[Support]: [16-rc1] Camera flags for disabling face and license plate recognition do not work as expected

[PilaScat]

Checklist

• I have updated to the latest available Frigate version.
• I have cleared the cache of my browser.
• I have tried a different browser to see if it is related to my browser.
• I have tried reproducing the issue in incognito mode to rule out problems with any third party extensions or plugins I have installed.

Describe the problem you are having

1. The camera flags intended to disable face recognition and license plate recognition do not function as expected. When these flags are set to "false" for specific cameras, Frigate ignores the configuration and proceeds to perform face and license plate recognition anyway. This makes it impossible to selectively disable these features per camera as intended.

2. The number of faces to be trained appears to be limited to 100. Once the system reaches this limit, Frigate automatically deletes older faces to make room for new ones. This continuous replacement makes training and management challenging.

Steps to reproduce

1. Configure a camera with the flags to disable face and license plate recognition.
2. Observe that recognition still occurs despite the flags.

Version

16-rc1

In which browser(s) are you experiencing the issue with?

No response

Frigate config file

mqtt:
  host: 192.168.1.20
  user: frigate
  password: 

detectors:
  coral:
    type: edgetpu
    device: pci:0

camera_groups:
  Casa:
    order: 1
    icon: LuHouse
    cameras:
      - Ingresso
      - Giardino
      - Entrata
      - Cucina
      - Salotto
  Nonno:
    order: 2
    icon: LuEar
    cameras:
      - Nonno-Soggiorno
      - Nonno-Orto
      - Nonno-Cortile
      - Nonno-Lavanderia
      - Nonno-Camera
      - Nonno-Corridoio
      - Nonno-Giardino
      - Nonno-Ingresso

record:
  enabled: true
  alerts:
    retain:
      days: 7
  detections:
    retain:
      days: 7

detect:
  enabled: true

face_recognition:
  enabled: true
  model_size: large

lpr:
  enabled: true
  model_size: large

go2rtc:
  streams:
    ### CASA ###
    Ingresso_main:
      - rtsp://192.168.1.31:554/Preview_01_main
      - ffmpeg:Ingresso_main#audio=opus
    Ingresso_sub:
      - rtsp://192.168.1.31:554/Preview_01_sub
      - ffmpeg:Ingresso_sub#audio=opus
    Giardino_main:
      - rtsp://192.168.1.32:554/Preview_01_main
      - ffmpeg:Giardino_main#audio=opus
    Giardino_sub:
      - rtsp://192.168.1.32:554/Preview_01_sub
      - ffmpeg:Giardino_sub#audio=opus
    Cucina:
      - tapo://
    Salotto:
      - tapo://
    Entrata:
      - tapo://
    ### NONNO ###
    Nonno-Soggiorno:
      - rtsp://192.168.2.30:554/Preview_01_main
      - ffmpeg:Nonno-Soggiorno#audio=opus
    Nonno-Orto_main:
      - rtsp://192.168.2.31:554/Preview_01_main
      - ffmpeg:Nonno-Orto_main#audio=opus
    Nonno-Orto_sub:
      - rtsp://192.168.2.31:554/Preview_01_sub
      - ffmpeg:Nonno-Orto_sub#audio=opus
    Nonno-Cortile_main:
      - rtsp://192.168.2.32:554/Preview_01_main
      - ffmpeg:Nonno-Cortile_main#audio=opus
    Nonno-Cortile_sub:
      - rtsp://192.168.2.32:554/Preview_01_sub
      - ffmpeg:Nonno-Cortile_sub#audio=opus
    Nonno-Giardino_main:
      - rtsp://192.168.2.37:554/Preview_01_main
      - ffmpeg:Nonno-Giardino_main#audio=opus
    Nonno-Giardino_sub:
      - rtsp://192.168.2.37:554/Preview_01_sub
      - ffmpeg:Nonno-Giardino_sub#audio=opus
    Nonno-Lavanderia:
      - tapo://
    Nonno-Camera:
      - tapo://
    Nonno-Corridoio:
      - tapo://
    Nonno-Ingresso:
      - tapo://
  webrtc:
    candidates:
      - 192.168.1.20:8555
      - stun:8555
      - 100.64.0.0/10

cameras:
  ### CASA ###
  Ingresso:
    live:
      streams:
        Ingresso_main: Ingresso_main
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/Ingresso_main
          input_args: preset-rtsp-restream
          hwaccel_args: preset-intel-qsv-h265
          roles:
            - record
        - path: rtsp://127.0.0.1:8554/Ingresso_sub
          input_args: preset-rtsp-restream
          hwaccel_args: preset-intel-qsv-h264
          roles:
            - detect
    detect:
      width: 640
      height: 360
    record:
      retain:
        days: 1
    motion:
      mask: 0,0,0.306,0,0.306,0.074,0,0.072
    lpr:
      enabled: false

