Adding Google Colab Support to All Notebooks

[sawradip]

Bringing the topic From #840

Google Colab Support

Several of the notebooks in the Openvino notebooks repository have a dependency on external data and utility functions, which are either contained in corresponding directory, or downloaded on the fly. For instance, some notebooks import data from the utils and data folders, which are not always present in remote environments like Google Colab (Much more accesible than currently used Binder, I feel).

sys.path.append("../utils")
import notebook_utils as utils

This makes it difficult for users to run the notebooks without manually copying the require directory into the notebook.

Thus, adding support for Google Colab would successfully make the notebooks accessible to an wider range of users, as well as make quick testing and tinkering much easier.

Major Problems

1. Installation: There are some libraries that are not pre-installed in Colab, that need to be installed for that specific model.
2. Common Functions: Some functions are such as download_file or read_image are implemented at a common place, and we import them, but when we open in Colab, the directory is not available.
3. Weights and Files: Some necessary files are also provided with the repository, and those are read or used in the corresponding directory. Those files can be dependency installation requirements.txt file, model weights in onnx or xml and bin format, testing images etc.

Solution

Simplest solutions are often best. I have tested this idea with 219-knowledge-graphs-conve notebook.
The solution I am thinking is,
Step 1: Clone this repository into the notebook.

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

Step 2: Change the directory to that corresponding notebook path.[Solves Problem 2 & 3]

%cd /content/openvino_notebooks/notebooks/219-knowledge-graphs-conve

Step3: Install the requirements.txt file.[Solves Problem 1]

! pip install -r ../../requirements.txt

Adding these 3 line(which can be turned into a simple code-block, where the only variable will be the notebook name) makes the notebooks Google Colab compatible.

Example

I have tried this approach that specific approach, because is has dependency on current directory files as well as distant relative directory.
You can check the Google Colab version here: 219-knowledge-graphs-conve

[sawradip]

@eaidova @adrianboguszewski Your thoughts?