Training model using Azure Machine Learning and Jupyter Notebook VM

What you will do

  • Create a Notebook VM to your Azure Machine Learning Workspace
  • Train a model using Azure Machine Learning and Notebook VM. Deploy the model to your Vision AI Dev Kit device

Pre-requisites

  • Vision AI DevKit camera
  • Azure Machine Learning workspace

Create Notebook VM

  • Log in to Azure Portal and access Azure Machine Learning workspace
  • Create a new Notebook VM
  • Create new Notebook VM

Clone example notebooks from GitHub

  • Add an access token to your GitHub account (if not exist yet). This is needed for being able to clone a private GitHub. In this case the password when cloning a GitHub is your token.
  • Open the Notebook VM Create new Notebook VM
  • Open “Terminal” Create new Notebook VM
  • Note that if you use Edge as your browser the terminal many not open directly. In that case you can access it from “Running” tab.

Create new Notebook VM

git clone --recursive https://github.com/microsoft/vision-ai-developer-kit

Prep your environment

  • Change the aml_config/config.json file to match the subscription details for your Azure ML Workspace
  • In case you want to use your own data go to the data folder and create a folder with your data, e.g. my_data.
    • In the my_data folder, copy your data. Each folder name is the label of the images in that folder. For example, the soda_cans folder looks like this:
      • soda_cans
      • coke
      • ice
      • pepsi
  • IMPORTANT: You MUST upload more than 35 images per folder so there’s enough data for training.

Run the notebook

  • Open the 02-mobilenet-transfer-learning-final.ipynb notebook

  • Set the Kernel to Python 3.6

Set kernel

  • Run the notebook by running each cell individually

Run notebook

  • In case you don’t want to run a cell you can change it to markdown

Markdown cell

Configure your IoT Hub details

  • Change the cell content to reflect your IoT Hub. Please note that you can use format:
    • Iot_hub_name=”youriothub”

IotHub Details

Updated: