Required components for OpenVINO runtime

[Shigeyoshi_O_Intel]

Hi,

May I ask what are the required components to run models which are created in another computer with same version of OpenVINO?

I find I can choose components when installing OpenVINO but cannot figure out which are must for runtime environment.

https://docs.openvinotoolkit.org/latest/openvino_docs_install_guides_installing_openvino_windows.html

Regards,

Ohara

Ohara

[Iffa_Intel]

Greetings,

First of all you need to ensure your environment satisfy the requirement of:

1.  Intel® Distribution of OpenVINO™ toolkit core components
2. CMake 2.8.12 or higher 64-bit
3. Python 3.5 - 3.7 64-bit

and the steps taken afterwards are to ensure functionality as in here: https://docs.openvinotoolkit.org/latest/openvino_docs_install_guides_installing_openvino_windows.html

If you are trying to convert from custom model such as YOLOv3, you need to convert your trained model into IR first which you can refer here: https://docs.openvinotoolkit.org/latest/openvino_docs_MO_DG_prepare_model_convert_model_tf_specific_Convert_YOLO_From_Tensorflow.html

However, if you are using the same openvino version and trying to migrate them into another environment, you still need to ensure you had done the installation steps & replace the exact folder that contain the one that you are trying to migrate.

Please note that different OS have different version of Openvino packages.

Sincerely,

Iffa

[Shigeyoshi_O_Intel]

Hello Iffa,

Thank you so much for the answer. I'm sorry that I didn't make myself clear enough. The question is about runtime. I think IE is required but MO is not. I wonder if you could tell me which of below are required?

• IE -> required
• MO -> not required?
• Model Zoo -> not required?
• Open CV -> ?
• CMake -> ?
• Python ->?

Regards,

Ohara

[Iffa_Intel]

In order to run Openvino samples or running your own model with Openvino or migrate it to another computer, you need to have the exact basic things as your previous successfully executed Openvino environment. 

Please help to go through the official Openvino Tutorial, the required components are already stated there: https://docs.openvinotoolkit.org/latest/openvino_docs_install_guides_installing_openvino_windows.html

Everything that being instructed in this tutorial ARE necessary.

If you had successfully installed  Intel® Distribution of OpenVINO™ toolkit core components

and followed the documentation that I shared previously, you should have these folders in your environment ( as the photo I attached). You should automatically have IE (Inference Engine) folder, ,MO(Model Optimizer) folder, and also open model zoo folder. These come as a package with the Toolkit installation which you don't have to manually search. Please note that these are pre-trained models which you could use and run directly.

You can move your old script/codes/models from old computer into exact same location in new computer. 

However, if you are trying to create or use a custom model which are not from in those folders, you need to do conversion from models, such as Tensorflow model, Mxnet model, Kaldi or ONNX into IR before you able to use it with Openvino.

Here is the official documentation:https://docs.openvinotoolkit.org/latest/openvino_docs_MO_DG_prepare_model_convert_model_Converting_Model.html

OpenCV is a library of programming functions mainly aimed at real-time computer vision where you need to install into Python library. However, if you had followed the installation steps, you should already have this since the script will auto install it for you.

CMake is a builder/compiler, you definitely need this ad you won't be able to compile your code without it. This is also stated in Openvino Official Tutorial, please help to read it.

Python is a programming language, for Openvino it is required, again every required components are stated in the official tutorial which I attached above.

Sincerely,

Iffa

[Iffa_Intel]

Greetings,

Intel will no longer monitor this thread since we have provided a solution. If you need any additional information from Intel, please submit a new question

Sincerely,

Iffa