Coral smart camera for Apple HomeKit

Read more on this project here!

This project shows how to build a Coral Edge TPU trainable smart camera and integrate it into Apple HomeKit. The project builds on HAP-python and consists of two main components:

• A Web-GUI to train the camera.
• A backend service to interact with HomeKit.

Installation

todo

Usage

Download model weights

Execute download.sh to download the pretrained embedding extractor. The extractor will be saved into the models directory.

Start the Application

To start the application you need to run both run.py and hap.py.

Imprint new weights

Access the application at yourhost:5000/. You can imprint new weights using the two buttons add_background and add_detection. Pressing on imprint_weights will update the classification model.