Frigate NVR: The Complete Guide to AI-Powered Home Security Cameras

Why self-hosting your security camera system is the smartest privacy move you can make in 2026

---

Picture this: It’s 2 AM, and your phone buzzes with a notification from your security camera. “Motion detected.” You groggily open the app to find… a stray cat walking across your lawn. Again. For the third time tonight.

If you’ve ever used cloud-based security cameras, you know this frustration all too well. Traditional motion detection is notoriously unreliable—triggering on shadows, swaying trees, passing cars, and literally nothing at all. But what if your cameras were actually intelligent? What if they could tell the difference between a delivery driver, a stray animal, and an actual intruder?

Enter Frigate NVR—the open-source, AI-powered Network Video Recorder that’s revolutionizing how homelab enthusiasts approach home security.

What is Frigate NVR?

Frigate NVR is a self-hosted, AI-driven network video recorder that processes your security camera feeds locally using real-time object detection. Unlike cloud-based systems that upload your footage to distant servers, Frigate runs entirely on your hardware, analyzing video streams with sophisticated AI models to identify people, vehicles, pets, and other objects.

The result? Dramatically fewer false alarms, complete control over your data, and zero monthly subscription fees.

Why Frigate Matters in 2026

The home security landscape has shifted dramatically. High-profile privacy breaches at companies like Ring and Nest have exposed the uncomfortable reality of cloud-based camera systems: your footage isn’t really yours. Law enforcement partnerships mean your videos could be accessed without your knowledge. And those monthly subscription fees? They add up to hundreds of dollars over just a few years.

Frigate has matured into a production-ready solution:

• Version 0.16 (August 2025) made face recognition and license plate recognition free for all users—features that cost extra with commercial systems
• Version 0.17 added NVIDIA GPU support with CUDA Graphs and AI-generated review summaries
• The community has grown exponentially, with active support channels and extensive documentation
• Hardware options have expanded beyond the Google Coral to include NVIDIA GPUs, Intel OpenVINO, and Hailo accelerators

Key Features and Capabilities

Real-Time AI Object Detection

Frigate’s core strength is its ability to identify objects in real-time using AI models like YOLO. Out of the box, it recognizes:

• People—The most common detection target
• Vehicles—Cars, trucks, motorcycles, bicycles
• Pets—Dogs, cats
• And 70+ other object types from the COCO dataset

This isn’t simple motion detection. The AI actually understands what it’s looking at, meaning you only get alerts for things that matter.

Zone-Based Detection

Define polygonal zones around your property and configure exactly what should trigger alerts in each area:

cameras:
  front_door:
    required_zones:
      - porch
    zones:
      porch:
        coordinates: 100,400 300,400 300,600 100,600
        objects:
          - person
    detect:
      enabled: true

Want to know when a car enters your driveway but ignore cars passing by on the street? Zones make that trivial.

Zero False Alert Configuration

With proper tuning—motion masks, zone requirements, and object filters—Frigate can achieve essentially zero false alerts. The system uses multiple layers of detection:

1. Motion detection filters out static scenes
2. Motion masks ignore problematic areas (trees, timestamps)
3. AI object detection identifies what actually moved
4. Zone requirements ensure objects are in relevant areas
5. Object filters set confidence thresholds

Feature-Rich Recording

• 24/7 continuous recording for complete coverage
• Event-based recording to save storage
• Hybrid mode with pre/post recording buffers
• Intelligent retention policies that automatically manage storage

Enrichment Features

Starting with version 0.16, Frigate includes powerful enrichment features at no cost:

• Face Recognition: Identify family members, frequent visitors, or unknown individuals
• License Plate Recognition (LPR): Track vehicles entering your property
• AI-Generated Summaries: Quick overviews of daily activity (v0.17+)

Seamless Home Assistant Integration

Frigate integrates beautifully with Home Assistant, enabling:

• Real-time camera feeds in your dashboard
• Binary sensors for detected objects
• Automation triggers (e.g., “turn on lights when person detected at night”)
• Mobile notifications with snapshots

Hardware Requirements

Here’s where Frigate gets interesting. Unlike cloud cameras that offload processing to someone else’s servers, you need hardware capable of running AI inference locally.

