Skip to content


Switch branches/tags
This branch is 1 commit ahead, 1010 commits behind BVLC:master.

Latest commit


Git stats


Failed to load latest commit information.
Latest commit message
Commit time

Caffe library with JNI to cope with SpeeDO.

Install instructions (Tested on Ubuntu 12.04)

Install caffe and all dependencies

We provide two ways to install caffe and its dependencies(OpenBLAS, glog, gflags, lmdb etc).

  • A. Manually install on all cluster nodes
  • B. Automatic deployment by cloudera parcels

A. Manually install on all cluster nodes

We provide a script to install caffe dependencies (you need to set JAVA_HOME), on each host, execute:

git clone
# install dependency
export JAVA_HOME=/usr/lib/jvm/java-8-oracle
# install caffe
make all javainstall && sudo make install

B. Automatic deployment by cloudera parcels

Instead of installing maually on each machinein the cluster, you can use scripts in parcel-scripts to create a parcel and deploy using Cloudera Manager:

# Install cloudera manager:
chmod u+x cloudera-manager-installer.bin && sudo ./cloudera-manager-installer.bin
# Generate caffe parcel, also including its main dependencies
git clone
cd parcel-scripts

The parcel and manifest is generated in parcel-scripts/caffe/target/output/ folder. containing two files:

ls -al caffe/target/output/
-rw-rw-r-- 1 ****** CAFFE-0.1-***.parcel
-rw-rw-r-- 1 ****** manifest.json

Then following the steps from Creating a Local Parcel Repository to deploy caffe on cloudera cluster.

You still needs to manually install following softwares on each machine:

sudo apt-get -y install libleveldb-dev libsnappy-dev libopencv-dev libboost-all-dev libhdf5-serial-dev

Verify installation by java tests

You can run java tests after installing to make sure your installation is correct. The unit test depends on mnist dataset, so you need to download mnist dataset using script in caffe repo:


Now you can run java tests:

make javatest



Original Caffe README below


Caffe is a deep learning framework made with expression, speed, and modularity in mind. It is developed by the Berkeley Vision and Learning Center (BVLC) and community contributors.

Check out the project site for all the details like

and step-by-step examples.

Join the chat at

Please join the caffe-users group or gitter chat to ask questions and talk about methods and models. Framework development discussions and thorough bug reports are collected on Issues.

Happy brewing!

License and Citation

Caffe is released under the BSD 2-Clause license. The BVLC reference models are released for unrestricted use.

Please cite Caffe in your publications if it helps your research:

  Author = {Jia, Yangqing and Shelhamer, Evan and Donahue, Jeff and Karayev, Sergey and Long, Jonathan and Girshick, Ross and Guadarrama, Sergio and Darrell, Trevor},
  Journal = {arXiv preprint arXiv:1408.5093},
  Title = {Caffe: Convolutional Architecture for Fast Feature Embedding},
  Year = {2014}


Caffe: a fast open framework for deep learning.







No packages published


  • C++ 79.1%
  • Python 8.6%
  • Cuda 4.4%
  • CMake 3.1%
  • Protocol Buffer 1.5%
  • MATLAB 1.0%
  • Other 2.3%