HOME/PCG
source $HOME/.local/bin/virtualenvwrapper.sh
export CUDA_VISIBLE_DEVICES='0'

(note the addition of PYTHONPATH compared to the tutorial in ILIAS)
Press Ctrl+O, Enter and Ctrl+X to save/exit the profile

Then (everytime you log on), type:

source ~/.profile
mkvirtualenv env

to start the environment.
The first time in the environment, install tensorflow:

easy_install -U pip        #ensure pip version
pip install --user --upgrade tensorflow-gpu
pip install --user -U git+https://github.com/ppwwyyxx/tensorpack.git

This should install tensorflow aswell as tensorpack (tensorpack didn't work for me without adding site-packages directory to $PYTHONPATH).

Test if it is correctly installed using

import tensorflow as tf
from tensorpack import *

If this doesn't fail, the packages should be installed correctly

tensorflow getting started: https://www.tensorflow.org/get_started/get_started

mariushobbhahn/Reading-scene-text-in-deep-convolutional-sequences

CGPraktikum

Setup