Deploy PyTorch ResNet18 model trained with Python using C++.

Develop using PyTorch Python API:

1. create virtualenv and install dependencies:
   virtualenv --system-site-packages -p python3.6 ./venv && source ./venv/bin/activate
   pip install torch torchvision opencv-python
   
2. train and save model converted to Torch Script:
   python train.py
   

Deploy using PyTorch C++ API:

1. install PyTorch:
   wget https://download.pytorch.org/libtorch/nightly/cpu/libtorch-shared-with-deps-latest.zip
   unzip libtorch-shared-with-deps-latest.zip -d /tmp
   
2. install OpenCV:
   wget https://github.com/opencv/opencv/archive/3.4.6.zip
   unzip opencv-3.4.6.zip -d /tmp
   cd /tmp/opencv-3.4.6
   mkdir build && cd build
   cmake -D CMAKE_BUILD_TYPE=RELEASE -D CMAKE_INSTALL_PREFIX=/tmp/opencv-3.4.6/install ..
   make -j8
   sudo make install
   
3. build and run c++ program:
   mkdir build && cd build
   cmake -DCMAKE_PREFIX_PATH="/tmp/libtorch;/tmp/opencv-3.4.6/install" ..
   make
   ./app