[Config Support]: upgraded from 0.16 to 0.17; seemed "deaf"; used AI config assistance/got confused; downgraded to 0.16 #22447

ElectronicBattle · 2026-03-15T10:57:11Z

ElectronicBattle
Mar 15, 2026

Describe the problem you are having

Hello all

short version:
Please could someone look over my 0.16 config and highlight those areas which are simply wrong or won't work if I try to u/g to 0.17 again? My Frigate seemed to go a bit "deaf" on 0.17 and I would like to have another go at upgrading.

specific versions:
0.16.4-4131252
0.17.0-f0d69f7

background:
0.16 has worked perfectly from my point of view but I am keen to embrace some of the improvements in 0.17. The recent u/g to 0.17 was easy to perform and the config was automatically translated at the point of upgrading. However, I noticed that the number of detections seemed to have reduced and although object classification was still accurate there seemed to be fewer of them.

I used https://docs.frigate.video "Ask AI" to recommend some config changes, which took quite some effort from me to implement due to mistakes I think I made, and trying to load it back into Frigate caused lots of errors, but in the end it resulted in a new config.

However, I again noticed what I believe might be poorer detection and classification. For example, this is what a typical camera timeline history looks like on 0.16:

but the same camera yesterday morning looked like this on 0.17 (prior to today's downgrade):

Now, I realise that 0.17 just shows a blank when there's nothing happening to be recorded:

but shouldn't there still be motion detection "noise" due to sun/clouds/shadows, wind blowing leaves on trees etc., which are then ignored by the classification process? I am just seeing blank everything on all eight cameras.

The number of object classification incidents seems to have dropped overnight (but I acknowledge that it could be coincidental that various wild animals are just happening to stay during these last dozen or so days but that seems less likely than me having messed something up!).

I tried to use "ask AI" again but things seem to have changed now; unlike a couple of weeks ago, I can't get it to help, instead seeing

What I need to do, in that case, is to establish a baseline by verifying that my config should work and then I can try again. I am somewhat overwhelmed by the documentation and changes to 0.17 and I have never been particularly familiar with the config yaml anyway!

Can you help please?

Version

0.16.4-4131252

Frigate config file

#280525_0.15.1 TRUENAS low CPU burden NVidia P620 doorbell dewarp TS remap
version: 0.16-0

ui:
  time_format: 24hour
mqtt:
  enabled: true
  host: 192.168.1.194
  port: 1883
  user: '{FRIGATE_MQTT_USER}'
  password: '{FRIGATE_MQTT_PASSWORD}'


model:
  path: plus://74c5218b34c06b120149615ce8e30e3a #mobiledet edgeTPU 16/1/26 was 22e90a15e32d970bafcc50069ef525c0 #mobiledet edgeTPU 2025.2 16/4/25 was 78cd2d4014fa853509fa2ce285a8649b 17/4/25 was 40842fbab04a0be3b3c360e90cc1f766 2025.0 23/3/25 was 9ef83f381bbbdc16adccda5201f1ed55 30/10/25
genai:
  enabled: false
  provider: openai #this doesn't work - openAI needs payment
  api_key: '{FRIGATE_GENAI_API_KEY}'
  model: gpt-4o #gemini-2.0-flash-lite-preview-02-05 #gemini-2.0-flash

semantic_search:
  enabled: true
  reindex: false #true
  model_size: small # large = GPU small = CPU



objects:
  filters:
    dog:
      min_score: .8
      threshold: .9
    cat:
      min_score: .9
      threshold: .85
    fox:
      min_score: .75
      threshold: .8
    squirrel:
      min_score: .65
      threshold: .8
    bird:
      min_score: .65
      threshold: .8
    hedgehog:
      min_score: .65
      threshold: .8
    face:
      min_score: .7
    package:
      min_score: .65
      threshold: .9
    amazon:
      min_score: .75
    ups:
      min_score: .75
    fedex:
      min_score: .75
    person:
      min_score: .65
      threshold: .85
    car:
      min_score: .65
      threshold: .85
    license_plate:
      min_score: .6
    bicycle:
      min_score: .65
      threshold: .85


  track:
    - person
    - face
    - car
    - license_plate
    - cat
    - dog
    - fox
    - bird
    - deer
    - squirrel
    - hedgehog
    - package
    - bicycle


review:
  alerts:
    labels:
      - person
      - face

      - car
      - license_plate

      - cat
      - dog
      - fox
      - deer
      - hedgehog
      - bicycle

  detections:
    #enabled: true
    labels:
      - bird
      - squirrel


snapshots:
  enabled: true # Optional: Enable writing jpg snapshot to /media/frigate/clips
  clean_copy: true #false  #true # Optional: save a clean PNG copy of the snapshot image
  timestamp: false # Optional: print a timestamp on the snapshots
  bounding_box: true # Optional: draw bounding box on the snapshots
  crop: false # Optional: crop the snapshot
  height: 768 # Optional: height to resize the snapshot to (default: original size)  #height: 800
  #required_zones: []# Optional: Restrict snapshots to objects that entered any of the listed zones (default: no required zones)
  retain: # Optional: Camera override for retention settings (default: global values)
    default: 5 # Required: Default retention days 
    objects: # Optional: Per object retention days
      person: 7
      face: 7
      cat: 7
      dog: 7
      bird: 3
      squirrel: 3
      fox: 7
      badger: 7
      deer: 7
      hedgehog: 7
      car: 7
      licence_plate: 7
      bicycle: 7


  quality: 100 # Optional: quality of the encoded jpeg, 0-100 

go2rtc:
  log:
    exec: trace
  streams:
    #greenhouse:
    #  - ffmpeg:rtsp://192.168.1.123:554/media=0/channel=0&level=0/track0#video=h264#audio=copy
    #  - ffmpeg:greenhouse#audio=opus
    #upper_patio:
    #  - ffmpeg:rtsp://192.168.1.123:554/media=0/channel=1&level=0/track0#video=h264#audio=copy
    #   - ffmpeg:upper_patio#audio=opus
    #ponds:
    #  - ffmpeg:rtsp://192.168.1.123:554/media=0/channel=2&level=0/track0#video=h264#audio=copy
    #  - ffmpeg:ponds#audio=opus
    #blue_gate:
    #  - ffmpeg:rtsp://192.168.1.123:554/media=0/channel=3&level=0/track0#video=h264#audio=copy
    #  - ffmpeg:blue_gate#audio=opus
    #lower_steps:
    #  - ffmpeg:rtsp://192.168.1.123:554/media=0/channel=4&level=0/track0#video=h264#audio=copy
    #  - ffmpeg:lower_steps#audio=opus
    #lower_patio:
    #  - ffmpeg:rtsp://192.168.1.123:554/media=0/channel=5&level=0/track0#video=h264#audio=copy
    #  - ffmpeg:lower_patio#audio=opus
    #shed:
    #  - ffmpeg:rtsp://192.168.1.123:554/media=0/channel=6&level=0/track0#video=h264#audio=copy
    #  - ffmpeg:shed#audio=opus
    #drive:
    #  - ffmpeg:rtsp://192.168.1.123:554/media=0/channel=7&level=0/track0#video=h264#audio=copy
    #  - ffmpeg:drive#audio=opus

    #doorbell_sub: # 102 is SD  ... this is working witrhout the TS remap
    #  - ffmpeg:rtsp://{FRIGATE_RTSP_USER}:{FRIGATE_RTSP_PASSWORD}@192.168.1.192:554/Streaming/Channels/102#video=h264#audio=copy#raw=-vf "lenscorrection=k1=-0.6:k2=0.2" 
    #  - ffmpeg:doorbell_sub#audio=opus

    doorbell_main: # 101 is HD & 102 is SD  ... this is a test WITH the TS remap
      - ffmpeg:rtsp://{FRIGATE_RTSP_USER}:{FRIGATE_RTSP_PASSWORD}@192.168.1.192:554/Streaming/Channels/101#video=h264#audio=copy#raw=-vf
        "split=2[main][sub];[main]lenscorrection=k1=-0.6:k2=0.2[corrected_main];[sub]crop=150:25:1300:1300,lenscorrection=k1=-0.6:k2=0.2[cropped];[corrected_main][cropped]overlay=0:5"
      - ffmpeg:doorbell_sub#audio=opus

    doorbell_sub: # 101 is HD & 102 is SD  ... this is a test WITH the TS remap
      - ffmpeg:rtsp://{FRIGATE_RTSP_USER}:{FRIGATE_RTSP_PASSWORD}@192.168.1.192:554/Streaming/Channels/102#video=h264#audio=copy#raw=-vf
        "split=2[main][sub];[main]lenscorrection=k1=-0.6:k2=0.2[corrected_main];[sub]crop=150:25:1300:1300,lenscorrection=k1=-0.6:k2=0.2[cropped];[corrected_main][cropped]overlay=0:5"
      - ffmpeg:doorbell_sub#audio=opus


