Machine Learning on a Raspberry Pi: The Complete Guide to Running TensorFlow Lite on Your Raspberry Pi 3 2023 (Mac)

Wiktoria Kasprzak
2 min readJul 5, 2023

--

Picture captured from the image_segmentation model identifying dog and background.

Introduction

The previous article in this series, titled “Machine Learning on a Raspberry Pi: The Complete Guide to Setting Up Your Raspberry Pi 3 2023 (Mac)”, covered the initial setup of a Raspberry Pi 3 and configuring it for Machine Learning tasks. This guide will provide step-by-step instructions for running TensorFlow Lite models on your Raspberry Pi 3 using a Mac.

Prerequisites

Before proceeding with running TensorFlow Lite models on your Raspberry Pi 3, make sure you have completed the setup process outlined in the previous article.

Setting up TensorFlow Lite on Raspberry Pi

1. Clone the TensorFlow Examples Repository

To get started, navigate to the TensorFlow GitHub repository and clone the repository to your Raspberry Pi. Open a terminal on your Raspberry Pi and run the following command:

git clone https://github.com/tensorflow/examples.git

This will create a local copy of the TensorFlow Examples repository.

2. Access the TensorFlow Lite Models

Navigate into the TensorFlow Examples directory and then into the lite folder. From there, go into the examples folder. This can be done using the command below. This is where you will find various TensorFlow Lite models.

cd examples/lite

3. Select a Model

Within the examples folder, you will find several TensorFlow Lite models. Note that the selected model should have a raspberry_pi folder associated with it. At the time of writing, these include:

  • Audio classification
  • Image classification
  • Image segmentation
  • Object detection
  • Pose estimation
  • Sound classification
  • Video classification

Choose the model that aligns with your project’s requirements.

4. Execute the Main Program

Once you have selected the model of your choice, navigate into its directory, which should have the name corresponding to the model’s topic, for example image_segmentation. Inside the model’s directory, you will find the main program file, such as segment.py in this case. Execute the main program using the following command:

python model_file_name.py

Alternatively, you could run the model by opening the file using the GUI and executing the programme by clicking the respective button under the build drop-down in the Geany IDE:

Geany build drop-down in VNC viewer.

Conclusion

You have successfully set up and run a TensorFlow Lite model on your Raspberry Pi 3 using your Mac! This guide has provided you with a comprehensive overview of the steps involved in running TensorFlow Lite on a Raspberry Pi.

--

--