Cannot Make Standalone Model

[Brown_Kramer__Joshua]

I am trying to embed a call to OpenVINO inference engine in an existing system in Windows 10.  I have a model and I have successfully run inference with it on the GPU with a python script.  Now I want to do inference in c++.  We don't want to go through the entire install process for every machine we put our system on, so ideally we would just ship the necessary dlls.  I have copied the contents of openvino\inference_engine\bin\intel64\Debug to the bin directory of my project.  At first there were several missing dlls, including ngraphd.dll and tbb_debug.dll.  I tracked these down (I think) and put them in the bin directory as well.  I no longer get a missing dll execption, but I do get the following:

Below is my cpp code, which I am calling from C#.  Any help would be great.

#include <inference_engine.hpp>
#include "pch.h"
#include "OpenVinoBridge.h"
#include <string>

using namespace InferenceEngine;
using namespace OpenVinoBridge; 

ExecutableNetwork executable_network;
void InferenceEngineBridge::Initialize()
{
    //Set up OpenVino inference 
    Core ie;
    std::string input_model_base = "resnet50-depthFrame7-048313-2365p0991-1402p3900";
    CNNNetwork network = ie.ReadNetwork(input_model_base + ".xml", input_model_base + ".bin");
    network.setBatchSize(1);
    executable_network = ie.LoadNetwork(network, "GPU");
}

[SIRIGIRI_V_Intel]

Are you able to run the cpp code on the system with OpenVINO installed? If yes, you can use Deployment Manager which creates a deployment package by assembling the model, IR files, your application, and dependencies into a runtime package for your target device.

Regards,

Ram prasad

[Brown_Kramer__Joshua]

I am trying to run the particular cpp code I pasted above on a system that has OpenVINO installed and it is not working.  I have been able to run inference through a python script on this system.

[Brown_Kramer__Joshua]

Some more info: I have been able to build the cpp examples and run them (although on Windows, reading jpegs seems to be broken).  I tried the Deployment Manager, which seems to make a copy of the inference engine and its dependencies as well as setupvars.bat.  I am confused about what to do with that.  I am trying to build a managed cpp project that compiles down to a DLL.  Is there some general guidance about what files have to be pointed to and where in order for things to work?

Thanks for your help.

-Josh Brown Kramer

[SIRIGIRI_V_Intel]

The required dlls are copied to the <path_to_openvino_deploy_package>/deployment_tools. Make sure the dlls are exist and run the setupvars.bat file located in <path_to_openvino_deploy_package>/bin. Now you are ready to the application.

It is recommended to use the supported languages(c, c++, python) in OpenVINO 2020.2

Regards,

Ram prasad