Minimum vs Recommended Hardware

Component	Minimum	Recommended	High-End
RAM	4GB	8-16GB	16GB+
CPU	Modern multi-core	Intel i5/i7 with Quick Sync	Multi-core server CPU
Storage	SSD preferred	SSD + HDD for archives	Enterprise SSD/NVR HDD
Network	Gigabit	Gigabit	10GbE for large setups

AI Accelerators: The Big Change

Important note for 2026: The Google Coral Edge TPU, once the go-to accelerator for Frigate, is no longer recommended for new installations. While still supported, newer options offer significantly better performance.

Current Recommendations:

NVIDIA GPUs (Best Performance)

• Requirements: CUDA 12.x support, Compute Capability 5.0+ (Maxwell or newer)
• Performance: RTX 4060 achieves ~6ms detection latency with TensorRT
• Best for: Large installations (12+ cameras), users wanting cutting-edge performance

Intel OpenVINO (Best Value)

• Options: Integrated GPUs (6th Gen Intel Core+) or Intel Arc discrete GPUs
• Benefits: Low power, cost-effective, works with existing Intel systems
• Best for: Budget builds, low-power setups

Hailo-8/8L (Balanced Choice)

• PCIe-based accelerator with automatic model selection
• High performance without GPU power consumption
• Best for: New installations wanting dedicated AI hardware

Google Coral (Legacy Support)

• Still works and supported, but superseded by newer options
• Best for: Existing installations, low-power systems like Raspberry Pi

Camera Requirements

• Protocol: RTSP support required
• Video: H.264 video, AAC audio for best compatibility
• Substreams: Highly recommended—use low-res stream for detection (saves AI resources), high-res for recording
• Resolution: Detection stream at 720p-1080p; recording can be 4K

Installation Guide: Docker Setup

Frigate runs in Docker, making installation straightforward across platforms. Here’s a complete setup guide:

Step 1: Create Directory Structure

mkdir -p /opt/frigate/config
mkdir -p /opt/frigate/media

Step 2: Create docker-compose.yml

version: "3.9"
services:
  frigate:
    container_name: frigate
    privileged: true
    restart: always
    image: ghcr.io/blakeblackshear/frigate:stable
    shm_size: "1024mb"
    volumes:
      - /opt/frigate/config:/config
      - /opt/frigate/media:/media/frigate
      - /etc/localtime:/etc/localtime:ro
    ports:
      - "5000:5000"      # Web UI (unauthenticated)
      - "8971:8971"      # Web UI (authenticated)
    environment:
      - FRIGATE_RTSP_PASSWORD=your_secure_password

Step 3: Start Frigate

docker compose up -d
docker logs frigate  # Check the logs for the initial password

Step 4: Access the Web UI

Navigate to http://your-server-ip:5000 to access Frigate’s web interface. From here, you can configure cameras, tune detection settings, and view live feeds.

Configuration Basics

Frigate’s configuration lives in /config/config.yml. Here’s a practical starting point:

mqtt:
  enabled: true
  host: 192.168.1.100
  user: frigate
  password: your_mqtt_password

detectors:
  openvino:
    type: openvino
    device: GPU

cameras:
  front_door:
    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://camera-ip:554/stream1
          roles:
            - record
        - path: rtsp://camera-ip:554/stream2
          roles:
            - detect
    detect:
      enabled: true
      width: 1280
      height: 720
    record:
      enabled: true
      retain:
        days: 7
        mode: motion
    zones:
      porch:
        coordinates: 100,400 300,400 300,600 100,600
        objects:
          - person
    required_zones:
      - porch

This configuration:

1. Sets up MQTT for Home Assistant integration
2. Uses Intel OpenVINO for AI detection (adjust for your hardware)
3. Configures a camera with dual streams (detect + record)
4. Defines a zone and requires objects to enter it before triggering alerts
5. Records for 7 days on motion

Achieving Near-Zero False Alerts

The key to frustration-free security? Layer your detection filters:

cameras:
  front_yard:
    motion:
      threshold: 30          # Higher = less sensitive
      contour_area: 100      # Filter small motion areas
    motion_masks:
      - coordinates: "0,0,200,0,200,100,0,100"  # Mask tree area
    objects:
      track:
        - person
        - car
      filters:
        person:
          min_score: 0.65    # Minimum AI confidence
          min_area: 1000     # Minimum object size
          threshold: 0.7     # Detection threshold

Home Assistant Integration

Frigate and Home Assistant are a perfect pairing. Here’s how to connect them:

Step 1: Install via HACS

HACS → Integrations → Search "Frigate" → Download → Restart Home Assistant

Step 2: Add the Integration

Settings → Devices & Services → Add Integration → Frigate

Enter your Frigate URL (typically http://192.168.1.x:5000) and authenticate.

