Home Assistant seems to "lag" since adding Google Coral

Hey all, I’m running 2022.8.6 on Home Assistant Blue (kinda new to HA and not that tech-savvy). I recently got frigate up and running, which was a miracle for me. I have a Google Coral occupying one of my four USB ports - a ZigBee USB stick occupies another and the other two are unused. Since integrating Frigate and the Coral USB…Home Assistant seems to lag (probably not the right word) and miss automations/deadlines for automations. Sometimes they’re delayed and sometimes they don’t run at all. My dashboard is not very responsive when turning on and off lights, or the state is incorrect on the dashboard for some things. I can’t really explain why this is going on and don’t really know much about the logs or how to read them. Apologies for getting in over my head. Any help would be so appreciated!

That wouldn’t make much sense as the coral improves performance. Did you have frigate running before the coral?

In any case, overall resource usage of the system (CPU, RAM, etc.) would be helpful to know

I did have Frigate running on a previous Home Assistant Blue (not this exact hardware, same specs though). I didn’t have the Coral at that time.

The Frigate add-on is running about 88% Add-On CPU Usage and 15% Add-On Ram Usage. Hope this is what you were asking for - sorry if not!

Not sure how to add my frigate.yml file - hope this looks…normal?

mqtt:
  host: <ip address of HA instance is here>
  port: 1883
  topic_prefix: frigate
  client_id: frigate
  user: <my user is here>
  password: <my password is here>
  stats_interval: 300

record:
  expire_interval: 7

cameras:
  doorbell:
    ffmpeg:
      inputs:
        - path: <the doorbell camera path is here>
          roles:
            - detect
            - record
    detect:
      width: 1600
      height: 1200
      fps: 3
    objects:
      track:
        - person
        - dog
        - bicycle
        - cat
        - car
      filters:
        person:
          min_score: 0.50
          threshold: 0.70
        dog:
          min_score: 0.50
          threshold: 0.60
        bicycle:
          min_score: 0.50
          threshold: 0.75
        cat:
          min_score: 0.50
          threshold: 0.60
        car:
          min_score: 0.70
          threshold: 0.90
    snapshots:
      enabled: True
      timestamp: False
      bounding_box: False
      crop: True
      height: 500
      retain:
        default: 7
    record:
      enabled: True
      retain:
        days: 0
      events:
        retain:
          default: 7
          mode: motion
        pre_capture: 15
        post_capture: 15

  garage:
    ffmpeg:
      inputs:
        - path: <my garage camera path is here>
          roles:
            - record
            - detect
    detect:
      width: 1920
      height: 1080
      fps: 3
      stationary:
        interval: 50
        threshold: 100
    objects:
      track:
        - person
        - dog
        - bicycle
        - cat
        - car
      filters:
        person:
          min_score: 0.50
          threshold: 0.70
        dog:
          min_score: 0.50
          threshold: 0.60
        bicycle:
          min_score: 0.50
          threshold: 0.75
        cat:
          min_score: 0.50
          threshold: 0.60
        car:
          min_score: 0.80
          threshold: 0.90
    snapshots:
      enabled: True
      timestamp: False
      bounding_box: False
      crop: True
      height: 500
      retain:
        default: 7
    record:
      enabled: True
      retain:
        days: 0
      events:
        retain:
          default: 7
          mode: active_objects
        pre_capture: 15
        post_capture: 15

  patio:
    ffmpeg:
      inputs:
        - path: <my patio camera path is here>
          roles:
            - record
            - detect
    detect:
      width: 1920
      height: 1080
      fps: 3
      stationary:
        interval: 50
        threshold: 100
    objects:
      track:
        - person
        - dog
        - cat
      filters:
        person:
          min_score: 0.60
          threshold: 0.75
        dog:
          min_score: 0.50
          threshold: 0.65
        cat:
          min_score: 0.75
          threshold: 0.90
    snapshots:
      enabled: True
      timestamp: False
      bounding_box: False
      crop: True
      height: 500
      retain:
        default: 7
    record:
      enabled: True
      retain:
        days: 0
      events:
        retain:
          default: 7
          mode: active_objects
        pre_capture: 15
        post_capture: 15

  sirius_crate:
    ffmpeg:
      inputs:
        - path: <my pup's camera path is here>
          roles:
            - record
            - detect
    detect:
      width: 1920
      height: 1080
      fps: 3
      stationary:
        interval: 50
        threshold: 100
    objects:
      track:
        - person
        - dog
        - cat
      filters:
        person:
          min_score: 0.75
          threshold: 0.90
        dog:
          min_score: 0.50
          threshold: 0.65
        cat:
          min_score: 0.75
          threshold: 0.90
    snapshots:
      enabled: True
      timestamp: False
      bounding_box: False
      crop: True
      height: 500
      retain:
        default: 7
    record:
      enabled: True
      retain:
        days: 0
      events:
        retain:
          default: 7
          mode: active_objects
        pre_capture: 15
        post_capture: 15

detectors:
  coral:
    type: edgetpu
    device: usb

rtmp:
  enabled: false

birdseye:
  enabled: True
  width: 1280
  height: 720
  quality: 8
  mode: objects

live:
  height: 640
  quality: 8

I have stopped the Frigate add-on, but the performance issues remain. Is there a chance that the Coral is causing issues just by being plugged in via USB? It seems to improve when I unplug the Coral entirely.

Thanks for your help so far!

1 Like