← すべてのガイド
Frigate NVR
A self-hosted network video recorder with real-time AI object detection — it only alerts and saves clips for actual people, cars, or animals, instead of every motion event a plain security camera would trigger on.
対象: Any Docker host (a Google Coral USB accelerator is recommended but not required)security-camerasNVRAI-detection公式ホームページ ↗公式ドキュメント ↗
このガイドは上記リンク先にあるFrigate NVR公式ドキュメントをもとに作成されています。ディストリビューションやプロジェクトのバージョンは時間とともに変わるため、本番サーバーでこれらのコマンドを実行する前に、必ず公式ドキュメントで正確なパッケージ名・バージョン名を確認してください。
1Create a config.yml describing your cameras
Each camera entry needs its RTSP stream URL and which objects to detect (person, car, dog, etc.); Frigate's docs have a full reference for every option.
2Run the official container
A shared-memory size of at least 64MB per camera is required -- undersizing this is the most common first-run problem.
docker run -d \ --name frigate \ --restart unless-stopped \ --shm-size=256m \ -v ./frigate/config.yml:/config/config.yml \ -v ./frigate/storage:/media/frigate \ -v /etc/localtime:/etc/localtime:ro \ -p 5000:5000 -p 8554:8554 \ --device /dev/dri/renderD128 \ ghcr.io/blakeblackshear/frigate:stable3Open the Frigate web UI
Review live camera feeds, detection events, and saved clips; connect Frigate to Home Assistant for notifications via the official integration.
http://your-server-ip:5000