Frigate+ LPR does not create license_plate boxes, but standalone LPR detector works

[cho0p]

Describe the problem you are having

Hi,

I am testing LPR with Frigate 0.17.1 using a newly trained Frigate+ model.

Setup:

Intel i7-14700 CPU
Frigate version: 0.17.1-416a9b7
Frigate+
Model: yolov9s-relu6
Detector: 2x EdgeTPU
LPR enabled

Test clip and live camera both use a clearly readable license plate

Minimal LPR config:

lpr:
enabled: true
device: CPU (tried with and without)
debug_save_plates: true

When I use Frigate+ native license plate detection:

objects:
track:
- person
- dog
- car
- motorcycle
- license_plate

LPR does not work. In the debug view I get a car box, but no usable license_plate box on the actual plate. Logs show messages like:

No attributes to parse.
Not a processing license plate for non car/motorcycle object.
Skipping LPR for non-stationary car object ... Detected in 3 concurrent frames, threshold to run is 4

When I remove license_plate from objects.track, the same test clip works correctly:

objects:
track:
- person
- dog
- car
- motorcycle
# - license_plate

Then Frigate uses the separate YOLOv9 LPR detector and the plate is detected/read correctly, e.g. 0220 JVG with high confidence.

So the standalone YOLOv9 LPR detector works, but the Frigate+ native license_plate path does not produce usable plate attributes/boxes for the same scene.

Is this expected behavior in 0.17.1, or is there a known issue with Frigate+ license_plate attributes?

Version

0.17.1-416a9b7

Frigate config file

ui:
  timezone: Europe/Berlin
mqtt:
  host: 192.168.8.x
  user: xxx
  password: xxx
  topic_prefix: frigate
auth:
  failed_login_rate_limit: 1/second;5/minute;20/hour
  trusted_proxies:
    - xxx
logger:
  default: info
  logs:
    frigate.data_processing.common.license_plate: debug
lpr:
  enabled: true
  debug_save_plates: true

cameras:

  test:
    ffmpeg:
      hwaccel_args: []
      input_args: -re -stream_loop -1 -fflags +genpts
      inputs:
        - path: /media/frigate/car-stopping-h264.mp4
          roles:
            - detect
    detect:
      enabled: true
    record:
      enabled: false
    snapshots:
      enabled: false

