Frigate vs Blue Iris: Which NVR Should You Use?
Quick Verdict
Frigate is the better choice if you run Docker on Linux and want AI-powered object detection. Blue Iris wins if you’re on Windows and need a polished, mature NVR with decades of development behind it. For most self-hosters running Linux servers, Frigate is the clear pick.
Overview
Frigate is an open-source NVR built specifically around real-time AI object detection. It runs in Docker, integrates tightly with Home Assistant, and uses Google Coral TPUs or OpenVINO for hardware-accelerated detection. It’s the most popular self-hosted NVR in the homelab community.
Blue Iris is a commercial Windows-only NVR software ($70 one-time license). It’s been around since 2007 and supports virtually every IP camera on the market. It has its own AI detection via DeepStack/CodeProject.AI integration, but it’s fundamentally a Windows application — no Docker, no Linux.
Feature Comparison
| Feature | Frigate | Blue Iris |
|---|---|---|
| Platform | Linux (Docker) | Windows only |
| License | MIT (free, open-source) | $69.95 one-time |
| AI Object Detection | Built-in (Coral TPU, OpenVINO, CPU) | Via DeepStack/CodeProject.AI plugin |
| Home Assistant Integration | Native, first-class | Via MQTT/API (third-party) |
| Camera Protocol Support | RTSP, RTMP, HTTP | RTSP, ONVIF, HTTP, MJPEG, 700+ models |
| Mobile App | Web UI (PWA) | Dedicated iOS/Android app ($5) |
| Recording Modes | Continuous, detect, motion | Continuous, triggered, scheduled |
| Sub-stream Support | Yes (detect on sub, record on main) | Yes |
| Multi-server | No (single instance) | No (single instance) |
| Hardware Acceleration | Coral TPU, Intel QSV, VAAPI | Intel QSV, NVIDIA NVENC, AMD AMF |
| Max Cameras | Limited by hardware | 64 officially |
| Active Development | Very active (monthly releases) | Active (regular updates) |
Installation Complexity
Frigate runs as a single Docker container. The configuration is a YAML file where you define cameras, detection zones, and recording settings. If you already run Docker, you can be up in 15 minutes. The main complexity is configuring the AI detection hardware (Coral USB or OpenVINO).
# Minimal Frigate config.yml
mqtt:
host: your-mqtt-broker
cameras:
front_door:
ffmpeg:
inputs:
- path: rtsp://user:pass@camera-ip:554/stream1
roles: [detect, record]
detect:
width: 1280
height: 720Blue Iris requires a Windows machine (or VM), a one-time license purchase, and installation via a standard Windows installer. The GUI is comprehensive but has a steep learning curve — it’s packed with options accumulated over 17+ years of development. Camera setup is point-and-click with ONVIF auto-discovery.
Frigate wins on simplicity if you’re already in the Docker ecosystem. Blue Iris wins if you prefer GUI-based configuration.
Performance and Resource Usage
| Metric | Frigate | Blue Iris |
|---|---|---|
| RAM (10 cameras) | 1-2 GB | 2-4 GB |
| CPU (10 cameras, no HW accel) | High (AI detection is CPU-heavy) | Moderate |
| CPU (with HW acceleration) | Very low (Coral handles detection) | Low (QSV handles encoding) |
| Recommended Hardware | Coral TPU ($30) + any Linux box | Dedicated Windows PC, i5+ |
| Idle Power | 5-15W (on mini PC) | 30-60W (typical Windows PC) |
Frigate with a Coral TPU is remarkably efficient — the TPU handles all AI inference at ~100ms per frame, leaving the CPU almost idle. Without a Coral, Frigate’s CPU usage is significantly higher.
Blue Iris is heavier on RAM and benefits from a dedicated Windows machine. It handles transcoding and recording well with Intel QSV but lacks the purpose-built AI acceleration path that Frigate has with Coral.
Community and Support
Frigate has 20,000+ GitHub stars, an active Discord server, and strong integration with the Home Assistant community. Documentation is solid and improving with every release. Issues get addressed quickly. The project is maintained by Blake Blackshear with a growing contributor base.
Blue Iris has a large, long-standing user forum (ipcamtalk.com), years of YouTube tutorials, and a mature knowledge base. Commercial support is available via email. The developer (Ken Tobin) has maintained the product solo for nearly two decades.
Both have excellent community support, but through different channels. Frigate leans on GitHub and Discord; Blue Iris leans on forums and YouTube.
Use Cases
Choose Frigate If…
- You run Linux and Docker
- You want AI object detection (person, car, animal) with minimal CPU usage
- You use Home Assistant and want tight integration
- You prefer open-source and free software
- You want to run on low-power hardware (mini PC, Raspberry Pi 5)
- You value automation (MQTT events → Home Assistant automations)
Choose Blue Iris If…
- You run Windows and prefer a GUI-based NVR
- You need support for obscure or older camera models (ONVIF auto-discovery)
- You want a mature, battle-tested NVR with 17+ years of development
- You need a dedicated mobile app with push notifications
- You’re managing a larger camera deployment (20+ cameras)
- You don’t use Docker or Home Assistant
Final Verdict
For the typical self-hoster running Docker on Linux, Frigate is the better choice. Its AI detection with Coral TPU support is unmatched in the open-source NVR space, Home Assistant integration is first-class, and it runs efficiently on modest hardware.
Blue Iris is an excellent NVR, but it’s locked to Windows. If you’re already committed to a Windows server or need the broadest possible camera compatibility, it’s worth the $70 license. But requiring a dedicated Windows machine is a significant limitation for most self-hosting setups.
The real deciding factor is your platform: Linux/Docker → Frigate. Windows → Blue Iris. If you have the choice, go with Frigate.
Frequently Asked Questions
Can Frigate and Blue Iris run on the same network?
Yes. They’re independent NVR systems. You can even point them at the same cameras (via RTSP sub-streams) if you want to test both. Just ensure your cameras support multiple RTSP connections.
Does Blue Iris work in a Windows VM?
Yes, but with caveats. GPU passthrough is needed for hardware acceleration (QSV/NVENC), and USB passthrough is needed if you want to use a Coral TPU with a DeepStack alternative. Many users run Blue Iris in a VM successfully, but a bare-metal Windows install performs better.
Is Frigate’s AI detection accurate enough to replace Blue Iris alerts?
Yes. Frigate’s built-in detection models (based on SSD MobileNet) reliably detect people, cars, dogs, cats, and other objects. False positive rates are low, especially with properly configured detection zones. With a Coral TPU, detection runs at ~100ms per frame — fast enough for real-time alerting.
What about Frigate+ (paid)?
Frigate+ is an optional paid service ($50/year) that provides a custom AI model trained on your specific cameras. It improves detection accuracy for your environment but isn’t required — the free built-in model works well for most setups.
Related
Get self-hosting tips in your inbox
New guides, comparisons, and setup tutorials — delivered weekly. No spam.