[Config Support]: Wanted to upgrade from 14 to 15 and migrate from LXC to Docker. No detectors work

[ozhound]

Describe the problem you are having

My version 14 of frigate was working fine using the IGPU on my proxmox system via the frigate LXC script. I wanted to upgrade to 15 but the LXC required a fresh install, however this has not been updated. I decided to migrate to a docker VM, I firstly started with the Docker LXC, but this couldn't use my iGPU is it simply wasn't detected and wasn't passed from the host to the LXC (even though it worked fine with the frigate LXC) however with the docker LXC at least the CPU detection worked (as the config below is currently set)

So then I tried to install a full docker VM from the tteck script repository. This all went fine, I migrated my config, however when the system started it kept crashing complaining about a detector not been found. I'm simply using the open vino and CPU detection method which is worked in the past on exactly the same hardware, but now frigate is stuck in a boot loop. I don't understand why the open vino using CPU no longer works.

Can someone point me in the right direction

Version

0.15.0-cea210d

Frigate config file

mqtt:
  enabled: true
  host: 192.168.1.25
  user: homeassistant
  
go2rtc:
  streams:
    Driveway:
      - tapo://x@192.168.1.193
      - ffmpeg:Driveway#audio=aac
    Laundry:
      - tapo://x@192.168.1.191
      - ffmpeg:Laundry#audio=aac
    MasterBed:
      - tapo://x@192.168.1.192
      - ffmpeg:MasterBed#audio=aac
    backyard:
      - tapo://x@192.168.1.190
      - ffmpeg:backyard#audio=aac
#audio: # <- enable audio events for all camera
#  enabled: true
snapshots:
  enabled: true
  clean_copy: true
  timestamp: false
  bounding_box: true
  crop: false
  height: 175
  required_zones: []
  retain:
    default: 10
    objects:
      person: 15
  quality: 100
record:
  enabled: true
  retain:
    days: 3
    mode: all
  alerts:
    pre_capture: 10
    post_capture: 10
    retain:
      days: 2
      mode: motion
  detections:
    pre_capture: 10
    post_capture: 10
    retain:
      days: 10
      mode: active_objects
  preview:
      # Options are: very_low, low, medium, high, very_high
    quality: very_high
cameras:
  backyard: # <------ Name the camera
    enabled: true
    ffmpeg:
      output_args:
        record: preset-record-generic-audio-aac
      inputs:
        - path: rtsp://127.0.0.1:8554/backyard?video&audio # <--- the name here must match the name of the camera in restream
          input_args: preset-rtsp-restream
          roles:
            - record
            - detect
 #           - audio
    detect:
      width: 1280
      height: 720
      fps: 5
    zones:
      Lawn:
        coordinates: 
          0.544,0.475,0.485,0.575,0.201,0.607,0.008,0.82,0,1,0.999,0.994,0.922,0.832,0.898,0.623,0.781,0.485
        loitering_time: 0
        inertia: 3
      Foliage:
        coordinates: 
          0.801,0.2,0.752,0.088,0.744,0,0.693,0.01,0,0,0,0.823,0.204,0.603,0.485,0.571,0.546,0.469,0.779,0.491
        loitering_time: 0
        objects: person
        inertia: 3
      Carport:
        coordinates: 0.803,0.197,0.901,0.262,0.899,0.622,0.786,0.486
        loitering_time: 0
        inertia: 3
    motion:
      mask: 0.003,0.084,0.17,0.082,0.172,0.038,0.354,0.05,0.35,0,0.001,0.001
      threshold: 40
      contour_area: 20
      improve_contrast: true
    objects:
      mask:
        - 0.145,0.402,0.151,0.527,0.199,0.523,0.192,0.393
        - 0.379,0.334,0.407,0.337,0.404,0.408,0.379,0.401
      filters:
        person:
          mask:
            - 0.608,0.31,0.574,0.421,0.636,0.424,0.632,0.319
            - 0.379,0.334,0.407,0.337,0.401,0.49,0.367,0.494
            - 0.145,0.402,0.151,0.527,0.199,0.523,0.192,0.393
            - 0.379,0.334,0.407,0.337,0.404,0.408,0.379,0.401
  Laundry: # <------ Name the camera
    enabled: true
    ffmpeg:
      output_args:
        record: preset-record-generic-audio-aac
      inputs:
        - path: rtsp://127.0.0.1:8554/Laundry?video&audio # <--- the name here must match the name of the camera in restream
          input_args: preset-rtsp-restream
          roles:
            - record
            - detect
    detect:
      width: 1280
      height: 720
      fps: 5
    motion:
      mask: 0.001,0.049,0.362,0.046,0.36,0.002,0,0
      threshold: 30
      contour_area: 20
    zones:
      All:
        coordinates: 0.001,0.313,0.613,0.01,0.917,0.014,0.997,0.018,0.997,0.986,0.009,0.984
        loitering_time: 0
    objects:
      filters:
        person:
          mask: 0.621,0,0.834,0,0.816,0.407,0.814,0.51,0.656,0.491
  MasterBed: # <------ Name the camera
    enabled: true
    ffmpeg:
      output_args:
        record: preset-record-generic-audio-aac
      inputs:
        - path: rtsp://127.0.0.1:8554/MasterBed?video&audio # <--- the name here must match the name of the camera in restream
          input_args: preset-rtsp-restream
          roles:
            - record
            - detect
    detect:
      width: 1280
      height: 720
      fps: 5
    motion:
      mask:
        - 0.001,0.001,0.357,0,0.356,0.05,0.002,0.042
        - 0.054,0.117,0.052,0.039,0.106,0.043,0.121,0.212,0.133,0.307,0.107,0.36,0.091,0.424,0.047,0.419,0.044,0.233
      threshold: 30
      contour_area: 30
    objects:
      filters:
        person:
          mask: 
            0,0.043,0.156,0.048,0.165,0.4,0.104,0.426,0.111,0.511,0.102,0.517,0.09,0.433,0.001,0.44
    zones:
      All:
        coordinates: 0.163,0.184,0.579,0.119,0.995,0.143,0.995,0.99,0.008,0.994,0.01,0.192
        loitering_time: 0
  Driveway: # <------ Name the camera
    enabled: true
    ffmpeg:
      output_args:
        record: preset-record-generic-audio-aac
      inputs:
        - path: rtsp://127.0.0.1:8554/Driveway?video&audio # <--- the name here must match the name of the camera in restream
          input_args: preset-rtsp-restream
          roles:
            - record
            - detect
    detect:
      width: 1280
      height: 720
      fps: 10
    motion:
      mask: 0.001,0.049,0.362,0.046,0.36,0.002,0,0
      threshold: 30
      contour_area: 20
    objects:
      track:
        - person
        - car
        - bird
        - airplane
        - bicycle
        - motorcycle
        - bus
        - cat
        - dog
        - cell phone
      filters:
        car: {}
        person:
          mask: 0.753,0.299,0.755,0.251,0.81,0.245,0.807,0.295
    zones:
      Entryway:
        coordinates: 0.969,0,0.957,0.482,0.663,0.4,0.677,0.373,0.672,0.001
        loitering_time: 5
        inertia: 3
        objects: person
      MainDriveway:
        coordinates: 
          1,0.495,0.999,0.997,0.004,0.999,0,0.635,0.095,0.549,0.182,0.475,0.28,0.404,0.41,0.311,0.41,0.35,0.621,0.392,0.658,0.401
        loitering_time: 3
        inertia: 3
      Roadway:
        coordinates: 
          0.002,0.632,0.178,0.476,0.279,0.402,0.426,0.302,0.521,0.237,0.641,0.169,0.636,0.105,0.473,0.164,0.25,0.281,0.115,0.366,0,0.449
        loitering_time: 0
        inertia: 3
      Bus_Stop:
        coordinates: 0.496,0.001,0.503,0.147,0.599,0.103,0.598,0.003
        loitering_time: 5
        objects: person
        inertia: 3
      Henley:
        coordinates: 
          0.004,0.144,0.232,0,0.49,0.003,0.638,0,0.637,0.084,0.493,0.124,0.495,0.145,0.406,0.184,0.259,0.254,0,0.419
        loitering_time: 0
        inertia: 3
      Grass:
        coordinates: 
          0.67,0.005,0.677,0.214,0.634,0.229,0.639,0.357,0.615,0.394,0.411,0.344,0.41,0.315,0.52,0.239,0.636,0.174,0.638,0.02
        loitering_time: 0
        objects: person
    review:
      alerts:
        required_zones:
          - Entryway
          - MainDriveway
      detections:
        required_zones:
          - Entryway
          - MainDriveway
          - Roadway
          - Grass
detectors:
  ov:
    type: openvino
    device: CPU
model:
  width: 300
  height: 300
  input_tensor: nhwc
  input_pixel_format: bgr
  path: /openvino-model/ssdlite_mobilenet_v2.xml
  labelmap_path: /openvino-model/coco_91cl_bkgr.txt
version: 0.15.1