  Giardino:
    live:
      streams:
        Giardino_main: Giardino_main
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/Giardino_main
          input_args: preset-rtsp-restream
          hwaccel_args: preset-intel-qsv-h265
          roles:
            - record
        - path: rtsp://127.0.0.1:8554/Giardino_sub
          input_args: preset-rtsp-restream
          hwaccel_args: preset-intel-qsv-h264
          roles:
            - detect
    detect:
      width: 1536
      height: 576
    record:
      retain:
        days: 1
    mqtt:
      quality: 100
    motion:
      mask: 0,0,0.21,0,0.211,0.074,0,0.072
    objects:
      filters:
        person:
          min_area: 1500

  Cucina:
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/Cucina
          input_args: preset-rtsp-restream
          hwaccel_args: preset-intel-qsv-h264
          roles:
            - record
            - detect
    onvif:
      host: 192.168.1.33
      port: 2020
      user: PilaScat
      password:
    detect:
      width: 2560
      height: 1440
    objects:
      filters:
        person:
          min_area: 50000
    motion:
      mask: 0,0,0.357,0,0.356,0.049,0,0.049
    face_recognition:
      enabled: false
    lpr:
      enabled: false

  Salotto:
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/Salotto
          input_args: preset-rtsp-restream
          hwaccel_args: preset-intel-qsv-h264
          roles:
            - record
            - detect
    onvif:
      host: 192.168.1.34
      port: 2020
      user: PilaScat
      password:
    detect:
      width: 2560
      height: 1440
    objects:
      filters:
        person:
          min_area: 20000
    motion:
      mask: 0,0,0.357,0,0.356,0.049,0,0.049
    face_recognition:
      enabled: false
    lpr:
      enabled: false

  Entrata:
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/Entrata
          input_args: preset-rtsp-restream
          hwaccel_args: preset-intel-qsv-h264
          roles:
            - record
            - detect
    onvif:
      host: 192.168.1.35
      port: 2020
      user: PilaScat
      password:
    record:
      retain:
        days: 1
    detect:
      width: 2560
      height: 1440
    objects:
      filters:
        person:
          min_area: 20000
    motion:
      mask: 0,0,0.357,0,0.356,0.049,0,0.049
    face_recognition:
      enabled: false
    lpr:
      enabled: false

  ### NONNO ###
  Nonno-Soggiorno:
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/Nonno-Soggiorno
          input_args: preset-rtsp-restream
          hwaccel_args: preset-intel-qsv-h264
          roles:
            - record
            - detect
    onvif:
      host: 192.168.2.30
      port: 8000
      user: admin
      password: 
    detect:
      width: 2560
      height: 1440
    objects:
      filters:
        person:
          min_area: 20000
    motion:
      mask: 0,0,0.294,0,0.294,0.051,0,0.052
    face_recognition:
      enabled: false
    lpr:
      enabled: false

  Nonno-Orto:
    live:
      streams:
        Nonno-Orto_main: Nonno-Orto_main
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/Nonno-Orto_main
          input_args: preset-rtsp-restream
          hwaccel_args: preset-intel-qsv-h265
          roles:
            - record
        - path: rtsp://127.0.0.1:8554/Nonno-Orto_sub
          input_args: preset-rtsp-restream
          hwaccel_args: preset-intel-qsv-h264
          roles:
            - detect
    detect:
      width: 640
      height: 360
    motion:
      mask: 0.001,0,0.308,0,0.307,0.07,0,0.07
    lpr:
      enabled: false

  Nonno-Cortile:
    live:
      streams:
        Nonno-Cortile_main: Nonno-Cortile_main
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/Nonno-Cortile_main
          input_args: preset-rtsp-restream
          hwaccel_args: preset-intel-qsv-h265
          roles:
            - record
        - path: rtsp://127.0.0.1:8554/Nonno-Cortile_sub
          input_args: preset-rtsp-restream
          hwaccel_args: preset-intel-qsv-h264
          roles:
            - detect
    onvif:
      host: 192.168.2.32
      port: 8000
      user: admin
      password: 
    detect:
      width: 640
      height: 360
    motion:
      mask: 0.001,0,0.308,0,0.307,0.07,0,0.07