  Tor:
    ffmpeg:
      hwaccel_args: preset-intel-qsv-h265
      inputs:
        - path: rtsp://192.168.8.2:8554/Tor/main
          input_args: preset-rtsp-restream
          roles:
            - detect
    detect:
      width: 3840
      height: 2160
      stationary:
        interval: 50
        threshold: 200
    objects:
      filters:
        car:
          mask: 
            0,0,1,0,1,0.89,0.586,0.762,0.242,0.602,0.25,0.463,0.154,0.48,0.136,0.511,0.062,0.563,0.044,0.652,0.052,1,0,1
        person:
          mask: 0.492,0.367,0.519,0.505,0.469,0.624,0.337,0.618,0.327,0.428
    motion:
      mask: 
        0.726,0,0.667,0.207,0.48,0.334,0.368,0.414,0.281,0.417,0.205,0.407,0.201,0.379,0.158,0.384,0.15,0.485,0.057,0.571,0,0.734,0,0
    ui:
      order: 1
  Strasse:
    ffmpeg:
      inputs:
        - path: rtsp://192.168.8.2:8554/Strasse/main
          input_args: preset-rtsp-restream
          roles:
            - detect
    detect:
      width: 2560
      height: 1920
      stationary:
        interval: 50
        threshold: 200
    motion:
      mask: 
        1,0,1,0.652,0.801,0.676,0.757,0.516,0.669,0.518,0.621,0.597,0.534,0.639,0.422,0.65,0.324,0.611,0,0.64,0,0
    objects:
      filters:
        car: {}
    ui:
      order: 2
  Auffahrt:
    ffmpeg:
      hwaccel_args: preset-intel-qsv-h265
      inputs:
        - path: rtsp://192.168.8.2:8554/Auffahrt/main
          input_args: preset-rtsp-restream
          roles:
            - detect
    detect:
      width: 3840
      height: 2160
      stationary:
        interval: 50
        threshold: 200
    objects:
      filters:
        car:
          min_score: 0.85
          mask: 1,0,0,0,0,0.581,0.472,0.485,0.812,0.748,1,0.752,1,0.345
        person:
          mask: 0,0,1,0,1,0.325,0.515,0.414,0.48,0.484,0,0.584
    motion:
      mask: 
        0,0,1,0,1,0.329,0.806,0.418,0.629,0.435,0.489,0.437,0.489,0.622,0.164,0.683,0.109,0.795,0.047,0.797,0,0.84
    ui:
      order: 3
  Pool:
    ffmpeg:
      hwaccel_args: preset-intel-qsv-h265
      inputs:
        - path: rtsp://192.168.8.2:8554/Pool/main
          input_args: preset-rtsp-restream
          roles:
            - detect
    ui:
      order: 4
    detect:
      width: 3840
      height: 2160
      stationary:
        interval: 50
        threshold: 200
    motion:
      mask:
        - 1,0,1,0.347,0.884,0.334,0.889,0.279,0.66,0.152,0.541,0.227,0.273,0.175,0.198,0.243,0.144,0.483,0.074,0.498,0.07,0.65,0,0.673,0,0
        - 0,0.777,0.209,0.856,0.482,0.901,0.79,0.889,1,0.818,1,0.684,0.914,0.625,0.225,0.63
    objects:
      filters:
        car:
          mask: 0,1,0,0,1,0,1,1
        person:
          mask: 
            0.224,0.666,0.101,0.762,0.252,0.838,0.481,0.856,0.684,0.846,0.838,0.824,0.996,0.773,1,0.741,0.893,0.671,0.515,0.672
      mask: 0.616,0.317,0.677,0.317,0.678,0.5,0.612,0.498
  Parkplatz:
    ffmpeg:
      inputs:
        - path: rtsp://192.168.8.2:8554/Parkplatz/main
          input_args: preset-rtsp-restream
          roles:
            - detect
    detect:
      width: 2560
      height: 1920
      stationary:
        interval: 50
        threshold: 200
    motion:
      mask: 
        0,1,0.467,1,0.055,0.575,0.044,0.427,0.114,0.387,0.284,0.349,0.342,0.191,0.594,0.168,0.801,0.167,1,0,0,0
    objects:
      filters:
        car:
          mask: 1,0,1,1,0,1,0,0
    ui:
      order: 5
  Kabuff:
    ffmpeg:
      inputs:
        - path: rtsp://192.168.8.2:8554/Kabuff/main
          input_args: preset-rtsp-restream
          roles:
            - detect
    detect:
      width: 2560
      height: 1920
      stationary:
        interval: 50
        threshold: 200
    motion:
      mask:
        - 1,0,1,0.469,0.71,0.374,0.361,0.314,0,0.326,0,0
        - 0.414,0.944,0.617,0.891,0.738,0.762,1,0.648,1,1,0,1,0,0.818,0.132,0.91
    objects:
      filters:
        car:
          mask: 0,1,1,1,1,0,0,0
    ui:
      order: 6
  Wintergarten:
    ffmpeg:
      inputs:
        - path: rtsp://192.168.8.2:8554/Wintergarten/main
          input_args: preset-rtsp-restream
          roles:
            - detect
    ui:
      order: 6
    detect:
      width: 2560
      height: 1920
      stationary:
        interval: 50
        threshold: 200
    motion:
      mask:
        - 0,0,0,0.275,0.241,0.202,0.368,0.175,0.49,0.162,0.575,0.168,0.641,0.181,0.691,0.191,0.704,0.096,0.825,0.136,1,0.184,1,0
        - 0.098,0.404,0.512,0.276,0.667,0.338,0.217,0.547
        - 1,0.527,0.918,0.808,0.929,1,1,1
        - 0,0.787,0.285,0.824,0.354,1,0,1
    objects:
      filters:
        car:
          mask: 0,1,0,0,1,0,1,1
        person:
          mask: 0.615,0.205,0.608,0.287,0.646,0.302,0.658,0.209
  Obstbaeume:
    ffmpeg:
      inputs:
        - path: rtsp://192.168.8.2:8554/Obstbaeume/sub
          input_args: preset-rtsp-restream
          roles:
            - detect
    ui:
      order: 7
    detect:
      width: 7680
      height: 2160
      stationary:
        interval: 50
        threshold: 200
    motion:
      mask: 
        0,0,0,0.119,0.173,0.117,0.299,0.174,0.348,0.246,0.367,0.363,0.39,0.47,0.418,0.532,0.438,0.615,0.494,0.615,0.545,0.646,0.639,0.707,0.726,0.764,0.77,0.809,0.829,0.848,0.88,0.866,0.924,0.869,1,0.832,1,0
    objects:
      filters:
        car:
          mask: 0,1,0,0,1,0,1,1
  Birds:
    ui:
      order: 8
    ffmpeg:
      inputs:
        - path: 
            rtsp://xxx:xxx@192.168.8.226:554/stream1#video=copy#audio=copy#audio=aac
          input_args: preset-rtsp-restream
          roles:
            - detect
    objects:
      track:
        - bird
    detect:
      width: 1280
      height: 720
    motion:
      mask: 
        0,1,0,0,1,0,1,1,0.877,1,0.882,0.901,0.842,0.761,0.751,0.661,0.656,0.61,0.528,0.58,0.403,0.582,0.295,0.615,0.182,0.71,0.126,0.847,0.129,1
    record:
      enabled: true
      alerts:
        retain:
          days: 90
          mode: motion
      detections:
        retain:
          days: 90
          mode: motion
  Wohnzimmer:
    ui:
      order: 9
    ffmpeg:
      inputs:
        - path: 
            rtsp://xxx:xxx@192.168.8.227:554/stream1#video=copy#audio=copy#audio=aac
          input_args: preset-rtsp-restream
    detect:
      enabled: false
record:
  enabled: true
  alerts:
    retain:
      days: 30
      mode: active_objects
  detections:
    retain:
      days: 30
      mode: active_objects
detectors:
  coral1:
    type: edgetpu
    device: pci:0
  coral2:
    type: edgetpu
    device: pci:1
ffmpeg:
  hwaccel_args: preset-intel-qsv-h264
objects:
  track:
    - person
    - dog
    - car
    - motorcycle
    - license_plate
  filters:
    person:
      min_score: 0.55
      threshold: 0.75
    dog:
      min_score: 0.65
      threshold: 0.85
    car:
      min_score: 0.6
      threshold: 0.8
    motorcycle:
      min_score: 0.6
      threshold: 0.8
detect:
  fps: 5
  stationary:
    interval: 50
    threshold: 50
  enabled: true
model:
  path: plus://xxx
motion:
  # Auflösung / Glättung
  frame_height: 150      # gute Detailtiefe für 4K
  frame_alpha: 0.02
  # Empfindlichkeit
  threshold: 35          # etwas über Default, filtert IR-Rauschen
  contour_area: 15       # bewusst RELATIV NIEDRIG für Fern-Erkennung
  # IR / Bildqualität
  improve_contrast: true
  lightning_threshold: 0.8
snapshots:
  # Optional: Enable writing jpg snapshot to /media/frigate/clips (default: shown below)
  enabled: true
  # Optional: save a clean PNG copy of the snapshot image (default: shown below)
  clean_copy: true
  # Optional: print a timestamp on the snapshots (default: shown below)
  timestamp: true
  # Optional: draw bounding box on the snapshots (default: shown below)
  bounding_box: true
  # Optional: crop the snapshot (default: shown below)
  crop: false
  # Optional: height to resize the snapshot to (default: original size)
#  height: 175
  # Optional: Restrict snapshots to objects that entered any of the listed zones (default: no required zones)
#  required_zones: []
  # Optional: Camera override for retention settings (default: global values)
  retain:
    # Required: Default retention days (default: shown below)
    default: 30
timestamp_style:
  position: tl
  format: '%d.%m.%Y %H:%M:%S'
version: 0.17-0

