Install Caffe2 for video models

installing PyTorch through conda does not install the necessary video libraries. You must install PyTorch/Caffe2 libraries from source to use video models

tested for CUDA 10.0, cudnn 7.4.2.12-cuda-10.0, openmpi 3.1.0

load necessary CUDA/cudnn libraries

module load cuda/10.0 module load cudnn/7.4.2.12-cuda-10.0

load MPI

module load openmpi

add to LD_LIBRARY_PATH

export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:

get PyTorch from source

git clone --recursive https://github.com/pytorch/pytorch

make sure to have opencv

conda install -c conda-forge opencv

ffmpeg

conda install -c conda-forge ffmpeg

install python-lmdb

conda install -c conda-forge python-lmdb

cd USE_FFMPEG=1 USE_OPENCV=1 USE_LEVELDB=0 USE_LMDB=1 USE_CUDA=1 python setup.py install

test installation

python -c 'from caffe2.python import core' 2>/dev/null && echo "Success" || echo "Failure"

VMZ/facebook PYTHONPATH

export PYTHONPATH=$PYTHONPATH: