Ubuntu+docker+tensorflow+opencv+tensorboard 安装

Centos7安装TensorFlow

1.1.安装Centos7

https://blog.csdn.net/monkey131499/article/details/51169210

2.安装Python3

查看当前Python版本信息,命令(python -v),Centos7默认的Python版本是2.7.5

 

下载Python3:

wget https://www.python.org/ftp/python/3.4.1/Python-3.4.1.tgz

解压编译安装

# tar zxvf Python-3.4.1.tgz

# cd Python-3.4.1

# ./configure

# make

# make install

 

 

本虚拟机不需要覆盖版本

 

若要覆盖

,。看文档

3.安装Python-pip

https://blog.csdn.net/yulei_qq/article/details/52984334

4.安装TensorFlow

 

https://blog.csdn.net/monkey131499/article/details/51169210

 

 

 

 

 

 

centos 7 下搭建 tensorflow+keras 深度学习环境

https://blog.csdn.net/huangfei711/article/details/78606159

 

 

 

 

Linux下安装Anaconda

https://jingyan.baidu.com/article/20b68a8893ae50796cec62b4.html

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Ubuntu安装

 

  1. Ubuntu安装

https://jingyan.baidu.com/article/3c48dd348bc005e10be358eb.html   

 

 

  1. ubuntu下如何安装Tensorflow

https://www.cnblogs.com/tsingke/p/7171270.html

 

  1. Ubuntu16.04下Anaconda安装完成后conda:找不到命令

https://blog.csdn.net/xianglao1935/article/details/80510494

 

 

  1. 4在anaconda中安装、切换python的版本:2.7~3.6

.https://blog.csdn.net/x_ym/article/details/78995472

 

  1. 装Docker

https://docs.docker.com/install/linux/docker-ce/ubuntu/    最新

或者

https://blog.csdn.net/yxgxy270187133/article/details/48492937    学习版

 

 

用  docker -v 测试

用  sudo docker run hello-world 测试

   

  1. 基于Docker的TensorFlow 环境搭建

 

https://www.cnblogs.com/dyufei/p/8027764.html

 

或者:https://www.jb51.net/article/135441.htm

 

https://www.jb51.net/article/135441.htm 使用说明

 

 

  1. 把Docker安装为自启动服务

https://blog.csdn.net/pipisorry/article/details/50803028

 

 

 

  1. docker + jupyter +tensorflow +opencv +tensorboard:

Tensorflow+opencv 组合镜像:

https://blog.csdn.net/chenming_hnu/article/details/70184543

 

组合镜像+tensorboard:

参考:https://blog.csdn.net/qq_29831979/article/details/79453641

docker run --name my-tf-1 -dit  -p 8888:8888 -p 6006:6006 tensorflow:tensorflowCV   即可!!

打开8888网站,然后会需要token,返回终端输入 sudo docker my-tf-1查看

 

这一步,可以通过

①# tensorboard --inspect --logdir="/notebooks/graph/dataflow" 看能否打开文件

②检查6006 的路径是否正确

 

 

下面的不对,可用于anaconda:

组合镜像+tensorboard:

https://www.cnblogs.com/dyufei/p/8094507.html

命令:

docker volume create --name notebooks 

docker volume create --name logs

sudo mkdir /home/dyufei/docker/notebooks

sudo mkdir /home/dyufei/docker/logs

 

posted @ 2018-08-06 14:56  sharryling  阅读(1624)  评论(0编辑  收藏  举报