docker-compose file or Docker CLI command

services:
  frigate:
    container_name: frigate
    image: ghcr.io/blakeblackshear/frigate:stable
    restart: unless-stopped
    stop_grace_period: 30s
    privileged: true
    shm_size: "12288mb"

    devices:
      - /dev/dri/renderD128:/dev/dri/renderD128
      - /dev/apex_0:/dev/apex_0
      - /dev/apex_1:/dev/apex_1
    volumes:
      - /etc/localtime:/etc/localtime:ro
      - /home/sebastian/frigate/config:/config
      - /home/sebastian/frigate/media:/media/frigate
      - type: tmpfs
        target: /tmp/cache
        tmpfs:
          size: 4000000000

    ports:
      - "5000:5000"
      - "8971:8971"
      - "8554:8554"
      - "8555:8555/tcp"
      - "8555:8555/udp"

    environment:
      TZ: Europe/Berlin
      LIBVA_DRIVER_NAME: iHD
      FRIGATE_RTSP_PASSWORD: "XXX"
      PLUS_API_KEY: "XXX"

Relevant Frigate log output

2026-05-28 18:58:59.305124415  [INFO] Preparing Frigate...
2026-05-28 18:58:59.482329912  [INFO] Starting Frigate...
2026-05-28 18:59:01.410228592  [2026-05-28 18:59:01] frigate.util.config            INFO    : Checking if frigate config needs migration...
2026-05-28 18:59:01.440470953  [2026-05-28 18:59:01] frigate.util.config            INFO    : frigate config does not need migration...
2026-05-28 18:59:03.010319496  [2026-05-28 18:59:03] frigate.app                    INFO    : Starting Frigate (0.17.1-416a9b7)
2026-05-28 18:59:03.502465499  [2026-05-28 18:59:03] peewee_migrate.logs            INFO    : Starting migrations
2026-05-28 18:59:03.505772154  [2026-05-28 18:59:03] peewee_migrate.logs            INFO    : There is nothing to migrate
2026-05-28 18:59:04.114796396  [2026-05-28 18:59:04] frigate.app                    INFO    : Recording process started: 1183
2026-05-28 18:59:04.460482860  [2026-05-28 18:59:04] frigate.app                    INFO    : Review process started: 1203
2026-05-28 18:59:04.461661594  [2026-05-28 18:59:04] frigate.app                    INFO    : go2rtc process pid: 119
2026-05-28 18:59:05.220724888  [2026-05-28 18:59:05] frigate.app                    INFO    : Embedding process started: 1213
2026-05-28 18:59:05.587984458  [2026-05-28 18:59:05] frigate.detectors.plugins.edgetpu_tfl INFO    : Attempting to load TPU as pci:0
2026-05-28 18:59:05.606251138  [2026-05-28 18:59:05] frigate.detectors.plugins.edgetpu_tfl INFO    : TPU found
2026-05-28 18:59:05.992036704  [2026-05-28 18:59:05] frigate.detectors.plugins.edgetpu_tfl INFO    : Attempting to load TPU as pci:1
2026-05-28 18:59:06.044034965  [2026-05-28 18:59:06] frigate.detectors.plugins.edgetpu_tfl INFO    : TPU found
2026-05-28 18:59:06.435301216  [2026-05-28 18:59:06] frigate.app                    INFO    : Output process started: 1386
2026-05-28 18:59:06.490827824  [2026-05-28 18:59:06] frigate.camera.maintainer      INFO    : Camera processor started for test: 1469
2026-05-28 18:59:06.553259002  [2026-05-28 18:59:06] frigate.camera.maintainer      INFO    : Capture process started for test: 1526
2026-05-28 18:59:06.587390502  [2026-05-28 18:59:06] frigate.camera.maintainer      INFO    : Camera processor started for Tor: 1590
2026-05-28 18:59:06.614943202  [2026-05-28 18:59:06] frigate.api.fastapi_app        INFO    : Starting FastAPI app
2026-05-28 18:59:06.677625817  [2026-05-28 18:59:06] frigate.api.fastapi_app        INFO    : FastAPI started
2026-05-28 18:59:06.693640807  [2026-05-28 18:59:06] frigate.camera.maintainer      INFO    : Capture process started for Tor: 1660
2026-05-28 18:59:06.712386874  [2026-05-28 18:59:06] frigate.camera.maintainer      INFO    : Camera processor started for Strasse: 1728