Relevant Frigate log output

2025-02-18 04:57:09.059272006  [2025-02-18 04:57:09] peewee_migrate.logs            INFO    : Starting migrations
2025-02-18 04:57:09.068178974  [2025-02-18 04:57:09] peewee_migrate.logs            INFO    : There is nothing to migrate
2025-02-18 04:57:09.068267294  [2025-02-18 04:57:09] frigate.app                    INFO    : Recording process started: 359
2025-02-18 04:57:09.068282350  [2025-02-18 04:57:09] frigate.app                    INFO    : Review process started: 360
2025-02-18 04:57:09.071099500  [2025-02-18 04:57:09] frigate.app                    INFO    : go2rtc process pid: 110
2025-02-18 04:57:09.082872297  [2025-02-18 04:57:09] detector.ov                    INFO    : Starting detection process: 375
2025-02-18 04:57:09.095492694  [2025-02-18 04:57:09] frigate.app                    INFO    : Output process started: 391
2025-02-18 04:57:09.111808493  [2025-02-18 04:57:09] frigate.app                    INFO    : Camera processor started for backyard: 406
2025-02-18 04:57:09.118285109  [2025-02-18 04:57:09] frigate.app                    INFO    : Camera processor started for Laundry: 408
2025-02-18 04:57:09.131392976  [2025-02-18 04:57:09] frigate.app                    INFO    : Camera processor started for MasterBed: 410
2025-02-18 04:57:09.134022049  [2025-02-18 04:57:09] frigate.app                    INFO    : Camera processor started for Driveway: 412
2025-02-18 04:57:09.146316721  [2025-02-18 04:57:09] frigate.app                    INFO    : Capture process started for backyard: 428
2025-02-18 04:57:09.179996423  [2025-02-18 04:57:09] frigate.app                    INFO    : Capture process started for Laundry: 443
2025-02-18 04:57:09.188322936  [2025-02-18 04:57:09] frigate.app                    INFO    : Capture process started for MasterBed: 446
2025-02-18 04:57:09.224019418  [2025-02-18 04:57:09] frigate.app                    INFO    : Capture process started for Driveway: 457
2025-02-18 04:57:09.268466064  Process detector:ov:
2025-02-18 04:57:09.268468630  Traceback (most recent call last):
2025-02-18 04:57:09.268469660    File "/usr/lib/python3.9/multiprocessing/process.py", line 315, in _bootstrap
2025-02-18 04:57:09.268473179      self.run()
2025-02-18 04:57:09.268474714    File "/opt/frigate/frigate/util/process.py", line 41, in run_wrapper
2025-02-18 04:57:09.268475733      return run(*args, **kwargs)
2025-02-18 04:57:09.268489264    File "/usr/lib/python3.9/multiprocessing/process.py", line 108, in run
2025-02-18 04:57:09.268494284      self._target(*self._args, **self._kwargs)
2025-02-18 04:57:09.268496104    File "/opt/frigate/frigate/object_detection.py", line 121, in run_detector
2025-02-18 04:57:09.268497399      object_detector = LocalObjectDetector(detector_config=detector_config)
2025-02-18 04:57:09.268498637    File "/opt/frigate/frigate/object_detection.py", line 68, in __init__
2025-02-18 04:57:09.268499759      self.detect_api = create_detector(detector_config)
2025-02-18 04:57:09.268522629    File "/opt/frigate/frigate/detectors/__init__.py", line 18, in create_detector
2025-02-18 04:57:09.268523993      return api(detector_config)
2025-02-18 04:57:09.268538021    File "/opt/frigate/frigate/detectors/plugins/openvino.py", line 40, in __init__
2025-02-18 04:57:09.268539332      self.interpreter = self.ov_core.compile_model(
2025-02-18 04:57:09.268540811    File "/usr/local/lib/python3.9/dist-packages/openvino/runtime/ie_api.py", line 543, in compile_model
2025-02-18 04:57:09.268542084      super().compile_model(model, device_name, {} if config is None else config),
2025-02-18 04:57:09.268543376  RuntimeError: Exception from src/inference/src/cpp/core.cpp:121:
2025-02-18 04:57:09.268544523  Exception from src/inference/src/dev/plugin.cpp:53:
2025-02-18 04:57:09.268545580  could not create a primitive
2025-02-18 04:57:24.721417  2025-02-18 04:57:09.268564099
2025-02-18 04:57:24.721428  2025-02-18 04:57:09.268565165
2025-02-18 04:57:09.403726974  [2025-02-18 04:57:09] frigate.api.fastapi_app        INFO    : Starting FastAPI app
2025-02-18 04:57:09.510386708  [2025-02-18 04:57:09] frigate.api.fastapi_app        INFO    : FastAPI started

Relevant go2rtc log output

2025-02-18 04:59:51.247694073  [INFO] Preparing new go2rtc config...
2025-02-18 04:59:51.628028222  [INFO] Starting go2rtc...
2025-02-18 04:59:51.705378690  04:59:51.705 INF go2rtc platform=linux/amd64 revision=b2399f3 version=1.9.2
2025-02-18 04:59:51.705381471  04:59:51.705 INF config path=/dev/shm/go2rtc.yaml
2025-02-18 04:59:51.705730749  04:59:51.705 INF [api] listen addr=:1984
2025-02-18 04:59:51.705776346  04:59:51.705 INF [rtsp] listen addr=:8554
2025-02-18 04:59:51.705914538  04:59:51.705 INF [webrtc] listen addr=:8555/tcp

Frigate stats

No response

Operating system

Proxmox

Install method

Docker Compose

docker-compose file or Docker CLI command

version: "3.9"
services:
  frigate:
    container_name: frigate
    privileged: true # this may not be necessary for all setups
    restart: unless-stopped
    stop_grace_period: 30s # allow enough time to shut down the various services
    image: ghcr.io/blakeblackshear/frigate:stable
    shm_size: "512mb" # update for your cameras based on calculation above
    devices:
      - /dev/bus/usb:/dev/bus/usb # Passes the USB Coral, needs to be modified for other versions
      - /dev/apex_0:/dev/apex_0 # Passes a PCIe Coral, follow driver instructions here https://coral.ai/docs/m2/get-started/#2a-on-linux
      - /dev/video11:/dev/video11 # For Raspberry Pi 4B
      - /dev/dri/renderD128:/dev/dri/renderD128 # For intel hwaccel, needs to be updated for your hardware
    volumes:
      - /etc/localtime:/etc/localtime:ro
      - /config/frigate:/config
      - /media/frigate:/media/frigate
      - type: tmpfs # Optional: 1GB of memory, reduces SSD/SD Card wear
        target: /tmp/cache
        tmpfs:
          size: 1000000000
    ports:
      - "8971:8971"
      #- "5000:5000" # Internal unauthenticated access. Expose carefully.
      - "8554:8554" # RTSP feeds
      - "8555:8555/tcp" # WebRTC over tcp
      - "8555:8555/udp" # WebRTC over udp
    environment:
      FRIGATE_RTSP_PASSWORD: "x"
      NEOReadDebugKeys: "1"
      OverrideGpuAddressSpace: "48"

