Requirements:
PyTorch1.7.1+, Python3.8+, Cuda11.0+
mmcv==1.3.5
Option 1: Step-by-step installation
a. Create a conda virtual environment and activate it. Then install required packages.
conda create --name dm-b python=3.8
//source activate(importent)
conda activate dense_matching_benchmark# this installs the right pip and dependencies for the fresh python
conda install ipython
conda install pip# install required packages from requirements.txt
pip install -r /(自己的路经）/requirements.txt -i https://pypi.tuna.tsinghua.edu.cn/simple
注：repuirement.txt文件里面有pytorch,删掉。torch>=1.7.1 torchvision>=0.8.2 torchaudio>=0.7.2 自己下载
b.Install PyTorch stable or nightly and torchvision following the https://pytorch.org/
conda install pytorch==1.7.1 torchvision==0.8.2 torchaudio==0.7.2 cudatoolkit=10.1 -c pytorch
conda 安装报错 specified in the package manifest cannot be found.
解决： conda clean --packages --tarballs 删除缓存的包重新下载
离线安装：
conda install --offline pytorch-1.3.1-py3.7_cuda10.1.243_cudnn7.6.3_0.tar.bz2
conda install --offline torchvision-0.4.2-py37_cu101.tar.bz2
conda install --offline cudatoolkit-10.1.243-h6bb024c_0.tar.bz2
