Frigate NVR
Frigate NVR is an open source network video recorder (NVR) that leverages real-time AI object detection. This tool performs all processing ...
Last verified:
What is Frigate NVR?
Frigate is an open source Network Video Recorder (NVR) built around real-time AI object detection for IP cameras. All processing is performed locally on your own hardware, ensuring camera feeds never leave your home for maximum privacy. It uses OpenCV and Tensorflow to perform realtime object detection locally, making it the most popular NVR for privacy-focused home automation enthusiasts.
Key features include local AI object detection that reduces false positives compared to traditional motion detection, smart streaming that updates camera feeds once per minute when no activity is detected, zones for fine-tuning events and alerts based on precise locations, and deep integration with Home Assistant, OpenHab, NodeRed, and any MQTT-supported platform. Frigate can run 100+ object detections per second with a supported AI accelerator like Google Coral, and it includes built-in face recognition and license plate recognition in version 0.16.
Frigate is designed for DIY home security enthusiasts, Home Assistant users, privacy-conscious individuals who want to remove cloud dependencies from their security cameras, and anyone looking for a complete locally-controlled security camera system with advanced AI features without paying for cloud processing.
The Review page lets users quickly review historical footage with bandwidth-optimized previews on a vertical timeline, while the Birdseye view provides a unified overhead view of all cameras tracking objects in real-time. Frigate supports package detection to track deliveries and recognizes common delivery service logos like UPS, Fedex, Amazon, and DHL.
Frigate NVR pricing
Pricing model: Freemium
Frigate is completely free and open source with no paid tiers for the core NVR functionality. Frigate+ is a optional annual subscription that provides access to Frigate+ models trained on actual security camera images from users worldwide, up to 12 fine-tuned models per year, AI suggested labels for uploads, and support for open source Frigate. Additional fine-tunings beyond 12 cost $5 each. Models trained during subscription remain available for download indefinitely even if subscription is not renewed. Facial and license plate recognition works with or without Frigate+ subscription.
Frigate NVR pros
- Open source with no cloud dependencies
- All processing performed locally on your hardware
- Real-time AI object detection with 100+ detections per second
- Reduces false positives vs traditional motion detection
- Deep Home Assistant integration with media browser and low latency camera entities
- Built-in face recognition in version 0.16
- Built-in license plate recognition in version 0.16
- Zones for precise location-based alerts
- Smart streaming conserves bandwidth by updating once per minute when no activity
- Supports Google Coral EdgeTPU AI accelerator
- Supports Intel OpenVINO detector for GPU acceleration
- Supports Hailo-8 AI accelerator
- Supports MemryX MX3 Accelerator
- Supports Rockchip platform NPU
- RTSP restreaming on port 8554
- WebRTC support for two-way talk on port 8555
- HTTP API for automation and integrations
- Birdseye view with unified overhead camera view
- Package detection to track deliveries
- Recognizes delivery service logos (UPS, Fedex, Amazon, DHL)
Frigate NVR cons
- Docker container installation requires technical knowledge
- Windows not officially supported
- Running in VMs (Proxmox, ESXi, Virtualbox) not recommended
- Requires MQTT broker for Home Assistant integration
- Shared memory (shm-size) must be calculated and configured for multiple cameras
- Raspberry Pi requires external powered USB hub for USB Coral
- Android TV box support not mentioned
- Intel GPUs not supported in Home Assistant OS
- Nvidia GPUs not supported in Home Assistant Apps
- Custom object labels cannot be added yet
Frequently asked questions about Frigate NVR
What is Frigate?
Frigate is an open source NVR built around real-time AI object detection for IP cameras. All processing is performed locally on your own hardware, and your camera feeds never leave your home. It uses OpenCV and Tensorflow to perform realtime object detection locally.
Do I need a Frigate+ subscription for facial and license plate recognition?
No. Facial and license plate recognition is built directly into Frigate starting with version 0.16. This works with or without a Frigate+ subscription. Frigate+ models just improve efficiency by detecting the location of faces and license plates at the same time as other objects.
What happens to my trained models if I cancel my Frigate+ subscription?
Models trained during your subscription are available for download indefinitely even if you decide not to renew. If you want a one-time purchase option, you can simply cancel your subscription and it won't renew at the end of the year.
Do I need Frigate+ to use Frigate?
You don't have to. Subscribers have direct access to the base models trained on the images from other Frigate+ users, but the core Frigate NVR functionality is completely free and open source.
Can I add my own custom object labels?
Only specific labels are supported. You can't add your own custom labels yet. This is a current limitation of the system.
What AI accelerators does Frigate support?
Frigate supports Google Coral EdgeTPU (USB and PCIe), Intel OpenVINO for GPU acceleration, Hailo-8 and Hailo-8L AI accelerators, MemryX MX3 Accelerator, Rockchip platform NPU, and Synaptics SL1680. CPU detection is also available as the default option.
How does Frigate reduce false positives?
Traditional NVRs rely on simple motion detection and can require hours of fine tuning to reduce false positive rates. Frigate offloads object detection to a supported AI accelerator, allowing even modest hardware to run advanced analysis to determine if the motion is actually a person, car, or other object of interest.
What is smart streaming in Frigate?
Frigate employs smart streaming where camera images update once per minute when no detectable activity is occurring to conserve bandwidth and resources. As soon as any motion or active objects are detected, cameras seamlessly switch to a live stream.
How do I install Frigate?
Frigate is a Docker container that can be run on any Docker host including as a Home Assistant App. It runs best with Docker installed on bare metal Debian-based distributions. The Home Assistant App is different from the integration - the integration is required to integrate Frigate into Home Assistant regardless of installation method.
What objects can Frigate detect?
Frigate includes object labels from Google Coral test data including person, bicycle, car, motorcycle, airplane, bus, train, boat, bird, cat, dog, horse, sheep, cow, elephant, bear, zebra, giraffe, and many more. With Frigate+ models, you can also detect wildlife like foxes, raccoons, squirrels, and deer, as well as waste bins and robot lawnmowers.