Object Detector

OpenVino

Screenshots of the Frigate UI's System metrics pages

No response

Any other information that may be helpful

Proxmox Host --> Docker VM (https://community-scripts.github.io/ProxmoxVE/scripts?id=docker-vm) --> Portainer 2.21.5

[hawkeye217]

While some users have found success, Proxmox is an unsupported installation method. See the official docs for suggestions: https://docs.frigate.video/frigate/installation#proxmox

Also, see this comment from a user: #15644 (reply in thread)

[Shred99]

I've had success running Frigate on proxmox as follows:

• Used the tteck script to create an LXC hosting Docker
• Installed Frigate 0.14.1 (now 0.15.0) in Docker

I wonder if OpenVino actually requires the GPU?

It was a bit fiddly getting the USB connected Coral and the GPU mapped through to the Docker container, but once it's working, it is stable and I had no problems upgrading to 0.15.0. Once you have the GPU mapped through to the docker container - try it again?

I had to install the usbutils in the LXC hosting docker and the proxmox host from memory ("apt install usbutils") so that I could see which bus the Coral device was on and debug things.

The following posts were helpful:

#5773

https://community.home-assistant.io/t/google-coral-usb-frigate-proxmox/383737

Doing an "lspcu -nnv | grep VGA" was useful for seeing how Proxmox saw the GPU.

[ozhound]

Thanks for the resources Shred99. the HA one is far too old to shed and light on my particular situation, however, this did give me motivation to try again. Which was worthwhile as i got it working finally.

So my issue was as follows.

Proxmox host detected the Coral as the generic

ID 18d1:9302 Global Unichip Corp

The docker LXC detected the same

ID 18d1:9302 Global Unichip Corp

However when lsusb was run on the frigate docker container it returned the following

ID 18d1:9302

I read on another post that the Coral needed to go through an initialization process before it would be detected and could be used by frigate.

Attempting to simply follow the setup and test instructions here https://coral.ai/docs/accelerator/get-started/ doesnt work as all the dependencies are outdated and no long installable by the pycoral command.

in order to get pycroal installed i had to run the pip install commands as referenced here

google-coral/edgetpu#771 (comment)

I also had to manually install Pillow and Numby as referenced at these 2 websites

pip install Pillow
pip install "numpy<2.0"

https://stackoverflow.com/questions/78641150/a-module-that-was-compiled-using-numpy-1-x-cannot-be-run-in-numpy-2-0-0

https://stackoverflow.com/questions/8863917/importerror-no-module-named-pil

only then was i able to run the command to test the Coral

python3 examples/classify_image.py \ --model test_data/mobilenet_v2_1.0_224_inat_bird_quant_edgetpu.tflite \ --labels test_data/inat_bird_labels.txt \ --input test_data/parrot.jpg

once this ran running the command lsusb FINALLY reported the initialized Coral (both in the proxmox host AND the docker LXC)

Bus 004 Device 005: ID 18d1:9302 Google Inc.

I updated the Frigate config to use the Coral TPU as a detector and it picked it up and used it straight away

The Coral-ai "getting started" page needs an update. but i read in my travels that google doesnt seem to be reposive to those requests as the board is no longer in development?

anyway. thanks again. i hope the above helps someone in the future