Step 3: Automation Examples

Notify when someone’s at the door:

automation:
  - alias: "Person at Front Door"
    trigger:
      - platform: state
        entity_id: binary_sensor.front_door_person
        to: "on"
    action:
      - service: notify.mobile_app
        data:
          message: "Person detected at front door"
          data:
            image: "/api/frigate/notifications/{{ state_attr('binary_sensor.front_door_person', 'event_id') }}/thumbnail.jpg"

Turn on lights when someone approaches:

automation:
  - alias: "Front Yard Motion Lights"
    trigger:
      - platform: state
        entity_id: binary_sensor.front_yard_person
        to: "on"
    condition:
      - condition: state
        entity_id: sun.sun
        state: "below_horizon"
    action:
      - service: light.turn_on
        target:
          entity_id: light.front_porch

The Frigate HACS card provides a beautiful dashboard with live streams, event timelines, and quick scrubbing through recordings.

Privacy Benefits vs Cloud Solutions

This is where Frigate truly shines. Let’s compare:

Aspect	Frigate (Self-Hosted)	Cloud Cameras (Ring/Nest)
Data location	Your hardware	Company servers
Third-party access	None without your consent	Law enforcement partnerships exist
Data breaches	Not applicable to you	Several high-profile incidents
Subscription required	No	Yes ($10-30/month)
Works offline	Yes	No
You control retention	Yes	Limited options
AI processing location	Your hardware	Their cloud

The Cloud Camera Privacy Problem

Cloud camera companies have faced numerous privacy controversies:

• Ring’s partnerships with over 2,000 law enforcement agencies
• Reports of employees accessing customer footage
• “Deleted” footage reportedly recoverable by authorities
• Data mining clauses in privacy policies
• No end-to-end encryption by default

With Frigate, your footage stays on your network. Period. You control who sees it, how long it’s kept, and how it’s accessed. For privacy-conscious homelab enthusiasts, this alone is reason enough to make the switch.

Cost Comparison

Let’s talk money. Because self-hosting saves a lot over time.

Initial Hardware Costs

Build	Components	Cost
Budget Frigate Build	Intel N100 Mini PC + 16GB RAM + 500GB SSD	~$280-330
Performance Build	Used Optiplex i5-8500 + 16GB RAM + 1TB SSD	~$300-400
Synology NAS Solution	DS220+ + 4TB HDD x2 + Camera Licenses	~$620

5-Year Total Cost of Ownership

Solution	Initial Cost	Annual Cost	5-Year Total
Frigate DIY (budget)	$300	$0	$300
Frigate DIY (performance)	$350	$0	$350
Synology NAS	$620	$0	$620
Commercial NVR Appliance	$500	$0	$500
Ring/Nest (4 cameras)	$800	$100	$1,300

The math is clear: Cloud cameras cost 3-4x more over five years while giving you less privacy, less control, and less functionality.

Frigate’s optional Frigate+ subscription (~$5-10/month) offers enhanced AI models but is entirely optional—all core features remain free.

Conclusion: Is Frigate Right for You?

Frigate NVR is ideal for:

• Homelab enthusiasts who already have hardware or enjoy DIY projects
• Privacy advocates who want complete control over their surveillance data
• Home Assistant users wanting seamless smart home integration
• Cost-conscious homeowners tired of monthly subscription fees
• Tech-savvy users comfortable with Docker and YAML configuration

It’s not ideal for:

• Users wanting a plug-and-play, zero-configuration experience
• Those without the technical inclination to manage a server
• People who need professional monitoring services

Getting Started

If you’re ready to take control of your home security:

1. Audit your existing cameras for RTSP support
2. Choose appropriate hardware based on your camera count
3. Follow the Docker installation guide above
4. Configure zones and masks to eliminate false alerts
5. Integrate with Home Assistant for automation magic

The world of self-hosted security cameras has never been more accessible. Frigate NVR delivers what cloud providers can’t: true ownership of your data, professional-grade AI detection, and freedom from monthly fees.

Your footage. Your network. Your control.

---

For more information, visit the official Frigate documentation or join the active community on Reddit and Discord.