    #doorbell_sub2:
    #  - ffmpeg:rtsp://{FRIGATE_RTSP_USER}:{FRIGATE_RTSP_PASSWORD}@192.168.1.192:554/Streaming/Channels/102#video=h264#audio=copy#raw=-vf
    #    "split=2[main][sub];[main]lenscorrection=k1=-0.6:k2=0.2[corrected_main];[sub]crop=300:40:1450:1480,lenscorrection=k1=-0.6:k2=0.2[cropped];[corrected_main][cropped]overlay=5:0"
    #  - ffmpeg:doorbell#audio=opus

    #doorbell_main2:
     # - ffmpeg:rtsp://{FRIGATE_RTSP_USER}:{FRIGATE_RTSP_PASSWORD}@192.168.1.192:554/Streaming/Channels/101#video=h264#audio=copy#raw=-vf
     #   "split=2[main][sub];[main]lenscorrection=k1=-0.6:k2=0.2[corrected_main];[sub]crop=300:40:1450:1480,lenscorrection=k1=-0.6:k2=0.2[cropped];[corrected_main][cropped]overlay=5:0"
     # - ffmpeg:doorbell#audio=opus

####rtsp is 30060 and webRTC is 30061 (TN Frigate app internals)
  webrtc:
    candidates:
      - 192.168.1.194:30061
      - stun:30061


ffmpeg:
  #####you can only have one line here#####
  hwaccel_args: preset-nvidia
  #hwaccel_args: preset-nvidia-hevc #gives black screens on cameras
  #hwaccel_args: preset-nvidia-cuvid #gives black screens on cameras
  output_args:
    record: preset-record-generic-audio-aac

detectors:
  coral:
    type: edgetpu
    device: usb
    detect:
      enabled: true

birdseye:
  enabled: true
  mode: continuous #continuous | objects | motion
  inactivity_threshold: 5



# record: # <----- Enable recording # was working
#  enabled: true # was working

# added from https://fixtse.com/blog/coral-ia-frigate start#

# added from https://fixtse.com/blog/coral-ia-frigate #


record:
  enabled: true
  retain:
    days: 3
    mode: motion

motion:
  improve_contrast: true

detect:
  width: 640
  height: 360

#record:
#  width: 1920
#  height: 1080

########################################################################################################################################################################################################################################

  enabled: true
cameras: #########GREENHOUSE##################GREENHOUSE##################GREENHOUSE##################GREENHOUSE##################GREENHOUSE#########GREENHOUSE#########
  greenhouse:
    genai:
      enabled: false # <- enable GenAI for this camera

    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://192.168.1.123:554/media=0/channel=0&level=1/track0 #level1 = SD 640x360 (checked CORRECT); track0 media=0 are the only working options tested in ffplay
          roles:
            - detect



        - path: rtsp://192.168.1.123:554/media=0/channel=0&level=0/track0 #level0 = HD 1920x1080(checked CORRECT);  track0 media=0 are the only working options tested in ffplay
          roles:
            - record

    objects:
      track:
        - face
        - person
        - cat
        - dog
        - bird
        - deer
        - squirrel
        - fox
        - badger #placeholder

    birdseye:
      mode: continuous #continuous | objects | motion
    # masks are black: they block out areas where you aren't interested in detecting things
      order: 1





#########UPPER_PATIO###############################################UPPER_PATIO###########################UPPER_PATIO###########################UPPER_PATIO###########################
    motion:
      mask: 0.006,0.012,0.253,0.015,0.253,0.058,0.005,0.055
  upper_patio:
    genai:
      enabled: false # <- enable GenAI for this camera
      use_snapshot: true
      objects:
        - fox

