Plug-and-Play AI: Hailo, Frigate, and Home Assistant Integration
Here’s everything you need to know about the Hailo integration into Frigate NVR and Home Assistant (HA):
The What
This integration combines the capabilities of the Frigate video management system, the HA home automation system and the Hailo AI acceleration into a robust, industry-grade, AI-powered home security system.
The Why
- Cost: Running on the Raspberry Pi 5 with a Raspberry Pi AI HAT+ this is a <$200 solution. The November 2025 pricing is:
- Hailo-8L Kit (13 TOPS): RPi 5 ($80) + Hailo-8L HAT ($70) = $150
- Hailo-8 Kit (26 TOPS): RPi 5 ($80) + Hailo-8 HAT ($110) = $190
- Performance: The Hailo chip offers massive acceleration capacity, achieving ultra-low inference latency of just 7 ms using high-throughput models like YOLOv6n, and up to 580 FPS (Hailo-8) and 350 FPS (Hailo-8L), while also supporting stronger, more accurate models. The Hailo processor provides critical CPU offloading and with many TOPS to spare, allows additional AI applications on the same setup.
- Reliability: The integration with both Frigate and HA is available in their official websites. Combined with our long-term partnership with Raspberry pi, this guarantees stable releases and long-term support
- Privacy: all AI processing is executed locally on the host device, without sending any data to the cloud.
The Wow:
With the help of our partners at Frigate and Home Assistant, we have made this super easy to use:
- No Code: Installation is incredibly simple. By using the Frigate Add-on for Home Assistant, the entire application setup is No Code. The HA operating system (HAOS) automatically handles the driver installation, making deployment instantaneous and accessible to everyone.
- Optimal Integration: Hailo allows Frigate to leverage advanced computer vision models. This high-speed processing, combined with the low-latency MQTT bridge, enables Home Assistant to execute instantaneous and accurate home automations.
The Guide
This guide provides a template for using Hailo with Frigate running on Home Assistant, and covers the full installation process and required setup of Frigate on Home Assistant.
Prerequisites:
-
Home Assistant running on Raspberry Pi (version 2025.5.3 or higher)
-
Pi AI HAT+
-
Basic familiarity with how to navigate within home assistant, eg., adding integrations & add-ons.
Setup steps - all inside Home Assistant:
-
Install the MQTT broker “Add-on”.
Enable Watchdog and “Start”.
- Add MQTT “Integration”: First option (MQTT) and then again the first option (“Use the official…”).
- Create MQTT “User”: Settings → People → Add person.
- In the “Add-on”'s page select “Add-on store” and then press the 3 dots and select “Repositories”.
Paste the following repository link: GitHub - blakeblackshear/frigate-hass-addons: Frigate hass.io addons
- Restart Home assistant: Developer tools → Check and restart → press “Check configuration” and then press “Restart”.
- Now the Frigate “Add-on” is available - select the “Full Access” option.
- Enable “Watchdog” and disable “Protection mode” and then “Start”.
- Go to Frigate and in Settings select “Configuration Editor”.
- Example configuration - Edit based on your details the lines with TODO:
# Full Frigate documentation: https://docs.frigate.video/configuration/reference/
mqtt:
enabled: true
host: URL path # TODO
user: name # TODO
password: pass # TODO
cameras:
Camera_1:
enabled: true
ffmpeg:
inputs:
- path: URL path # TODO
roles:
- detect
- record
# Below are global settings - will aplly to all cameras
detectors:
hailo8l: # Always 8l, even in case of 8
type: hailo8l # Always 8l, even in case of 8
device: PCIe
# Default model
model:
width: 640
height: 640
input_tensor: nhwc
input_pixel_format: rgb
input_dtype: int
model_type: yolo-generic # Default YOLOv6n
# Example with Custom YOLO Model (HEF compiled for Hailo8) with URL (e.g. YOLOv8m)
# model:
# width: 640
# height: 640
# input_tensor: nhwc
# input_pixel_format: rgb
# input_dtype: int
# model_type: yolo-generic
# path: https://hailo-model-zoo.s3.eu-west-2.amazonaws.com/ModelZoo/Compiled/v2.17.0/hailo8/yolov8m.hef
detect:
enabled: true
width: 1280
height: 720
fps: 5
objects:
track: # List of objects to track from model's labelmap. Since model is Yolo, then: https://docs.ultralytics.com/datasets/detect/coco/#dataset-yaml
- person
filters: # Optional: filters to reduce false positives for specific object types
person:
min_area: 5000
max_area: 100000
min_score: 0.5
threshold: 0.7
record: # Optional: Enable recording
enabled: true
detections:
pre_capture: 5
post_capture: 5
retain:
days: 1
mode: active_objects
snapshots: # Optional: Configuration for the jpg snapshots written to the clips directory for each tracked object
enabled: true
timestamp: true
bounding_box: true
crop: False
retain:
default: 1
objects:
person: 1
logger:
default: info
logs:
frigate.object_detection: debug
frigate.motion: debug
detector.hailo8l: debug
version: 0.16-0
Models are downloaded to: ~/addon_configs/ccab4aaf_frigate-fa/model_cache/
- Save and restart frigate:
NOTE:
- If Frigate output errors about the Hailo model, you should try power cycling the Pi. Just restarting home assistant is not enough as the HAT is connected using PCIe and if driver is updated it need to update its FW.
- Detections can be seen in real time (live) only via debug mode. Select the camera - and enter debug mode (via settings on the right).
- Allow bounding boxes - then inspect detection in real time.
- Detection bounding boxes can also be seen on recordings & snapshots: Navigate to “Explore”.
Select event, and then explore the Snapshot and the object life-cycle.
Explore the performance: Frigate → Settings → System metrics:
This is for a system with 4 cameras:
