I would like to share my project HASS-Deepstack-object and face detection which enables the use of deep learning models in the home. This project builds upon the Deepstack project which is an open source AI server that is cross platform, and runs on Windows, Mac, Linux and Jetson devices. Deepstack is written in python and Go, and makes it easy for developers to begin using AI in their projects, simply by running a docker container which exposes AI services via a restful API. My project integrates Deepstack object and face detection & recognition services into the popular open source home automation platform Home Assistant. The community has found a wide variety of interesting use cases for this project, including:
- Monitoring activity in a brick factory in Latin America
- Watching for intruding snakes in Thailand
- Monitoring Amazon parcel deliveries
- Checking that a motorcycle was locked up
- Checking when a chicken laid an egg
- Greeting people when they return home and playing a theme tune
- Counting visitor numbers at a shop
- Checking when a parking spot became available
I link below a couple of videos showing use of the project:
The project code is at GitHub - robmarkcole/HASS-Deepstack-object: Home Assistant custom component for using Deepstack object detection
Deepstack exposes a generic yolov5 object detection model that provides accurate detection of people, animals and vehicles. In addition Deepstack can expose custom object detection models. I have also been working on developing a custom model to detect fires, with the goal of deploying this to monitor for fires via camera feeds.
I hope this project inspires readers to think about novel applications in and around the home & business that could benefit from AI, and enables rapid prototyping and experimentation.