    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://192.168.1.123:554/media=0/channel=1&level=1/track0 #SD 640x360 level1 ### 130125 trying HD for det (here) and rec below as usual
          roles:
            - detect
        - path: rtsp://192.168.1.123:554/media=0/channel=1&level=0/track0
          roles:
            - record

    objects:
      track:
        - face
        - person
        - cat
        - dog
        - bird
        - deer
        - squirrel
        - fox
        - badger #placeholder
        - hedgehog
    zones:
      upperpatio:
        coordinates: 
          0.003,0.997,0.005,0.795,0.042,0.789,0.045,0.732,0.049,0.619,0.077,0.546,0.103,0.479,0.145,0.425,0.168,0.401,0.204,0.371,0.216,0.369,0.268,0.378,0.279,0.339,0.32,0.302,0.347,0.293,0.361,0.298,0.382,0.345,0.395,0.375,0.563,0.418,0.582,0.474,0.626,0.476,0.674,0.51,0.712,0.515,0.795,0.664,0.862,0.704,0.933,0.822,0.999,0.896,0.995,0.996
        loitering_time: 0
    motion:
      mask: 
        0.003,0.374,0.207,0.313,0.239,0.272,0.292,0.258,0.32,0.245,0.36,0.222,0.682,0.219,0.716,0.204,0.726,0.001,0.002,0.001

    review:
      alerts:
        required_zones: upperpatio
      detections:
        required_zones: upperpatio



    birdseye:
      mode: continuous #continuous | objects | motion
    # masks are black: they block out areas where you aren't interested in detecting things
      order: 2

#########UPPER_PATIO###############################################UPPER_PATIO###########################UPPER_PATIO###########################UPPER_PATIO###########################

  ponds: #################PONDS#######################################PONDS#########################PONDS#########################PONDS#########################PONDS#########################
    genai:
      enabled: false # <- enable GenAI for this camera
    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://192.168.1.123:554/media=0/channel=2&level=1/track0 #level1 = SD 640x360 (checked CORRECT); track0 media=0 are the only working options tested in ffplay
          roles:
            - detect


        - path: rtsp://192.168.1.123:554/media=0/channel=2&level=0/track0 #level0 = HD 1920x1080(checked CORRECT);  track0 media=0 are the only working options tested in ffplay
          roles:
            - record

    objects:
      track:
        - face
        - person
        - cat
        - dog
        - bird
        - deer
        - squirrel
        - fox
        - badger #placeholder

    birdseye:
      mode: continuous  #continuous | objects | motion
     # masks are black: they block out areas where you aren't interested in detecting things
      order: 3
#################PONDS#######################################PONDS#########################PONDS#########################PONDS#########################PONDS#########################

    motion:
      mask: 0.002,0.003,0.24,0.003,0.241,0.049,0.003,0.053
  blue_gate: #########BLUE_GATE###################################################BLUE_GATE#############################BLUE_GATE#############################BLUE_GATE#############################BLUE_GATE#############################
    genai:
      enabled: true # <- enable GenAI for this camera
    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://192.168.1.123:554/media=0/channel=3&level=1/track0 #level1 = SD; track0 media=0 are the only working options tested in ffplay
        #- path: http://truenas.local:8082/media=0/channel=3&level=1/track0 # didn't work: two streams avalble from .123 not from motioneye
        #- path: http://truenas.local:8082 # didn't work: two streams avalble from .123 not from motioneye
      #  input_args: preset-rtsp-restream

          roles:
            - detect
        - path: rtsp://192.168.1.123:554/media=0/channel=3&level=0/track0 #level0 = HD ;  track0 media=0 are the only working options tested in ffplay
          roles:
            - record

    objects:
      track:
        - face
        - person
        - cat
        - dog
        - bird
        - deer
        - fox
        - badger # placeholder
        - car
        - license_plate
        - bicycle

    birdseye:
      mode: continuous   #continuous | objects | motion
      # masks are black: they block out areas where you aren't interested in detecting things
      order: 4

    motion:
      mask:
        - 0,153,85,148,140,139,239,140,300,142,354,145,398,141,429,141,486,145,518,213,450,226,605,360,640,360,640,0,0,0
        - 479,360,423,337,411,310,390,288,369,268,352,248,340,228,297,223,267,247,231,253,212,204,188,200,152,218,107,227,88,222,73,234,63,250,54,266,45,282,27,292,0,287,0,360

