Extract module installation

These instructions were not tested, we found them lying around in some old docs. They may not work at all, but that's all we have.

Software requirements

• Pytorch 1.0, build nightly
• torchvision
• cocoapi
• yacs
• matplotlib
• GCC >= 4.99

Installation

1. Create a conda environment
   conda create --name maskrcnn_benchmark
conda activate maskrcnn_benchmark
2. Install python
   conda install ipython
3. Install needed libraries
   pip install ninja yacs cython matplotlib tqdm numpy opencv-python
4. Install PyTorch, below an example for CUDA 9.0. Instructions for other CUDA versions can be found here: https://pytorch.org/get-started/locally/
   conda install -c pytorch pytorch-nightly torchvision cudatoolkit=9.0
export INSTALL_DIR=$PWD
5. Install pycocotools
   cd $INSTALL_DIR
git clone https://github.com/cocodataset/cocoapi.git
cd cocoapi/PythonAPI
python setup.py build_ext install
6. Download the code
   cd $INSTALL_DIR
git clone https://github.com/mini-pw/2019L-ProjektZespolowy/tree/master/
cd 2019L-ProjektZespolowy/extract/model/maskrcnn-benchmark
7. Install the model
   python setup.py build develop
unset INSTALL_DIR