安装：
pip install opencv-python
pip install opencv-contrib-python

conda install m2w64-toolchain
conda install mingw #先安装mingw；
conda install libpython #再安装libpython：
pip install theano
conda install tensorflow-gpu1.10.0
pip install keras #再安装keras
conda install keras2.2.0
pip install tensorflow-gpu1.10.0 keras2.2.0
conda uninstall tensorflow
忘了安装cudatoolkit cudnn
使用pip install tensorflow-gpu Could not find ‘cudart64_90.dll’
把后两个pip 用conda代替了

conda install m2w64-toolchain
conda install mingw #先安装mingw；
conda install libpython #再安装
conda install keras-gpu==2.2.0
清华镜像的tf 缺少tf-base 所以完全conda安装使用不了gpu
keras可以全用conda来安装，加上cudatoolkit cudnn tf-base就好

使用powershell 启动jupyter，载入包出错

conda尽量不要 conda install 和pip install 混用
conda config --show-sources
conda list --show-channel-urls

版本搜索
anaconda search -t conda lifelines
conda install -c https://conda.anaconda.org/conda-forge lifelines
-c -i

回滚
1conda
conda list --revisions
conda install --rev 1

清除缓存
conda clean -i 清除缓存索引
conda clean -p 删除没有用的包
conda clean -t 删除tar包

conda update conda
conda update --strict-channel-priority --all

删除镜像channels
conda config --remove-key channels

目测conda安装opencv，直接包含了opencv-contrib

conda install -c conda-forge pcl
conda install -c open3d-admin open3d
conda config --remove-key channels

conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main/
conda config --set show_channel_urls yes

删除镜像
conda env remove -n 虚拟环境的名称

linux