tensorflow环境搭建centos环境
python安装方法
https://www.cnblogs.com/wintest/p/12057170.html
安装Anaconda3
参照博客：https://blog.csdn.net/shulixu/article/details/88553841
参照博客：https://www.jianshu.com/p/2831365a61c3
anaconda版本 python版本对应关系
https://blog.csdn.net/yuejisuo1948/article/details/81043823
mac安装anaconda
https://www.pythonheidong.com/blog/article/483998/0c142aace6824d05b438/
在线安装命令（下载sh文件）：
下载
wget https://repo.anaconda.com/archive/Anaconda3-5.2.0-Linux-x86_64.sh
执行sh文件：
bash Anaconda3-5.2.0-Linux-x86_64.sh
注意：有时候需要多次安装
bash Anaconda3-5.2.0-Linux-x86_64.sh -u
注意：安装过程为回车、yes
安装成功的最好文案为：thank you for installing anaconda3…
激活anaconda3的路径：
source ~/.bashrc
注意：~/.bashrc文件内容记录了anaconda3的安装路径，默认值为：/root/anaconda3
在Anaconda环境下搭建tensorflow
在这里插入图片描述
参照博客：https://blog.csdn.net/shulixu/article/details/88553841
1 建立一个名为tensorflow_env的运行环境，python=3.6
conda create --name tensorflow_env python=3.6
删除
conda remove --name tensorflow_env --all
2 激活tensorflow_env这个环境
conda activate tensorflow_env
激活 anaconda 环境
source activate
注意：先要激活tensorflow_env以后，再安装tensorflow
3 安装tensorflow
conda install tensorflow
注意：退出conda环境
conda deactivate
测试tensorflow是否安装成功
python
import tensorflow
注意：若不报错，则安装成功
登录tensorflow_env环境
登录tensorflow_env环境
source activate tensorflow_env
退出tensorflow_env环境
source deactivate
jupyter 安装
(tensorflow_env) [root@instance-02b383b1 ~]# jupyter notebook --generate-config
Writing default config to: /root/.jupyter/jupyter_notebook_config.py
tensorboard 安装
参照博客：https://blog.csdn.net/fendouaini/article/details/80344591
安装命令：
pip install tensorboard
注意：安装好了以后在路径：/root/anaconda3/bin 下面会有tensorbaord
tensorboard的使用
切换到目录
cd /root/anaconda3/bin
python tensorboard --logdir=/path/to/logs
系统会自动启动一个端口，输出日志为：
[root@VM_0_2_centos bin]# python tensorboard --logdir=/path/to/logs
TensorFlow installation not found - running with reduced feature set.
W0801 21:42:25.461625 140474065319680 plugin_event_accumulator.py:294] Found more than one graph event per run, or there was a metagraph containing a graph_def, as well as one or more graph events. Overwriting the graph with the newest event.
W0801 21:42:25.461767 140474065319680 plugin_event_accumulator.py:302] Found more than one metagraph event per run. Overwriting the metagraph with the newest event.
TensorBoard 1.14.0 at http://VM_0_2_centos:6006/ (Press CTRL+C to quit)
打开浏览器输入ip和端口就可以访问了：
错误问题解决
1、问题：No module named ‘PIL’
解决：链接：https://blog.csdn.net/m0_37422289/article/details/79284541
conda uninstall pillow
conda update pip
pip install pillow
2、问题： No module named ‘matplotlib’
解决：链接：https://blog.csdn.net/andyL_05/article/details/84994101
pip install -U matplotlib
3、问题： No module named ‘cv2’
解决：链接：https://blog.csdn.net/xnightmare/article/details/81416761
pip install opencv-python
4、问题：安装caffe
解决，链接：https://www.jianshu.com/p/cc16e2977e27
安装caffe的前置准备（OK）：http://caffe.berkeleyvision.org/install_yum.html
两个环境都是
caffe 安装到cmake …报错了
5、问题：cmake安装
解决：https://blog.csdn.net/qq_38378235/article/details/80839693
6、问题：mac could not find Protobuf
解决：https://www.jianshu.com/p/f8b789280df4
7、问题：安装boost
解决：https://blog.csdn.net/xzwspy/article/details/81603227