OpenVINO & OpenCV

Prepare for OpenVINO Installation

• Download Python 3.12 or later

CMD> python --version
Python 3.10.9

• Download Git 2.45.0 or later

CMD> git --version
git version 2.40.1.windows.1

Clone this repo

CMD> cd %USERPROFILE%
git clone https://github.com/litcoder/ovcv

Clone OpenVINO Notebooks under Virtual Environment

Let's assume you're creating this under your home directory.

CMD> cd %USERPROFILE%
C:\Users\USER>

Create a virtual environment.

CMD> python -m venv openvino_env

You should see prompt changes after activating the virtual environment.

CMD> cd openvino_env
CMD> Scripts\activate
(openvino_env) C:\Users\USER\openvino_env>

Now clone the OpenVINO Notebook.

git clone https://github.com/openvinotoolkit/openvino_notebooks.git

Install Dependencies & Launch

Install dependnet Python modules from requirements.txt. This will take some time.

python -m pip install --upgrade pip
pip install -r requirements.txt
python -m ipykernel install --user --name openvino_env

Launch Jupyter Notebook

jupyter lab notebooks

OpenVINO Hands on

001-hello-world

• Download & Copy ImageNet Korean Labels
  Download ImageNet Korea Labels which was modified from todak_todak_python then copy it into openvino_notebooks\notebooks\data\datasets\imagenet\.

• Open the label
  Modify file path to indicate the new label file and pass encoding="utf-8" to open() call to avoid UnicodeDecodeError.

# Convert the inference result to a class name.
imagenet_classes = open("../imagenet/imagenet_2012_ko.txt", encoding="utf-8").read().splitlines()

• Play with your own image

• Quiz: List up 5 next possible items

  • Hint: Numpy argpartition()