    zones:
      gravel:
        coordinates: 
          411,273,471,360,523,360,585,360,474,227,473,151,388,141,283,146,228,142,0,151,0,301,296,233
#########BLUE_GATE###################################################BLUE_GATE#############################BLUE_GATE#############################BLUE_GATE#############################BLUE_GATE#############################

  lower_steps: #########LOWER_STEPS#########LOWER_STEPS#######LOWER_STEPS#######LOWER_STEPS#######LOWER_STEPS#######LOWER_STEPS#######LOWER_STEPS#######LOWER_STEPS#######LOWER_STEPS#######
    genai:
      enabled: false # <- enable GenAI for this camera
    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://192.168.1.123:554/media=0/channel=4&level=1/track0 #level1 = SD; track0 media=0 are the only working options tested in ffplay
        #- path: http://truenas.local:8082/media=0/channel=5&level=1/track0 # didn't work: two streams avalble from .123 not from motioneye
        #- path: http://truenas.local:8082 # didn't work: two streams avalble from .123 not from motioneye
       #input_args: preset-rtsp-restream
          roles:
            - detect
        - path: rtsp://192.168.1.123:554/media=0/channel=4&level=0/track0 #level0 = HD ;  track0 media=0 are the only working options tested in ffplay
          roles:
            - record

    objects:
      track:
        - face
        - person
        - cat
        - dog
        - bird
        - deer
        - fox
        - squirrel
        - badger #placeholder
        - hedgehog

    birdseye:
      mode: continuous   #continuous | objects | motion
      order: 5
#########LOWER_STEPS#########LOWER_STEPS#######LOWER_STEPS#######LOWER_STEPS#######LOWER_STEPS#######LOWER_STEPS#######LOWER_STEPS#######LOWER_STEPS#######LOWER_STEPS#######

    motion:
      mask: 0.005,0.013,0.251,0.01,0.251,0.055,0.007,0.063
  lower_patio: #########LOWER_PATIO######LOWER_PATIO######LOWER_PATIO######LOWER_PATIO######LOWER_PATIO######LOWER_PATIO######LOWER_PATIO######LOWER_PATIO######LOWER_PATIO######LOWER_PATIO
    genai:
      enabled: false # <- enable GenAI for this camera
    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://192.168.1.123:554/media=0/channel=5&level=1/track0 #level1 = SD; track0 media=0 are the only working options tested in ffplay
        #- path: http://truenas.local:8082/media=0/channel=5&level=1/track0 # didn't work: two streams avalble from .123 not from motioneye
        #- path: http://truenas.local:8082 # didn't work: two streams avalble from .123 not from motioneye
       #input_args: preset-rtsp-restream
          roles:
            - detect
        - path: rtsp://192.168.1.123:554/media=0/channel=5&level=0/track0 #level0 = HD ;  track0 media=0 are the only working options tested in ffplay
          roles:
            - record

    objects:
      track:
        - face
        - person
        - cat
        - dog
        - bird
        - deer
        - fox
        - squirrel
        - badger #placeholder
        - hedgehog

    birdseye:
      mode: continuous   #continuous | objects | motion
      order: 6

      # masks are black: they block out areas where you aren't interested in detecting things
    motion:
      mask:
        - 0,148,92,104,184,55,253,0,269,154,272,305,318,125,354,107,345,56,406,60,417,131,504,240,570,360,640,360,640,0,0,0
        #- 0,219,72,169,140,139,146,79,218,73,291,72,362,133,430,143,450,121,488,120,523,98,640,95,640,0,0,0

    zones:
      lower patio:
        coordinates: 
          284,360,303,183,333,157,344,139,357,118,352,99,359,74,382,73,384,110,403,128,499,246,580,360
#########LOWER_PATIO######LOWER_PATIO######LOWER_PATIO######LOWER_PATIO######LOWER_PATIO######LOWER_PATIO######LOWER_PATIO######LOWER_PATIO######LOWER_PATIO######LOWER_PATIO



 #   detect:
 #    annotation_offset: -800


  shed: #########SHED##########################SHED#########################SHED#########################SHED#########################SHED#########################SHED#########################
    genai:
      enabled: false # <- enable GenAI for this camera
    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://192.168.1.123:554/media=0/channel=6&level=1/track0 #level1 = SD; track0 media=0 are the only working options tested in ffplay
        #- path: http://truenas.local:8082/media=0/channel=5&level=1/track0 # didn't work: two streams avalble from .123 not from motioneye
        #- path: http://truenas.local:8082 # didn't work: two streams avalble from .123 not from motioneye
       #input_args: preset-rtsp-restream
          roles:
            - detect

