A passion project I’ve been working on is a custom NVR and UI for a home security system.

The initial build of the project was installing Home Assistant OS on a old computer of mine. Then I integrated Frigate NVR. Then connecting the real time streaming protocol(RTSP) feed to both systems. Home assistant provided a nice UI for the camera feed. Frigate allows for machine learning detection of objects by the system. Using tensor flow you can specify detection of cars, people or wildlife which allows for efficient storage of camera footage since you don’t need to store static footage. This solution worked well but in some situations it locked you into a box so I decided to recreate the system using Django.

I started by using OpenCV to read in my RTSP feed. Then put the entire site behind user authentication. Next I creating a PostgreSQL database.

I am currently working on database models used to storage URLs of the footage. Installing Tensor Flow and set it up to detect people and cars. Then store the footage objects in the PostgreSQL database. Lastly I need to implement revolving storage where the website automatically removes old footage to make room for new.