2026-05-28 18:59:06.770909066  [2026-05-28 18:59:06] frigate.camera.maintainer      INFO    : Capture process started for Strasse: 1811
2026-05-28 18:59:06.802954492  [2026-05-28 18:59:06] frigate.camera.maintainer      INFO    : Camera processor started for Auffahrt: 1875
2026-05-28 18:59:06.832419943  [2026-05-28 18:59:06] frigate.camera.maintainer      INFO    : Capture process started for Auffahrt: 1945
2026-05-28 18:59:06.998920641  [2026-05-28 18:59:06] frigate.camera.maintainer      INFO    : Camera processor started for Pool: 2036
2026-05-28 18:59:07.185302198  [2026-05-28 18:59:07] frigate.camera.maintainer      INFO    : Capture process started for Pool: 2226
2026-05-28 18:59:07.224049750  [2026-05-28 18:59:07] frigate.camera.maintainer      INFO    : Camera processor started for Parkplatz: 2296
2026-05-28 18:59:07.254906817  [2026-05-28 18:59:07] frigate.camera.maintainer      INFO    : Capture process started for Parkplatz: 2424
2026-05-28 18:59:07.306053141  [2026-05-28 18:59:07] frigate.camera.maintainer      INFO    : Camera processor started for Kabuff: 2570
2026-05-28 18:59:07.378948610  [2026-05-28 18:59:07] frigate.camera.maintainer      INFO    : Capture process started for Kabuff: 2771
2026-05-28 18:59:07.401005099  [2026-05-28 18:59:07] frigate.camera.maintainer      INFO    : Camera processor started for Wintergarten: 2835
2026-05-28 18:59:07.425208878  [2026-05-28 18:59:07] frigate.camera.maintainer      INFO    : Capture process started for Wintergarten: 2912
2026-05-28 18:59:07.472050864  [2026-05-28 18:59:07] frigate.camera.maintainer      INFO    : Camera processor started for Obstbaeume: 2996
2026-05-28 18:59:07.526556739  [2026-05-28 18:59:07] frigate.camera.maintainer      INFO    : Capture process started for Obstbaeume: 3048
2026-05-28 18:59:07.542538436  [2026-05-28 18:59:07] frigate.camera.maintainer      INFO    : Camera processor started for Birds: 3117
2026-05-28 18:59:07.607497966  [2026-05-28 18:59:07] frigate.camera.maintainer      INFO    : Capture process started for Birds: 3194
2026-05-28 18:59:07.626191133  [2026-05-28 18:59:07] frigate.camera.maintainer      INFO    : Camera processor started for Wohnzimmer: 3258
2026-05-28 18:59:07.649300013  [2026-05-28 18:59:07] frigate.camera.maintainer      INFO    : Capture process started for Wohnzimmer: 3335
2026-05-28 19:01:37.748889770  [2026-05-28 19:01:37] frigate.data_processing.common.license_plate.mixin DEBUG   : test: Skipping LPR for non-stationary car object 1779987697.468444-m6krhs with no position changes.  (Detected in 3 concurrent frames, threshold to run is 4 frames)
2026-05-28 19:01:37.748889770  [LOGGING] Last message repeated 2 times
2026-05-28 19:01:40.453755527  [2026-05-28 19:01:40] frigate.data_processing.common.license_plate.mixin DEBUG   : test: No attributes to parse.
2026-05-28 19:01:40.453755527  [LOGGING] Last message repeated 11 times
2026-05-28 19:02:13.251581264  [2026-05-28 19:02:13] frigate.data_processing.common.license_plate.mixin DEBUG   : test: Not a processing license plate for non car/motorcycle object.
2026-05-28 19:02:13.534136407  [2026-05-28 19:02:13] frigate.data_processing.common.license_plate.mixin DEBUG   : test: No attributes to parse.
2026-05-28 19:02:13.534136407  [LOGGING] Last message repeated 2 times
2026-05-28 19:02:14.123641257  [2026-05-28 19:02:14] frigate.data_processing.common.license_plate.mixin DEBUG   : test: Not a processing license plate for non car/motorcycle object.
2026-05-28 19:02:14.339465657  [2026-05-28 19:02:14] frigate.data_processing.common.license_plate.mixin DEBUG   : test: No attributes to parse.
2026-05-28 19:02:14.339465657  [LOGGING] Last message repeated 13 times