        - path: rtsp://192.168.1.123:554/media=0/channel=6&level=0/track0 #level0 = HD ;  track0 media=0 are the only working options tested in ffplay
          roles:
            - record

    objects:
      track:
        - face
        - person
        - cat
        - dog
        - bird
        - deer
        - fox
        - squirrel
        - badger #placeholder
        - hedgehog

    birdseye:
      mode: continuous   #continuous | objects | motion
      order: 7

#########SHED##########################SHED#########################SHED#########################SHED#########################SHED#########################SHED#########################



#########DRIVE##########################DRIVE#########################DRIVE#########################DRIVE#########################DRIVE#########################DRIVE#########################
    motion:
      mask: 0.008,0.014,0.252,0.014,0.25,0.057,0.008,0.054
  drive:
    genai:
      enabled: false # <- enable GenAI for this camera

    enabled: true # this enables or disables the "drive" camera
    ffmpeg:
      inputs:
        - path: rtsp://192.168.1.123:554/media=0/channel=7&level=1/track0 #level1 = SD; track0 media=0 are the only working options tested in ffplay
          roles:
            - detect
        - path: rtsp://192.168.1.123:554/media=0/channel=7&level=0/track0 #level0 = HD ;  track0 media=0 are the only working options tested in ffplay
          roles:
            - record

    objects:
      track:
        - face
        - person
        - car
        - bicycle
        - cat
        - dog
        - bird
        - deer
        - fox
        - squirrel
        - badger #placeholder


    zones:
      drive_gravel:
        coordinates: 
          0.256,0.37,0.355,0.375,0.344,0.548,0.693,0.536,0.991,0.866,0.999,0.991,0.207,0.995,0.2,0.595,0.082,0.522
        loitering_time: 0
    motion:
      mask: 
        0.002,0.36,0.179,0.304,0.27,0.269,0.276,0.12,0.361,0.137,0.358,0.349,0.487,0.351,0.658,0.356,0.71,0.358,0.998,0.539,0.993,0.003,0.008,0

    review:
      alerts:
        required_zones: drive_gravel
      detections:
        required_zones: drive_gravel

    birdseye:
      mode: continuous #continuous | objects | motion
    # masks are black: they block out areas where you aren't interested in detecting things
      order: 8

#########DRIVE##########################DRIVE#########################DRIVE#########################DRIVE#########################DRIVE#########################DRIVE#########################  



###############doorbell############################################doorbell############################################doorbell############################################doorbell#############################
###############doorbell############################################doorbell############################################doorbell############################################doorbell#############################
###############doorbell############################################doorbell############################################doorbell############################################doorbell#############################
  doorbell:
    genai:
      enabled: false # <- enable GenAI for this camera true false
    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/doorbell_sub
          roles:
            - detect
        - path: rtsp://127.0.0.1:8554/doorbell_main
          roles:
            - record

    objects:
      track:
        - face
        - person
        - cat
        - dog
        - bird
        - deer
        - fox
        - squirrel
        - badger #placeholder
        - hedgehog

    birdseye:
      mode: continuous   #continuous | objects | motion
      order: 9

#timestamp_style:
#      format: "%d/%m/%Y %H:%M:%S"
#      position: "tr"
#      color:
#        red: 0 #255
#        green: 0 #255
#        blue: 0 #255
#      thickness: 1
#      effect: solid 

################################################################################################################################################################################
################################################################################################################################################################################
################################################################################################################################################################################


 #   detect:
 #     annotation_offset: -1000

logger:
  logs:
      #frigate.record.maintainer: debug
    frigate.object_detection: debug
      #watchdog.blue_gate: debug

camera_groups:
  Woodlands2:
    order: 1
    icon: LuActivitySquare
    cameras:
      - drive
      - upper_patio
      - doorbell
  birdseye:
    order: 2
    icon: LuAmpersands
    cameras: birdseye
  Sitting_room:
    order: 3
    icon: LuAlertCircle
    cameras: doorbell
face_recognition:
  enabled: false
  model_size: small
lpr:
  enabled: true
classification:
  bird:
    enabled: true

