Installing Caffe on Ubuntu 16.04 LTS with CUDA

# Below are instructions for building Caffe, assuming CUDA 8.0 & CUDNN are alrady installed.

# get dependencies
sudo apt install liblmdb-dev libhdf5-dev libgoogle-glog-dev libgflags-dev libatlas-base-dev libprotobuf-dev libleveldb-dev libsnappy-dev libopencv-dev libhdf5-serial-dev protobuf-compiler python-numpy python-yaml
sudo apt install --no-install-recommends libboost-all-dev

# get Caffe latest
git clone https://github.com/BVLC/caffe.git
cd caffe

# copy template config to real config
cp Makefile.config.example Makefile.config

# edit Makefile.conf, uncommenting USE_CUDNN and adding Debian-specific INCLUDE & LIBRARY paths per below
-# USE_CUDNN := 1
+USE_CUDNN := 1
-INCLUDE_DIRS := $(PYTHON_INCLUDE) /usr/local/include
-LIBRARY_DIRS := $(PYTHON_LIB) /usr/local/lib /usr/lib
+INCLUDE_DIRS := $(PYTHON_INCLUDE) /usr/local/include /usr/include/hdf5/serial/
+LIBRARY_DIRS := $(PYTHON_LIB) /usr/local/lib /usr/lib /usr/lib/x86_64-linux-gnu/hdf5/serial/
 
# build Caffe
make all -j8
make test -j8
make runtest

# make distribution in distribute/
make distribute

# fetch models
python scripts/download_model_binary.py models/bvlc_reference_caffenet
python scripts/download_model_binary.py models/bvlc_googlenet
python scripts/download_model_binary.py models/bvlc_reference_rcnn_ilsvrc13