Frigate is a network video recorder that performs realtime object detection on IP camera feeds using local processing.
Frigate solves the problem of monitoring multiple IP cameras while identifying objects of interest without relying on cloud services. It runs object detection locally on the camera streams, allowing it to recognize people, cars, animals, and other objects as they appear. The system stores video and maintains a searchable database of detected events, letting users review what happened and when specific objects were present.
Frigate suits users who want to self-host their video surveillance with privacy guarantees and no dependency on external services. It works well for home automation setups, particularly those already using Home Assistant, since it integrates directly via MQTT and the Home Assistant add-on ecosystem. The tool is designed for setups with multiple IP cameras feeding RTSP streams and benefits from hardware accelerators like Google Coral TPUs to handle real-time inference efficiently. Users managing their own infrastructure who need local control and want to avoid subscription-based NVR services should consider it.
Development on the project shows consistent activity with regular commits addressing bug fixes and feature improvements. The maintainer responds actively to issues and pull requests, indicating engaged stewardship. The codebase reflects ongoing refinement of core detection and video processing logic rather than major architectural shifts. Documentation is maintained alongside code changes to keep setup and configuration guidance current.
