Skip to content

Latest commit

 

History

History
80 lines (56 loc) · 1.87 KB

MANUAL_INSTALL.md

File metadata and controls

80 lines (56 loc) · 1.87 KB
Raw

1. Install Caffe on PYNQ

  • Install dependencies
sudo apt-get install libprotobuf-dev libleveldb-dev libsnappy-dev libopencv-dev libhdf5-serial-dev protobuf-compiler
sudo apt-get install --no-install-recommends libboost-all-dev
sudo apt-get install libgflags-dev libgoogle-glog-dev liblmdb-dev
  • Install Protobuf 3

pip3 install protobuf

  • Add SWAP USB

Since PYNQ has limited RAM, in order to install Caffe extra memory will be required in the form of swap memory. In my test, a USB with size 2GB is large enough.

mkswap /dev/sda
swapon /dev/sda
  • Install Caffe
cd /home/xilinx
git clone https://github.com/BVLC/caffe.git`

Copy the PYNQ version of Makefile.config (provided under PYNQ-Classification/tools/CAFFE_ON_PYNQ) to caffe root directory

make all
make test
make runtest
  • Unmount SWAP USB

swapoff /dev/sda

  • Install pycaffe with Python 3
cd python
for req in $(cat requirements.txt); do sudo pip install $req; done
echo "export PYTHONPATH=$(pwd):$PYTHONPATH " >> ~/.bash_profile # to be able to call "import caffe" from Python after reboot
source ~/.bash_profile # Update shell 
cd .
  • Export PYTHONPATH

Add the following line at the end of ".bashrc" file

export PYTHONPATH=/home/xilinx/caffe/python

2. Install Theano with Lasagne on PYNQ

pip install -r https://raw.githubusercontent.com/Lasagne/Lasagne/v0.1/requirements.txt
pip install Lasagne==0.1

3. Run CNN Demos on PYNQ

cd /home/xilinx/jupyter_notebooks
git clone https://github.com/awai54st/PYNQ-Classification.git

The following two demos show the implementions of LeNet and CIFAR-10 on PYNQ:

/home/xilinx/jupyter_notebooks/PYNQ-Classification/PYNQ_SIDE/Theano/Lenet/Using a Caffe Pretrained Network - LeNet5.ipynb
/home/xilinx/jupyter_notebooks/PYNQ-Classification/PYNQ_SIDE/Theano/CIFAR_10/Using a Caffe Pretrained Network - CIFAR10.ipynb