1. 安装anaconda3(用于maskrcnn)
sh Anaconda3-5.2.0-Linux-x86_64.sh
## 加入环境变量后,python默认为python3.5(anaconda3),原来的python变为python2
sudo apt-get install python-pip
相应的tensorflow:
pip install tensorflow-gpu==1.4.0 easydict opencv-python
2.创建python2.7(用于tf-faster-rcnn)
推荐:
bash Anaconda2-4.1.1-Linux-x86_64.sh -b -p $HOME/softwares/anaconda3/envs/py2
rm -f $HOME/softwares/anaconda3/envs/py2/bin/conda*
rm -f $HOME/softwares/anaconda3/envs/py2/conda-meta/conda-*
rm -f $HOME/softwares/anaconda3/envs/py2/bin/activate
rm -f $HOME/softwares/anaconda3/envs/py2/bin/deactivate
cd $HOME/softwares/anaconda3/envs/py2/bin
ln -s ../../../bin/conda .
ln -s ../../../bin/activate .
ln -s ../../../bin/deactivate .
不推荐:
conda create --name py2 python=2.7
这种方式安装的python2.7没有numpy等等,很麻烦。。。
相应的tensorflow:
pip install tensorflow-gpu==1.0.1 easydict opencv-python
3. 切换
source activate py2
4. 退出
source deactivate
参考: