Install OpenVINO Toolkit Python Guide

OpenVINO Toolkit speeds up deep learning inference. It works well with Intel hardware. This guide shows you how to install it for Python.

We cover system requirements, pip setup, and conda setup. We also test the installation. Follow these steps to get started fast.

System Requirements

You need a computer with an Intel CPU, GPU, or VPU. Python 3.7 to 3.11 is supported. Use a 64-bit operating system like Windows, Linux, or macOS.

Make sure you have pip or conda installed. Check your Python version with python --version. This ensures compatibility.

Install Using pip

Using pip is the simplest method. It installs OpenVINO and its dependencies. Open your terminal or command prompt.

Run this command:

 
pip install openvino

This installs the core runtime. You can also install extra tools. For example, use pip install openvino-dev for model conversion tools.

Wait for the installation to finish. It may take a few minutes. Check the success message.

Install Using conda

Conda is another option. It works well with virtual environments. First, create a new environment to avoid conflicts.

 
conda create -n openvino_env python=3.9
conda activate openvino_env
conda install -c conda-forge openvino

This sets up a clean environment. Activate it every time you use OpenVINO. This method is great for managing multiple projects.

Verify Installation

After installation, verify it works. Open Python and import the library. Use this code:

 
# Check OpenVINO version
import openvino as ov
print("OpenVINO version:", ov.__version__)

Run the code. You should see a version number. For example:

OpenVINO version: 2024.0.0

If you see an error, check your Python path. Reinstall if needed. This confirms the toolkit is ready.

Install Additional Components

Some tasks need extra packages. For model optimization, install openvino-dev. It includes model converters and benchmarks.

 
pip install openvino-dev

For neural network compression, use openvino-pot. This helps reduce model size. Install it with pip:

 
pip install openvino-pot

These tools extend OpenVINO's capabilities. They are optional but useful for advanced work.

Run a Simple Inference

Test OpenVINO with a sample model. Download a pre-trained model from the Open Model Zoo. Use this code:

 
# Load a model and run inference
import openvino as ov
core = ov.Core()
model = core.read_model("path/to/model.xml")
compiled_model = core.compile_model(model, "CPU")
# Prepare input data
import numpy as np
input_data = np.random.randn(1, 3, 224, 224).astype(np.float32)
result = compiled_model([input_data])
print("Inference complete:", result)

Replace "path/to/model.xml" with your model. The output shows inference results. This proves your setup works.

Troubleshooting Common Issues

Some users face errors. A common one is DLL load failed. This happens on Windows. Install Microsoft Visual C++ Redistributable to fix it.

Another issue is module not found. Check your environment. Activate the correct conda environment or use the right Python interpreter.

If pip fails, upgrade it first. Use pip install --upgrade pip. Then retry the installation. These steps solve most problems.

Best Practices

Always use a virtual environment. This keeps dependencies isolated. Use venv or conda for this purpose.

Keep OpenVINO updated. New versions improve performance. Run pip install --upgrade openvino regularly.

Test with small models first. This helps you learn the API. Then move to larger projects for better efficiency.

Conclusion

Installing OpenVINO for Python is straightforward. Use pip or conda based on your preference. Verify the installation with a simple import.

Add extra tools like openvino-dev for advanced tasks. Run a sample inference to confirm everything works. Follow best practices for a smooth experience.

Now you are ready to accelerate your AI projects. Start optimizing models with OpenVINO today.