OpenVINO之二：Windows环境下OpenVINO安装与配置

OpenVINO系列：
OpenVINO之一：OpenVINO概述
OpenVINO之二：Windows环境下OpenVINO安装与配置
OpenVINO之三：Linux环境下OpenVINO安装与配置

1 下载并安装OpenVINO

1-1 下载OpenVINO

下载地址：https://software.seek.intel.com/openvino-toolkit?os=windows
需要先注册，邮箱会收到下载链接和Serial Number来激活软件。

1-2 开始安装

双击下载好的OpenVINO开始安装

1-3 选择安装路径：

接着一直下一步就好，提示python版本不满足>=3.6，可以先忽略安装。

1-4 检查依赖库

Microsoft Visual Studio* with C++ 2017 or 2015, with MSBuild, and the Build Tools for Microsoft Visual Studio
CMake>=3.4
Python>=3.6.5
一般安装vs2015或者2017的，第一项已经满足。

1-5 安装完成

选择过会重启

2 安装依赖包

2-1 下载 Microsoft Visual Studio* with C++ 2017 or 2015, with MSBuild, and the Build Tools for Microsoft Visual Studio

Build Tools for Microsoft Visual Studio：
https://visualstudio.microsoft.com/zh-hans/downloads/

Build Tools for Microsoft Visual Studio：
https://visualstudio.microsoft.com/downloads/#build-tools-for-visual-studio-2015
https://visualstudio.microsoft.com/downloads/#build-tools-for-visual-studio-2017

2-2 下载 CMake >=3.4

下载地址：https://cmake.org/download/

2-3 下载 Python >= 3.6

推荐使用Anaconda安装python
下载地址：https://www.anaconda.com/distribution/

3 设置环境变量

运行

C:\Program Files (x86)\IntelSWTools\openvino\bin\setupvars.bat

4 配置模型优化器（Model Optimizer）

进入目录下

C:\Program Files (x86)\IntelSWTools\openvino\deployment_tools\model_optimizer\install_prerequisites

运行

install_prerequisites.bat

也可以根据需要运行对应的脚本

Caffe : install_prerequisites_caffe.bat
TensorFlow : install_prerequisites_tf.bat
ONNX (Caffe2, PyTorch, MXNet，ML. NET，TensorRT，Microsoft CNTK) : install_prerequisites_onnx.bat
MXNet : install_prerequisites_mxnet.bat
Kaldi : install_prerequisites_kaldi.bat

5 运行示例

运行示例检验是否安装成功。

进入目录：

C:\Program Files (x86)\IntelSWTools\openvino\deployment_tools\demo

5-1 图像分类

下载并将转换caffe模型为Intermediate Representation (IR)文件，并运行图像分类测试

demo_squeezenet_download_convert_run.bat

5-2 目标检测与识别

下载预训练模型用于车辆检测与车辆属性识别和车牌检测与车牌识别

demo_security_barrier_camera.bat

其他预训练的模型： https://github.com/opencv/open_model_zoo

可能遇到的问题

1 pip 安装下载不到

Could not fetch URL https://pypi.org/simple/tensorflow/: There was a problem confirming the ssl certificate: HTTPSConnectionPool(host=‘pypi.org’, port=443): Max retries exceeded with url: /simple/tensorflow/ (Caused by SSLError(“Can’t connect to HTTPS URL because the SSL module is not available.”)) - skipping
Could not find a version that satisfies the requirement tensorflow>=1.2.0 (from -r …\requirements.txt (line 1)) (from versions: )
No matching distribution found for tensorflow>=1.2.0 (from -r …\requirements.txt (line 1))

到对应install_prerequisites_xxx.bat下找到

pip install --user -r …\requirements%postfix%.txt

找到对应的requirements_xxx.txt文件，手动添加下载源安装
OpenVINO视觉加速库依赖包打包下载
如：

pip install tensorflow=1.2.0 -i https://pypi.tuna.tsinghua.edu.cn/simple

如果还是下载不到就只能去清华镜像源手动下载安装了
打包下载：OpenVINO视觉加速库依赖包 https://download.csdn.net/download/shanglianlm/11114728

参考资料:
1 OpenVINO™ toolkit Documentation
2 Install Intel® Distribution of OpenVINO™ toolkit for Windows* 10
3 Install Intel® Distribution of OpenVINO™ toolkit for Linux*
4 Introduction to Intel® Deep Learning Deployment Toolkit