Openvino（Windows10 安装OpenVINO）

注意：OpenVINO官方的安装说明非常详细。

请参考：https://docs.openvinotoolkit.org/latest/_docs_install_guides_installing_openvino_windows.html

 

本博客是记录家里windows10安装openVINO的过程；

 

 

-1、下载OpenVINO

官网的下载链接：可能很卡，本博客给出文章所涉及的openvino安装文件

w_openvino_toolkit_p_2019.3.379.exe

-2、安装OpenVINO

双击w_openvino_toolkit_p_2019.3.379.exe执行安装；

如果是使用默认安装位置的话，是安装在"C:\Program Files (x86)\IntelSWTools\"文件夹下，我自己是选择安装到D盘，即：

D:\D02_AppLearn\D02A10_OpenVINO\installer

如上提示我，如果需要使用VS2019作为IDE,那么我需要安装CMake3.14。

另外检测到没有Intel GPU,注意此GPU非彼GPU，这里是指的Intel自家的集成显卡（核显），似乎Intel也在做自己的独显GPU，不过目前这里就是指的是核显。

另外这个警告可以无视，因为有GPU运行的异步，不一定赶得上CPU的异步，CPU性能还是相当重要的！

 

-3、初始化OpenVINO

在cmd终端进行初始化。

C:\Users\Administrator>D: 

D:\>cd D:\D02_AppLearn\D02A10_OpenVINO\installer\IntelSWTools\openvino_2019.3.379\bin

D:\D02_AppLearn\D02A10_OpenVINO\installer\IntelSWTools\openvino_2019.3.379\bin>setupvars.bat
Python 3.7.3
ECHO 处于关闭状态。
PYTHONPATH=D:\D02_AppLearn\D02A10_OpenVINO\installer\IntelSWTools\openvino_2019.3.379\deployment_tools\open_model_zoo\tools\accuracy_checker;
D:\D02_AppLearn\D02A10_OpenVINO\installer\IntelSWTools\openvino_2019.3.379\python\python3.7;
D:\D02_AppLearn\D02A10_OpenVINO\installer\IntelSWTools\openvino_2019.3.379\python\python3;
[setupvars.bat] OpenVINO environment initialized

D:\D02_AppLearn\D02A10_OpenVINO\installer\IntelSWTools\openvino_2019.3.379\bin>

如上图openVINO初始化成功了。由于我想使用的python版本的OpenVINO,

下面需要安装openVINO涉及的python三方包。

 

-4、安装OpenVINO依赖的python三方包

在安装文件夹下./IntelSWTools/openvino_2019.3.379/deployment_tools/model_optimizer/requirements_tf.txt

对应我的本机：D:\D02_AppLearn\D02A10_OpenVINO\installer\IntelSWTools\openvino_2019.3.379\deployment_tools\model_optimizer\requirements_tf.txt

使用清华镜像源进行快速安装：

 pip install -i  https://pypi.tuna.tsinghua.edu.cn/simple/  tensorflow==1.14.0

 pip install -i  https://pypi.tuna.tsinghua.edu.cn/simple/  networkx==2.2

 pip install -i  https://pypi.tuna.tsinghua.edu.cn/simple/ defusedxml==0.6.0

安装完成后查看python的三方包：

 

-5、测试OpenVINO推理包是否可以导入

C:\Users\Administrator>D:

D:\>cd D:\D02_AppLearn\D02A10_OpenVINO\installer\IntelSWTools\openvino_2019.3.379\bin

D:\D02_AppLearn\D02A10_OpenVINO\installer\IntelSWTools\openvino_2019.3.379\bin>setupvars.bat
Python 3.7.3
ECHO 处于关闭状态。
PYTHONPATH=D:\D02_AppLearn\D02A10_OpenVINO\installer\IntelSWTools\openvino_2019.3.379\deployment_tools\open_model_zoo\tools\accuracy_checker;D:\D02_AppLearn\D02A10_OpenVINO\installer\IntelSWTools\openvino_2019.3.379\python\python3.7;D:\D02_AppLearn\D02A10_OpenVINO\installer\IntelSWTools\openvino_2019.3.379\python\python3;
[setupvars.bat] OpenVINO environment initialized

