Image Classification with Raspberry Pi and Google Coral USB

Introduction

This is an example based on Coral USB that performs Image Classification on Raspberry Pi. This example uses TensorFlow Lite with Python on a Raspberry Pi to perform real-time image classification using images streamed from the Pi Camera. It also uses HyperPixel 4.0 Display as the display output.

Features

• Real-time Image classification
• Small and compact demo

Hardware Connection

• Connect Raspberry Pi Camera v1.3 to the Raspberry Pi
• Connect the Coral Accelerator USB to the Raspberry Pi via USB
• Connect the HyperPixel 4.0 via RPI 40-pin

Software Preparations

• Before you begin, you need to set up your Raspberry Pi with Raspbian (preferably updated to Buster).
• Configure the camera as follow:
• Update the firmware using the following in the terminal:

sudo apt update
sudo apt full-upgrade

• Configure camera support, use sudo raspi-config, select Interface Options and select Camera and Enable the Camera options.
• Install the TensorFlow Lite runtime:

pip3 install https://dl.google.com/coral/python/tflite_runtime-2.1.0.post1-cp37-cp37m-linux_armv7l.whl

• Download the Example file from Coral:

git clone https://github.com/tensorflow/examples --depth 1

• Use the script to install python packages and download models and labels:

cd examples/lite/examples/image_classification/raspberry_pi

# The script takes an argument specifying where you want to save the model files
bash download.sh /tmp

Coral USB Accelerating

• Follow Coral to install Coral USB into the Raspberry Pi

Now, open classify_picamera.py file and change as follow:

• Add the following line at the top:

from tflite_runtime.interpreter import load_delegate

• And then find the line that initializes the Interpreter, which looks like this:

interpreter = Interpreter(args.model)

And change it to specify the Edge TPU delegate:

interpreter = Interpreter(args.model,
experimental_delegates=[load_delegate('libedgetpu.so.1.0')])

and save the changes.

• Run the Example!

python3 classify_picamera.py \
  --model /tmp/mobilenet_v1_1.0_224_quant.tflite \
  --labels /tmp/labels_mobilenet_quant_v1_224.txt

Display on HyperPixel

Run the one-line installer:

curl -sSL https://get.pimoroni.com/hyperpixel4 | bash

And choose Raspberry Pi 3B+ model.