Skip to content

Latest commit

 

History

History
159 lines (105 loc) · 7.07 KB

DEMO Setup.md

File metadata and controls

159 lines (105 loc) · 7.07 KB
Raw

Setup for AIML20 demos

These tasks were performed ahead of time for the AIML20 presentation. Follow these same steps to prepare for the remaining demo scripts.

Clone the repository to your local machine

You will need the contents of https://github.com/microsoft/ignite-learning-paths-training-aiml/tree/master/aiml20 on your local machine. The easiest way to do this is to visit the Developer's Guide to AI Learning Paths repository and click the "Clone or download" button. We will refer to files relative to the aiml20 folder.

Sign up for an Azure Subscription

If you don't already have an Azure subscription, you can sign up here and also get $200 in free Azure credits to use.

Find your Azure Subscription ID

In the Azure Portal, sign in and click on "Subscriptions" in the left menu bar. Click on the Subscription Name you will be using, and copy the "Subscription ID" shown there. You'll need it later when you create resources.

Alternatively, run az account show in the Azure CLI and copy the id value shown.

Start fresh.

If you've run the demos before, delete these resource groups if they exist:

  • aiml20-demo

You can use the Azure Portal, or run this command in the Azure CLI:

az group delete --name aiml20-demo

Deploy the Tailwind Traders website.

Click the button below. This will deploy TailwindTraders-Website from Github, using an ARM template to enable the Personalizer integration and ONNX-based Shop by Photo feature. (More details about this deployment.)

Deploy to Azure

In the form that appears, select the following options:

  • Subscription: Select the subscription in which to deploy the resources

  • Resource Group: resources will be created in the resource group you specify. We recommend choosing "New" and entering the name aiml20-demo. When you're done, you can delete this resource group to shut down the site and delete all associated resources.

  • Location: The Azure region where the resources will be deployed. You must be able to deploy SQL Database and App Services in that region.

    Note: Since Personalizer is currently only available in WestUS2 and WestEurope, it will be deployed there regardless of what you choose.

  • Site Name: This will be used in the site's URL and visible publicly, and must be globally unique. To avoid clashes, choose aiml20-xy replacing xy with your initials, but we will refer to this name as just aiml20 in these scripts. (If you choose a name that is in use, you will get "Conflict" errors during the deployment.)

  • Site Location: Enter the short version of "Location" above, e.g. westus2. (You can get a list of short names in the Azure CLI with: az account list-locations -o table).

  • Deployment mode: Choose standalone

  • SQL Login: Enter twt

  • SQL Password: generate and use a secure password (it must include punctuation and mixed case, but do not use ;). You won't need it for our demos, so no need to write it down.

  • Enable Personalizer: choose true

  • Repo URL: accept the default, https://github.com/microsoft/TailwindTraders-Website

  • Branch: accept the default, master

(As a backup, you can also use the Repo URL https://github.com/revodavid/TailwindTraders-Website with the branch aiml20. This was forked on 2019-10-25 and is known to work.)

Check "I agree to the terms and condtions" and click "Purchase".

Allow at least 15 minutes for the site to deploy.

The deployed website URL will be of the form SITENAME.azurewebsites.net (using the Site Name you provided above), or you can find it as follows:

  • click "Go To Resource" under "Next Steps"

  • Click the "App Service" resource

  • Look at the "URL" value displayed in the right pane

The website URL will be displayed after the "Setting up Source Control" step, or you can inspect the "App Service" resource.

Install the "Simple" ONNX model

Follow the instructions in DEMO ONNX deployment.md under the heading "Load the Simple ONNX model". This will degrade the "Shop by Photo" tool in the app to only recognize hammers and drills.

Configure Visual Studio Code

Install the extension Azure Account. (On Windows, you will also need to install node.js.) In VS Code, log into Azure with the "Azure: Sign In" command (use Control-Shift-P to open the Command Palette). To run Azure CLI commands from a script in VS Code, use "Terminal: Run Selected Text in Azure Terminal" to copy commands.)

Alternatively you can install the Azure CLI on your local Windows, MacOS or Linux machine. If you don't have it installed, you can also launch the Azure Cloud Shell and run these commands from a browser window.

Prepare Visual Studio for demo

  • Open vision_demo.sh
  • launch a Cloud Shell with "Azure: Open Bash In Cloud Shell". (If you prefer, you can use the Azure CLI locally.)

Open browser pages ready to demo.

Find the image files on your local machine

Once you have cloned the repository, you can find the training images in the aiml20/CV Training Images folder. You will use these images to train the Custom Vision model. The folder contains the following subfolders:

  • drills
  • hammers
  • hard hats
  • pliers
  • screwdrivers

These images will be used to test the Computer Vision service and create a model with the Custom Vision service.

These images were sourced from Wikimedia Commons and used under their respective Creative Commons licenses. See the file ATTRIBUTIONS.md for details.

Additional test images can be found in the test images folder. These images will not be used in training, but will be used to test that our models are working.

Next Step

Computer Vision