Install method

Docker Compose

Object Detector

Coral

Screenshots of the Frigate UI's System metrics pages

Any other information that may be helpful

No response

[inkeep-ai-assistant[bot]]

The behavior you're seeing is expected based on how Frigate+ license plate detection works. When using a Frigate+ model with license_plate in your objects to track, the license_plate detection is treated as an attribute of the car or motorcycle object, not as a standalone detection that triggers LPR independently. (1)

Your log messages reveal the issue:

• "No attributes to parse." - The Frigate+ model is detecting a car but not returning a license_plate attribute box on that car detection
• "Skipping LPR for non-stationary car object ... Detected in 3 concurrent frames, threshold to run is 4" - The car isn't meeting the stationary threshold for LPR to run

This suggests your Frigate+ model may not be detecting license_plate as an attribute on the car objects in your scene. The standalone YOLOv9 detector works because it runs independently and searches the entire car region for plates. (1)

A few things to check:

1. Verify license_plate detection in debug view - Watch the debug view (Settings → Debug) to confirm whether license_plate bounding boxes are appearing on the detected cars. If not, the Frigate+ model isn't detecting plates in your scene. (1)

2. Adjust object filters for license_plate - You may need to lower the min_score and/or threshold for the license_plate object if plates aren't being detected: (1)

objects:
  track:
    - car
    - motorcycle
    - license_plate
  filters:
    license_plate:
      threshold: 0.7
      min_score: 0.5

(1), (2)

3. Check plate size - The min_area default is 1000 pixels. If your plates are small in frame, they may not meet the threshold. (1)

4. Consider the stationary requirement - For moving vehicles, LPR continuously refines recognition, but your logs show the car isn't meeting the frame threshold. With your looped test clip, the car may be appearing/disappearing in ways that reset the frame counter. (1)

If your Frigate+ model consistently fails to detect license_plate attributes on cars where plates are clearly visible, you may want to submit those images to Frigate+ for model training/fine-tuning, annotating both the car and the license_plate in the snapshots. (1)

📚 Sources:

1. License Plate Recognition (LPR)
2. Full Reference Config

---

Have another question? Just tag @inkeep.

[hawkeye217]

The AI's response is correct here.

If your Frigate+ model is not detecting license_plate with a car and you've tried the AI's steps above to lower the min_score, I'd suggest switching to a Frigate+ base model to rule out fine-tuning issues on your model.

[cho0p]

Just tried it with

    license_plate:
      threshold: 0.5
      min_score: 0.3

I do not see a single box in the debug view for the license plate.. The car is 1 m in front of the camera. You can see the plate crystal clear. I thought the 2026.1 Base Models for Frigate+ do detect license plates.

[hawkeye217]

I thought the 2026.1 Base Models for Frigate+ do detect license plates.

Yes, they do. Are you running a fine-tuned Frigate+ model or a base model?

[cho0p]

It is a fine-tuned Frigate+ model but I never used or trained license_plate before.

[hawkeye217]

If you've submitted images of cars and have not annotated license_plate on any of them, then you've effectively tuned out license plates from your fine-tuned model.

[cho0p]

Thanks @hawkeye217 ! Indeed I have uploaded lots of cars and not annotated license_plate in the past. What would be the best way forward? Start from scratch with a new account and fine-tuned model with new training images or keep the current models and start uploading and annotating license_plate?

[hawkeye217]

No need to start over, the work you've done is already useful. You can just go back through your images and annotate license plates. In the meantime, if you do need to detect plates with a Frigate+ model, you can use a base model. Or, just remove license_plate from your list of objects to track in your config to use the secondary detector until the new Frigate+ model is ready.