openvino部署yolov5

openvino部署yolov5

配置Visual Studio环境

打开属性管理器（视图—其他试图—属性管理器）

VC++目录

• 包含目录：配置openvino和opencv的include目录
• 库目录：配置openvino和opencv的lib依赖目录

链接器—输入

• 附加依赖项：配置需要用到的lib文件

包含目录
C:\open\Intel\openvino_2021\opencv\include
C:\open\Intel\openvino_2021\opencv\include\opencv2
C:\open\Intel\openvino_2021\deployment_tools\inference_engine\include

库目录
C:\open\Intel\openvino_2021\opencv\lib
C:\open\Intel\openvino_2021\deployment_tools\inference_engine\lib\intel64\Release

附加依赖项
inference_engine.lib
inference_engine_c_api.lib
inference_engine_ir_reader.lib
inference_engine_legacy.lib
inference_engine_lp_transformations.lib
inference_engine_onnx_reader.lib
inference_engine_preproc.lib
inference_engine_transformations.lib
opencv_calib3d452.lib
opencv_core452.lib
opencv_dnn452.lib
opencv_features2d452.lib
opencv_flann452.lib
opencv_gapi452.lib
opencv_highgui452.lib
opencv_imgcodecs452.lib
opencv_imgproc452.lib
opencv_ml452.lib
opencv_objdetect452.lib
opencv_photo452.lib
opencv_stitching452.lib
opencv_video452.lib
opencv_videoio452.lib

注意这里的配置包含库和附加依赖项需要区分debug和release环境，以上给出的是release环境下的配置，若在debug环境下配置：

• 需要将包含库目录换成

C:\open\Intel\openvino_2021\deployment_tools\inference_engine\lib\intel64\Debug

• 需要将openvino和opencv的lib文件名后面加上d，具体名称可以打开库目录查看

配置环境变量

配置完以上依赖后，运行程序会提示缺少xxx.dll，根据名称在openvino目录下搜索可以找到这些缺失的dll文件的所在目录，将其加入系统的环境变量Path即可

C:\open\Intel\openvino_2021\deployment_tools\ngraph\lib
C:\open\Intel\openvino_2021\inference_engine\tbb\bin
C:\open\Intel\openvino_2021\opencv\bin

此外还提示缺失vcruntime140_1d.dll缺失，在网上下载并复制到以下目录

C:\Windows\System32
C:\Windows\SysWOW64