win10 anaconda tensorflow环境配置

1.      设置好GPU开发环境，安装cuda8.0和cudnn5.1

2.      安装Anaconda3-4.3.1-Windows-x86_64，默认Python版本为3.6

3.   安装完以后，打开Anaconda Prompt，输入清华的仓库镜像，更新包更快：

conda config --add channelshttps://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/

conda config --set show_channel_urls yes

4.      建立TensorFlow空间：conda create -n tensorflow python=3.5，设置Python版本为3.5

5.      激活TensorFlow空间：activate tensorflow

6.       安装TensorFlow：pip install--ignore-installed --upgrade tensorflow_gpu-1.0.0-cp35-cp35m-win_amd64.whl

7.  测试TensorFlow：

import tensorflow as tf

hello = tf.constant('Hello, TensorFlow!')

sess = tf.Session()

print(sess.run(hello))

 

 

设置SSD运行环境

1．  安装numpy（anaconda 离线安装）：pipinstall numpy-1.12.1-cp35-none-win_amd64.whl

2．  安装matplotlib（anaconda 离线安装）：pipinstall matplotlib-2.0.1-cp35-cp35m-win_amd64.whl

3．  安装opencv（离线）：anaconda中安装：pipinstall opencv_python-3.2.0+contrib-cp35-cp35m-win_amd64.whl

4．  下载TensorFlow版本的SSD：git clone https://github.com/balancap/SSD-Tensorflow.git，或者下载压缩包解压

5．  解压TensorFlowssd目录下的/checkpoint里的ssd_300_vgg.ckpt.zip，得到模型参数

6．  安装pycharm-community-2017.1.2.exe，python编辑器，File->Setting->Project:Python->ProjectInterpreter:修改Python版本到TensorFlow工作空间下的python

7．  在notebook下新建工程，新建test_ssd.py文件

8．  在pycharm中打开ssd_notebook.ipynb，复制非注释的内容至test_ssd.py下

9．  修改test_ssd.py：

# Test on some demoimage and visualize output.
#path = '../demo/'
#image_names = sorted(os.listdir(path))
#print(image_names)
#for it  in image_names:
cam=cv2.VideoCapture(0)
success, img = cam.read()
while success:
    #img = cv2.imread(path+it)#mpimg.imread(path + it)
    t1=cv2.getTickCount()
    rclasses, rscores, rbboxes =  process_image(img)
    visualization.bboxes_draw_on_img(img,rclasses, rscores, rbboxes, visualization.colors_plasma)
    t2=cv2.getTickCount()
    print('time consumption:%.3f ms'%(1000*(t2-t1)/cv2.getTickFrequency()))
    cv2.imshow('test',img)
    c=cv2.waitKey(1)
    if c==27:
       break
    # visualization.bboxes_draw_on_img(img, rclasses,rscores, rbboxes, visualization.colors_plasma)
    #visualization.plt_bboxes(img, rclasses,rscores, rbboxes)
    success, img = cam.read()

10．             运行程序test_ssd.py