Skip to content

Latest commit

 

History

History
32 lines (20 loc) · 1.11 KB

README.md

File metadata and controls

32 lines (20 loc) · 1.11 KB

IERN Environment Installation

Manual installation

The provided exported anoconda environment file is mainly tested on NVIDIA GeForce RTX 2080 Ti with cuda 10. If it does not work on your machine, please consider installing the environment manually by following the instructions below.

  • initialize a python 3.8 environment
conda create -n iern python=3.8.2
conda activate iern
  • install pytorch 1.4.0. Choose a correct CUDA version from here. Below we provide the commands for cuda10
conda install pytorch==1.4.0 torchvision==0.5.0 cudatoolkit=10.1 -c pytorch
  • install related python dependencies
pip install scikit-learn==0.23.2 matplotlib==3.1.3 tqdm==4.45.0 seaborn==0.11.0 tensorboard==2.7.0
pip install protobuf==3.19.0  # fix bug for tensorboard
pip install imutils==0.5.3 opencv-python==4.3.0.36 dlib==19.20.0  # for preprocessing the input dataset

Miscs

Please follow the instructions to use PyTorch 1.4.0, it might encounter some unexpected bugs if the PyTorch version is higher than the recommended version.