Introduction to TensorFlow on Jetson Nano

Objectives

• Installation of OS on Jetson Nano
• Setup of Tensor flow for jetson nano

Things

1. Jetson Nano
2. 32GB memory card
3. Card Reader
4. Power adapter for jetson 5v-2A
5. Monitor, Keyboard and Mouse

Let’s begin!

Jetson is a cheap on board computing solution provided by NVIDIA. The board is capable of handling computer vision and machine learning applications better than other boards available in the market in that segment. This is due to the presence of the 128 CUDA core. The Jetson Nano Developer Kit brings the power of modern AI to a low-cost platform, enabling a new wave of innovation from makers, inventors, developers and students. They can build AI projects that weren’t previously possible and take existing projects to the next level — mobile robots and drones, digital assistants, automated appliances and more.

Installation of OS on Jetson Nano

1. The OS can be downloaded for the NVIDIA website which is a 5GB image file. Once downloaded extract it to your PC.

2. Once the image is extracted we will use Rufus a freeware to burn the image to memory card

3. Install the downloaded application and launch it

4. Click on select and look for the extracted image file and select it, then click on start to begin the process.

5. Now insert the SD onto the Jetson Board as shown below and connect the power adapter monitor and keyboard to it.

6. The first boot would take 3-4 mins.

• Review and accept NVIDIA Jetson software EULA
• Select system language, keyboard layout, and time zone
• Create username, password, and computer name
• Log in

Once the initial setup is complete you will land to the screen shown below

Installing TensorFlow-GPU on the Jetson Board

1. Open the terminal on the device and type the following commands

       sudo apt-get update
       sudo apt-get upgrade
       sudo apt-get install python3-pip libhdf5-serial-dev hdf5-tool
       sudo pip3 install --pre --extra-index-url https://developer.download.nvidia.com/compute/redist/jp/v42 tensorflow-gpu 

 In case of errors like HTTP 404 Not found use the following command

pip3 install — pre — extra-index-url https://developer.download.nvidia.com/compute/redist/jp/v42 tensorflow-gpu

Hello World with tensorflow

gedit hello_world.py

Enter the following code to verify

To run the code save it, execute the following command

python3 hello_world.py