D:\D02_AppLearn\D02A10_OpenVINO\installer\IntelSWTools\openvino_2019.3.379\bin>python
Python 3.7.3 (v3.7.3:ef4ec6ed12, Mar 25 2019, 22:22:05) [MSC v.1916 64 bit (AMD64)] on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> from openvino.inference_engine import IENetwork, IECore
>>>

如上图可以正常导入推理机：from openvino.inference_engine import IENetwork, IECore

 

-6、使用OpenVINO运行YOLOV3或者Tiny_yolo模型。

本博客将给出优化后的网络和权重，可以直接加载得到结果，另外将提供一个“玩具总动员”视频。

-----------------------------------------------------------------------------------------------------------------------------------------

---附注---

另外cv2在OpenVINO安装阶段，安装到OpenVINO中，在执行初始化后，就可以导入cv2了。如下：
python -c "import sys; print(sys.executable)"

python -c "import cv2"

C:\Users\Administrator>python -c "import sys; print(sys.executable)"
D:\D00_Python3\D00A2_python3.7.3\install\python.exe

C:\Users\Administrator>python -c "import cv2"
Traceback (most recent call last):
  File "", line 1, in 
ModuleNotFoundError: No module named 'cv2'

C:\Users\Administrator>

如果进行OpenVINO初始化后，可以导入到cv2，如下：

D:

D:\>cd D:\D02_AppLearn\D02A10_OpenVINO\installer\IntelSWTools\openvino_2019.3.379\bin

D:\D02_AppLearn\D02A10_OpenVINO\installer\IntelSWTools\openvino_2019.3.379\bin>setupvars.bat

python -c "import cv2; print(cv2.__version__)"

C:\Users\Administrator>python -c "import sys; print(sys.executable)"
D:\D00_Python3\D00A2_python3.7.3\install\python.exe


C:\Users\Administrator>python -c "import cv2"
Traceback (most recent call last):
  File "", line 1, in 
ModuleNotFoundError: No module named 'cv2'


C:\Users\Administrator>D:


D:\>cd D:\D02_AppLearn\D02A10_OpenVINO\installer\IntelSWTools\openvino_2019.3.379\bin


D:\D02_AppLearn\D02A10_OpenVINO\installer\IntelSWTools\openvino_2019.3.379\bin>setupvars.bat
Python 3.7.3
ECHO 处于关闭状态。
PYTHONPATH=D:\D02_AppLearn\D02A10_OpenVINO\installer\IntelSWTools\openvino_2019.3.379\deployment_tools\open_model_zoo\tools\accuracy_checker;D:\D02_AppLearn\D02A10_OpenVINO\installer\IntelSWTools\openvino_2019.3.379\python\python3.7;D:\D02_AppLearn\D02A10_OpenVINO\installer\IntelSWTools\openvino_2019.3.379\python\python3;
[setupvars.bat] OpenVINO environment initialized


D:\D02_AppLearn\D02A10_OpenVINO\installer\IntelSWTools\openvino_2019.3.379\bin>python -c "import cv2; print(cv2.__version__)"
4.1.2-openvino

D:\D02_AppLearn\D02A10_OpenVINO\installer\IntelSWTools\openvino_2019.3.379\bin>

-----------------------------------------------------------------------------------------------------------------------------------------

 

 下面将使用官方的脚本、制定CPU进行运行：

-a、脚本：

.\IntelSWTools\openvino_2019.3.379\inference_engine\demos\python_demos\object_detection_demo_yolov3_async\object_detection_demo_yolov3_async.py

（即我的本机：D:\D02_AppLearn\D02A10_OpenVINO\installer\IntelSWTools\openvino_2019.3.379\inference_engine\demos\python_demos\object_detection_demo_yolov3_async\object_detection_demo_yolov3_async.py）

 

-b、由于本计算机只有CPU，需要指定设备的输入参数为CPU，并制定指令集

.\IntelSWTools\openvino_2019.3.379\deployment_tools\inference_engine\bin\intel64\Release\cpu_extension_avx2.dll

（即我的本机：D:\D02_AppLearn\D02A10_OpenVINO\installer\IntelSWTools\openvino_2019.3.379\deployment_tools\inference_engine\bin\intel64\Release\cpu_extension_avx2.dll）

-c、将视频以及脚本、模型放入如下文件夹内：

E:\E02_HomeWork\E02_A1_OpenVINO

-d、运行如下代码：

初始化openVINO后，进入脚本文件夹(E:\E02_HomeWork\E02_A1_OpenVINO)下运行如下代码：

python object_detection_demo_yolov3_async.py -i ./movie.mp4 -m ./IR_tiny_yolo/road_tiny.xml -d CPU -l "D:\D02_AppLearn\D02A10_OpenVINO\installer\IntelSWTools\openvino_2019.3.379\deployment_tools\inference_engine\bin\intel64\Release\cpu_extension_avx2.dll"

如下：

C:\Users\Administrator>d:

D:\>cd D:\D02_AppLearn\D02A10_OpenVINO\installer\IntelSWTools\openvino_2019.3.379\bin

D:\D02_AppLearn\D02A10_OpenVINO\installer\IntelSWTools\openvino_2019.3.379\bin>setupvars.bat
Python 3.7.3
ECHO 处于关闭状态。
PYTHONPATH=D:\D02_AppLearn\D02A10_OpenVINO\installer\IntelSWTools\openvino_2019.3.379\deployment_tools\open_model_zoo\tools\accuracy_checker;D:\D02_AppLearn\D02A10_OpenVINO\installer\IntelSWTools\openvino_2019.3.379\python\python3.7;D:\D02_AppLearn\D02A10_OpenVINO\installer\IntelSWTools\openvino_2019.3.379\python\python3;
[setupvars.bat] OpenVINO environment initialized

D:\D02_AppLearn\D02A10_OpenVINO\installer\IntelSWTools\openvino_2019.3.379\bin>E:

E:\>cd E:\E02_HomeWork\E02_A1_OpenVINO

E:\E02_HomeWork\E02_A1_OpenVINO>python object_detection_demo_yolov3_async.py -i ./movie.mp4 -m ./IR_tiny_yolo/road_tiny.xml -d CPU -l "D:\D02_AppLearn\D02A10_OpenVINO\installer\IntelSWTools\openvino_2019.3.379\deployment_tools\inference_engine\bin\intel64\Release\cpu_extension_avx2.dll"
[ INFO ] Creating Inference Engine...
[ INFO ] Loading network files:
        ./IR_tiny_yolo/road_tiny.xml
        ./IR_tiny_yolo/road_tiny.bin
[ INFO ] Preparing inputs
MFX: Can't initialize session
[ INFO ] Loading model to the plugin
[ INFO ] Starting inference...
To close the application, press 'CTRL+C' here or switch to the output window and press ESC key
To switch between sync/async modes, press TAB key in the output window
[ INFO ] Layer detector/yolo-v3-tiny/Conv_12/BiasAdd/YoloRegion parameters:
[ INFO ]          classes : 10
[ INFO ]          num     : 3
[ INFO ]          coords  : 4
[ INFO ]          anchors : [10.0, 14.0, 23.0, 27.0, 37.0, 58.0]
[ INFO ] Layer detector/yolo-v3-tiny/Conv_9/BiasAdd/YoloRegion parameters:
[ INFO ]          classes : 10
[ INFO ]          num     : 3
[ INFO ]          coords  : 4
[ INFO ]          anchors : [81.0, 82.0, 135.0, 169.0, 344.0, 319.0]
[ INFO ] Layer detector/yolo-v3-tiny/Conv_12/BiasAdd/YoloRegion parameters:
[ INFO ]          classes : 10
[ INFO ]          num     : 3
[ INFO ]          coords  : 4
[ INFO ]          anchors : [10.0, 14.0, 23.0, 27.0, 37.0, 58.0]
......

运行界面：

 本提供提供复现的视频以及脚本、权重。

https://download.csdn.net/download/jn10010537/12519639

https://download.csdn.net/download/jn10010537/12519625