Skip to content
No description, website, or topics provided.
Python C++ Java TeX C Shell M
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.


Build Status Coverage Status Development of a deep convolutional neural network for image recognition using the sparse filtering algorithm.

The research paper published by Stanford University can be found at:


export PYTHONPATH='path/to/the/directory/where/you/downloaded/the/code'
cd sparse/demos/sparse_filtering

# Training phase
python --train [--unsupervised | --supervised | --all] [--input "/path/to/train.txt"] [--model "model_name"]

# Execution phase
python [--input path/to/train.txt] [--model "model_name"]
python [--input path/to/val.txt] [--model "model_name"]
python ../../../util/tsne_python/
cd sparse/util/plibsvm-3.18
./ [-o "output_folder"] [-n "network_folder"]


The zip file with all the training data, validation data and annotations taken from the TRECVID dataset can be downloaded from Amazon S3. The zip file with the images of cats, birds, dogs and lamps taken from the ImageNet dataset can be downloaded from Amazon S3

All the input files (train.txt and val.txt) must be written with the following format:

image_path label
dog/dog_001.png 0
cat/cat_002.png 0
bird/dird_001.png 0

To generate these files for the ImageNet dataset you can use the '' and '' scripts available in the folder 'sparse/util/_data'. To generate the input files for the trecvid dataset you can use the '' script in the folder 'sparse/util/_data/trecvid'.

./  num_of_desired_images
./gen_val  num_of_desired_images

python  num_of_train_images  num_of_test_images


Firstly install mkl intel library! (Building numpy, scipy and numexpr from source will give a speed increase with respect to the precompiled versions downloaded through apt-get.)

UBUNTU 14.04

sudo apt-get install python-dev
sudo apt-get install libfreetype6-dev
sudo apt-get install python-numpy
sudo apt-get install python-scipy python-matplotlib ipython
sudo apt-get install python-numexpr
sudo apt-get install libblas-dev
sudo easy_install networkx
sudo apt-get install python-skimage
sudo apt-get install python-pydot
sudo apt-get install python-mpi python-mpi4py
sudo -E pip install --upgrade sklearn

MAC OS X 10.9

export CFLAGS=-Qunused-arguments
export CPPFLAGS=-Qunused-arguments

sudo -E pip install --upgrade numpy
sudo -E pip install --upgrade gfortran
sudo -E pip install --upgrade scipy
sudo -E pip install --upgrade cython
sudo -E pip install --upgrade scikit
sudo -E pip install --upgrade matplotlib
sudo -E pip install --upgrade pydot
brew install graphviz
sudo -E pip install --upgrade pillow
sudo -E pip install --upgrade scikit-learn
You can’t perform that action at this time.