Relevant Frigate log output

n/a

Relevant go2rtc log output

n/a

Frigate stats

n/a

Operating system

Other Linux

Install method

Docker Compose

docker-compose file or Docker CLI command

docker compose via TrueNAS app

Object Detector

Coral

Screenshots of the Frigate UI's System metrics pages

n/a

Any other information that may be helpful

n/a

Answered by hawkeye217

Mar 15, 2026

The correct syntax would be:

detectors:
  coral:
    type: edgetpu
    device: usb

detect:
  enabled: true
  width: 640
  height: 360

However, this would apply a detect width and height of 640x360 to all of your cameras, as this is defining a detect section at the global level of your config. I'm not sure that's what you want.

You can specify the detect width and height at the camera level, eg:

detectors:
  coral:
    type: edgetpu
    device: usb

detect:
  enabled: true
  
cameras:
  camera_one:
    ffmpeg:
      ...
    detect:
      width: 640
      height: 360
    ...

View full answer

2026-03-15T10:58:15Z

inkeep-ai-assistant[bot]
bot Mar 15, 2026

I can review your 0.16 config and highlight areas that may cause issues or are potentially misconfigured based on the available documentation.

Issues Identified in Your Config

1. Zone name with spaces
Your lower_patio camera has a zone named lower patio (with a space). Zone names should not contain spaces as this can cause parsing issues. (1)

2. Duplicate enabled: true under record
You have:

record:
  enabled: true
  retain:
    days: 3
    mode: motion
# ... other settings ...
  enabled: true  # <-- This appears again before cameras section

(1)

This duplicate key could cause unexpected behavior.

3. Motion mask placement
Some of your motion masks are placed outside their camera blocks. For example, the greenhouse camera's motion mask appears to be defined after the camera block closes, just before upper_patio. Motion masks should be nested under the specific camera's motion: section. (2)

4. Object filter masks vs motion masks
Motion masks prevent motion from triggering detection, but they don't prevent objects from being detected if detection was triggered by motion in unmasked areas. If you're seeing fewer detections, verify your masks aren't over-masking areas where objects should be detected. (2)

5. Required zones configuration
For cameras like drive and upper_patio, you have required_zones set for alerts and detections. Objects must enter these zones to trigger alerts/detections. If zones are too restrictive, you'll see fewer events. (1)

Recommendations

Fix the zone name lower patio → lower_patio
Remove the duplicate enabled: true under record
Ensure each camera's motion mask is properly nested within that camera's configuration block
Review your motion masks using the Debug view with Motion Boxes enabled to verify they're not over-masking (2)

The blank timeline periods you're seeing in 0.17 when there's no activity is expected behavior - the timeline shows blank when no recordings are available. However, if you're seeing significantly fewer detections overall, the configuration issues above (especially the mask placement and zone restrictions) could be contributing factors.

📚 Sources:

Have another question? Just tag @inkeep.

0 replies

hawkeye217 · 2026-03-15T12:02:05Z

hawkeye217
Mar 15, 2026
Collaborator

Your config is slightly incorrect.

detectors:
  coral:
    type: edgetpu
    device: usb
    detect:
      enabled: true

should be:

detectors:
  coral:
    type: edgetpu
    device: usb

detect:
  enabled: true

This would be one reason why 0.17 was "deaf" - detection is disabled by default, and nesting detect underneath detectors is technically an invalid configuration that would not throw an error.

6 replies

ElectronicBattle Mar 15, 2026
Author

ahh ... I think I see. Thanks Josh !

(a) should I correct these mistakes in my current 0.16 config and run it in 0.16, or will it cause me errors due to it being for 0.17 only?

Naturally I can easily upgrade to 0.17 in order to experiment but I thought I'd check first before I confuse myself still further.

(b) is it the case that there's some form of (python?) yaml interpreter which parses the config to generate inputs to other python modules, and does that hypothetical yaml interpreter also get up-issued between versions so that its tolerance for errors changes and improves?

hawkeye217 Mar 15, 2026
Collaborator

You can make the change I suggested above and it will work just fine for both 0.16 and 0.17.

