Welcome to our collection of ready-to-run Jupyter* notebooks that helps you to quickly try out OpenVINO™ integration with TensorFlow. The notebooks carry out popular deep learning tasks like Image Classification, and Object Detection in TensorFlow and demonstrate it to developers how to leverage our simple two-liner API for optimized deep learning inference, without ever leaving the TensorFlow and Python* ecosystem.
The notebooks can be run on an Intel CPU running a supported version of the Ubuntu OS (currently 18.04 or 20.04)*. We recommend a Python* virtual environment to start the Jupyter* server.
You may need to install some additional libraries on Ubuntu* Linux. These steps work on a clean install of Ubuntu* Desktop 20.04, and should also work on Ubuntu* 18.04 and 20.10, and on Ubuntu* Server.
sudo apt-get update
sudo apt-get upgrade
sudo apt-get install python3-venv build-essential python3-dev git-all
If you have a CPU with an Intel Integrated Graphics Card, you can install the Intel Graphics Compute Runtime to enable inference on this device. The command for Ubuntu* 20.04 is:
Note: Only execute this command if you do not yet have OpenCL drivers installed.
sudo apt-get install intel-opencl-icd
First, let's clone this repo to get access to the notebooks
git clone https://github.com/openvinotoolkit/openvino_tensorflow
cd openvino_tensorflow
Now, create a Python* virtual environment and activate it
python3 -m venv openvino_tensorflow_env
source openvino_tensorflow_env/bin/activate
Let's install JupyterLab
python3 -m pip install jupyterlab
To launch a single notebook, like the TFHub Object Detection notebook
jupyter notebook examples/notebooks/OpenVINO_TensorFlow_TFHub_Object_Detection.ipynb
Alternatively, if you want to skip a local setup and want a stable runtime consider our docker instructions for runtime images. The images tagged latest start a Jupyter* server by default.
* Other names and brands may be claimed as the property of others.