Module 2: Install Azure IoT Edge on your Raspberry Pi

Index

Module 1: Create an Azure IoT Hub
Module 2: Install Azure IoT Edge on your Raspberry Pi
Module 3: Set up your development environment
Module 4: Create Azure Cognitive Services
Module 5: Build and deploy the solution
Module 6: Build and deploy the solution
Home

Hardware requirements

Raspberry Pi 4 recommended, 2GB or better
USB Camera
Speaker

Booting from high speed USB3 storage

As we will be building Docker images on the Raspberry Pi 4 so I would recommend a fast SD Card or a high speed USB3 Flash or SSD drive.

I use a Samsung USB 3.1 Flash Drive FIT Plus 128GB or USB3 SSD drive. See the 5 of the Fastest and Best USB 3.0 Flash Drives
For instruction on booting from USB3 see How to Boot Raspberry Pi 4 From a USB SSD or Flash Drive. Note, at the time of writing, set the FIRMWARE_RELEASE_STATUS to stable rather than beta.

Raspberry Pi set up

Create the Raspberry Pi OS Image

This project depends on ARM64 support and the recommended platform is Ubuntu Server 20.04 or 20.10 for Raspberry Pi. See How to install Ubuntu Server on your Raspberry Pi.

Start Raspberry Pi and update

Update and reboot

 sudo apt update && sudo apt full-upgrade && sudo reboot

Optionally overclock the Raspberry Pi 4.

Though not a requirement, the machine learning inference times will be improved by overclocking the Raspberry Pi 4. You will need a Raspberry Pi heat sink if you overclock. See the How to overclock Raspberry Pi 4 article for more information.

I use the following settings in the /boot/firware/config.txt.
```
 over_voltage=6
 arm_freq=2000
 gpu_freq=700
```
and reboot.

Install Azure IoT Edge on the Raspberry Pi

Follow the steps outlined in the Install or uninstall Azure IoT Edge for Linux guide. Follow the guidance for Ubuntu 18.04.

Configure the Azure IoT Edge device

You need to configure the Azure IoT Edge connection string.

Open the Azure IoT Edge config file.
```
 sudo nano /etc/iotedge/config.yaml
```

Scroll down to the Manual provisioning configuration section.

 # Manual provisioning configuration
 provisioning:
     source: "manual"
     device_connection_string: "<ADD DEVICE CONNECTION STRING HERE>"

Update the device_connection_string property with the Azure IoT Edge Connection string you saved to Notepad.
Press ctrl + X, then y to confirm you wish to save the config.
Press Enter to confirm the file name
Reboot the Raspberry Pi
```
 sudo reboot
```

Install the Docker Registry on the Raspberry Pi

Azure IoT Edge relies on Docker images being distributed from a Docker Registry. In production you would deploy Docker images from a registry such as Azure Container Registry.

When you are developing an Azure Iot Edge module it is faster to install a local container registry on the Raspberry Pi and deploy Docker images from the local registry to Azure IoT Edge.

Log into your Raspberry Pi

From the SSH session, run the following command.

 docker run -d -p 5000:5000 --restart=always --name registry registry:2

Do NOT close the SSH session.

Clone the image recognition solution to the Raspberry Pi

Log into your Raspberry Pi
```
 ssh pi@raspberrypi.local
```
From the SSH session, install the git client.
```
 sudo apt install -y git
```

From the SSH session, clone the solution repository to the Raspberry Pi

 git clone --depth 1 https://github.com/gloveboxes/Create-a-talking-image-recognition-solution-with-Azure-IoT-Edge-Azure-Cognitive-Services.git

NEXT