And yes, Frigate parses your yaml and uses Pydantic to validate the configuration. We often include "migrator" code that makes config changes automatically when you upgrade based on the new version's Pydantic config schema. This migrator code is keyed off the version in your config and runs before validation on the initial upgrade to make the changes so that the new config is valid and Frigate starts successfully.

ElectronicBattle Mar 15, 2026
Author

clever stuff! Unlike me.

I notice another probable error in my config whereby this detect a few lines down gets highlighted in red when I implement those detector corrections:

detect:
  width: 640
  height: 360

I think it is an "orphaned" detect statement and I think it needs to be combined into your corrected version like this:

detectors:
  coral:
    type: edgetpu
    device: usb

detect:
  enabled: true
      width: 640
      height: 360

I know that doing so gets rid of the red highlight so perhaps I should simply go ahead and try it.

hawkeye217 Mar 15, 2026
Collaborator

The correct syntax would be:

detectors:
  coral:
    type: edgetpu
    device: usb

detect:
  enabled: true
  width: 640
  height: 360

However, this would apply a detect width and height of 640x360 to all of your cameras, as this is defining a detect section at the global level of your config. I'm not sure that's what you want.

You can specify the detect width and height at the camera level, eg:

detectors:
  coral:
    type: edgetpu
    device: usb

detect:
  enabled: true
  
cameras:
  camera_one:
    ffmpeg:
      ...
    detect:
      width: 640
      height: 360
    ...

Answer selected by ElectronicBattle

ElectronicBattle Mar 15, 2026
Author

Yes - I see it now. Thanks a million!

I made a typo above; I did manage to get the indentation in the actual config yaml correct:

detect:
  enabled: true
  width: 640
  height: 360

At the moment I think all eight cameras work well enough by using the global low res 640x360 detection but the doorbell camera might be better adjusted; something to experiment with. I see how to do it.

I have improved my config and restarted and all is well, so I might now re-upgrade once again to 0.17 and see how I get on. Swapping about is easy and now that I have an improved yaml, I can probably make good progress without convincing myself that something's wrong (which might well not be the case at all).

Great help yet again - thanks Josh.

hawkeye217 Mar 15, 2026
Collaborator

Great! Glad it's working now, happy to help as always.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[Config Support]: upgraded from 0.16 to 0.17; seemed "deaf"; used AI config assistance/got confused; downgraded to 0.16 #22447

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{editor}}'s edit

{{editor}}'s edit

Uh oh!

Replies: 2 comments 6 replies

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Uh oh!

{{title}}

Uh oh!

Select a reply

Uh oh!

Uh oh!

[Config Support]: upgraded from 0.16 to 0.17; seemed "deaf"; used AI config assistance/got confused; downgraded to 0.16 #22447

Uh oh!

Uh oh!

ElectronicBattle Mar 15, 2026

Describe the problem you are having

Version

Frigate config file

Relevant Frigate log output

Relevant go2rtc log output

Frigate stats

Operating system

Install method

docker-compose file or Docker CLI command

Object Detector

Screenshots of the Frigate UI's System metrics pages

Any other information that may be helpful

Replies: 2 comments · 6 replies

Uh oh!

inkeep-ai-assistant[bot] bot Mar 15, 2026

Issues Identified in Your Config

Recommendations

Uh oh!

hawkeye217 Mar 15, 2026 Collaborator

Uh oh!

ElectronicBattle Mar 15, 2026 Author

Uh oh!

hawkeye217 Mar 15, 2026 Collaborator

Uh oh!

ElectronicBattle Mar 15, 2026 Author

Uh oh!

hawkeye217 Mar 15, 2026 Collaborator

Uh oh!

ElectronicBattle Mar 15, 2026 Author

Uh oh!

hawkeye217 Mar 15, 2026 Collaborator

ElectronicBattle
Mar 15, 2026

Replies: 2 comments 6 replies

inkeep-ai-assistant[bot]
bot Mar 15, 2026

hawkeye217
Mar 15, 2026
Collaborator

ElectronicBattle Mar 15, 2026
Author

hawkeye217 Mar 15, 2026
Collaborator

ElectronicBattle Mar 15, 2026
Author

hawkeye217 Mar 15, 2026
Collaborator

ElectronicBattle Mar 15, 2026
Author

hawkeye217 Mar 15, 2026
Collaborator