teachable-machine-pi

Implementation of Google Coral's Teachable Machine for Raspberry Pi Zero 2 W with Coral USB Accelerator.

This project is an implementation of the Teachable Machine example by Google Coral on a Raspberry Pi Zero 2 W with the Coral USB Accelerator.

Requirements

• Hardware:

  • Raspberry Pi Zero 2 W
  • Raspberry Pi Camera Module v2
  • Coral USB Accelerator

• Software:

  • Operating System: Raspbian GNU/Linux 10 (buster)
  • Python Version: 3.7 (default in Raspbian Buster)

Installation

1. Update the system:

   sudo apt-get update
   sudo apt-get upgrade

2. Install Edge TPU runtime:

   Add the Coral package repository and install the Edge TPU runtime:

   echo "deb https://packages.cloud.google.com/apt coral-edgetpu-stable main" | sudo tee /etc/apt/sources.list.d/coral-edgetpu.list
   curl https://packages.cloud.google.com/apt/doc/apt-key.gpg | sudo apt-key add -
   sudo apt-get update
   sudo apt-get install libedgetpu1-std

3. Install PyCoral library:

   Install the PyCoral library, which is required for communication with the Coral USB Accelerator:

   sudo apt-get install python3-pycoral

4. Clone the project repository:

   Clone the original Teachable Machine repository from Google Coral:

   git clone https://github.com/google-coral/project-teachable.git
   cd project-teachable

5. Install dependencies:

   Install the required Python dependencies:

   bash install_requirements.sh

Usage

1. Connect the Coral USB Accelerator:

   Plug the Coral USB Accelerator into the Raspberry Pi Zero 2 W.

2. Start Teachable Machine:

   Run the following script to start the Teachable Machine:

   bash run.sh

3. Train classes:

   • Key 1: Trains Class 1 with the current camera image.
   • Key 2: Trains Class 2 with the current camera image.
   • Key 3: Trains Class 3 with the current camera image.
   • Key 4: Trains Class 4 with the current camera image.
   • Key Q: Exits the program.

   Point the camera at an object and press the corresponding key to assign it to a class. Repeat this process for different objects and classes. Once trained, the system will classify detected objects in real-time.

Notes

• Python Version: Ensure you are using Python 3.7, as newer versions may not be compatible.
• Operating System: This project has been tested on Raspbian Buster. Other versions or distributions may cause compatibility issues.
• Performance: The Raspberry Pi Zero 2 W has limited resources. For better performance, a more powerful Raspberry Pi is recommended.

References

• Original Project: https://github.com/google-coral/project-teachable
• Coral Documentation: https://coral.ai/docs/accelerator/get-started/