  Nonno-Giardino:
    live:
      streams:
        Nonno-Giardino_main: Nonno-Giardino_main
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/Nonno-Giardino_main
          input_args: preset-rtsp-restream
          hwaccel_args: preset-intel-qsv-h265
          roles:
            - record
        - path: rtsp://127.0.0.1:8554/Nonno-Giardino_sub
          input_args: preset-rtsp-restream
          hwaccel_args: preset-intel-qsv-h264
          roles:
            - detect
    detect:
      width: 640
      height: 360
    motion:
      mask: 0.001,0,0.308,0,0.307,0.07,0,0.07

  Nonno-Lavanderia:
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/Nonno-Lavanderia
          input_args: preset-rtsp-restream
          hwaccel_args: preset-intel-qsv-h264
          roles:
            - record
            - detect
    onvif:
      host: 192.168.2.33
      port: 2020
      user: PilaScat
      password:
    detect:
      width: 2560
      height: 1440
    motion:
      mask: 0,0,0.357,0,0.356,0.049,0,0.049
    face_recognition:
      enabled: false
    lpr:
      enabled: false

  Nonno-Camera:
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/Nonno-Camera
          input_args: preset-rtsp-restream
          hwaccel_args: preset-intel-qsv-h264
          roles:
            - record
            - detect
    onvif:
      host: 192.168.2.34
      port: 2020
      user: PilaScat
      password:
    detect:
      width: 2560
      height: 1440
    motion:
      mask: 0,0,0.357,0,0.356,0.049,0,0.049
    face_recognition:
      enabled: false
    lpr:
      enabled: false

  Nonno-Corridoio:
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/Nonno-Corridoio
          input_args: preset-rtsp-restream
          hwaccel_args: preset-intel-qsv-h264
          roles:
            - record
            - detect
    onvif:
      host: 192.168.2.35
      port: 2020
      user: PilaScat
      password:
    detect:
      width: 2560
      height: 1440
    motion:
      mask: 0,0,0.357,0,0.356,0.049,0,0.049
    face_recognition:
      enabled: false
    lpr:
      enabled: false

  Nonno-Ingresso:
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/Nonno-Ingresso
          input_args: preset-rtsp-restream
          hwaccel_args: preset-intel-qsv-h264
          roles:
            - record
            - detect
    onvif:
      host: 192.168.2.36
      port: 2020
      user: PilaScat
      password:
    detect:
      width: 2560
      height: 1440
    motion:
      mask: 0,0,0.357,0,0.356,0.049,0,0.049
    face_recognition:
      enabled: false
    lpr:
      enabled: false

version: 0.16-0

docker-compose file or Docker CLI command

not necessary, same as always

Relevant Frigate log output

seems nothing

Relevant go2rtc log output

not necessary

Operating system

UNRAID

Install method

Docker CLI

Network connection

Wired

Camera make and model

not necessary

Screenshots of the Frigate UI's System metrics pages

not necessary

Any other information that may be helpful

No response

[hawkeye217]

• The camera flags intended to disable face recognition and license plate recognition do not function as expected. When these flags are set to "false" for specific cameras, Frigate ignores the configuration and proceeds to perform face and license plate recognition anyway. This makes it impossible to selectively disable these features per camera as intended.

I can't reproduce this. This has also been well-tested for months by other users during the 0.16 beta period. What makes you think that the recognition pipeline is actually running?

The code has explicit checks:

frigate/frigate/data_processing/common/license_plate/mixin.py

Line 1194 in 898d1de

if not self.config.cameras[camera].lpr.enabled:

frigate/frigate/data_processing/real_time/face.py

Line 166 in 898d1de

if not self.config.cameras[camera].face_recognition.enabled:

• The number of faces to be trained appears to be limited to 100. Once the system reaches this limit, Frigate automatically deletes older faces to make room for new ones. This continuous replacement makes training and management challenging.

This is not a bug. Please read the beta documentation for the save_attempts parameter. This is a user-configurable option that defaults to 100.

https://deploy-preview-16390--frigate-docs.netlify.app/configuration/face_recognition/#recognition

[PilaScat]

I can't reproduce this. This has also been well-tested for months by other users during the 0.16 beta period. What makes you think that the recognition pipeline is actually running?

I see on the train tab images from disabled cameras, seems not the way it should work

This is not a bug. Please read the beta documentation for the save_attempts parameter. This is a user-configurable option that defaults to 100.

Didn't saw this is the config, thank you

[NickM-27]

Were the images created while face recognition was disabled at the camera level?

[PilaScat]

Were the images created while face recognition was disabled at the camera level?

Interesting question, seems like it 🤦‍♂️

[PilaScat]

I take this opportunity to say that selecting faces with the control key is not reliable.