Repolode
Submit
blakeblackshear

blakeblackshear/frigate

Self-hosted NVR that runs realtime object detection on local IP camera feeds, built for Home Assistant integration.

Open-source alternative to:RingGoogle Nest

View on GitHubWebsite
Stars
36,842
Forks
3,740
Open issues
97
Last commit
22 hours ago

What is it?

Frigate is a network video recorder (NVR) for IP cameras that runs object detection locally rather than in the cloud. It uses OpenCV and TensorFlow to find objects in camera streams, and it is designed to pair closely with Home Assistant through a custom component.

To keep resource use low, Frigate runs a lightweight motion detector first and only invokes object detection in regions where motion occurs. Detection runs in separate processes, and the project favors realtime responsiveness over analyzing every frame. A GPU or AI accelerator is recommended for better performance.

Recordings are kept according to retention settings tied to detected objects, and 24/7 recording is supported. Frigate can re-stream camera feeds over RTSP to reduce connections to the camera, offers WebRTC and MSE live views, and publishes events over MQTT for other systems.

Key features

  • 01Local object detection on IP camera streams using OpenCV and TensorFlow
  • 02Motion detection gates object detection so it only runs where needed
  • 03Recording with retention rules based on detected objects, plus 24/7 recording
  • 04RTSP re-streaming and WebRTC/MSE live view
  • 05MQTT event publishing and a Home Assistant custom component

Use cases

  • Home security monitoring that alerts on people or vehicles from existing IP cameras
  • Reducing cloud subscription costs by keeping camera footage on local storage
  • Automating Home Assistant actions based on detected objects in camera views

Quick start

  1. 1

    Read the documentation for installation steps

  2. 2

    Visit the documentation site for setup instructions

Things to know

  • A GPU or AI accelerator is highly recommended for good performance.
  • Designed mainly for integration with Home Assistant.

In collections

Computer vision libraries & tools

Topics

#nvr#object-detection#home-automation#home-assistant#ip-camera#rtsp#mqtt#computer-vision#ai#camera

FAQ

What is frigate?
Self-hosted NVR that runs realtime object detection on local IP camera feeds, built for Home Assistant integration. Frigate is a network video recorder (NVR) for IP cameras that runs object detection locally rather than in the cloud. It uses OpenCV and TensorFlow to find objects in camera streams, and it is designed to pair closely with Home Assistant through a custom component.
Is frigate free to use? What license does it have?
frigate is open source under the MIT license. Check the license terms before using it in commercial products.
Is frigate still maintained?
Actively maintained. The last commit to frigate was on Oct 23, 2026.
What can I use frigate for?
Home security monitoring that alerts on people or vehicles from existing IP cameras Reducing cloud subscription costs by keeping camera footage on local storage Automating Home Assistant actions based on detected objects in camera views
How do I install frigate?
Read the documentation for installation steps Visit the documentation site for setup instructions
Where can I get frigate?
The source code and installation instructions are on GitHub: https://github.com/blakeblackshear/frigate

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects