[Detector Support]: M.2 Coral not detected by frigate docker container

[sefininio]

Describe the problem you are having

m.2 coral identified by host machine.

$ ls /dev/apex_0
/dev/apex_0
$ lspci
0000:00:00.0 PCI bridge: Broadcom Inc. and subsidiaries Device 2712 (rev 21)
0000:01:00.0 System peripheral: Global Unichip Corp. Coral Edge TPU
0001:00:00.0 PCI bridge: Broadcom Inc. and subsidiaries Device 2712 (rev 21)
0001:01:00.0 Ethernet controller: Device 1de4:0001
$ lspci -nn | grep 089a
0000:01:00.0 System peripheral [0880]: Global Unichip Corp. Coral Edge TPU [1ac1:089a]

udev rules on host: KERNEL=="apex_0", MODE="0666"

$ ls -l /dev/apex_0
crw-rw-rw- 1 root root 120, 0 Feb 27 16:17 /dev/apex_0

I am running frigate on a RaspberryPie 5 with m.2 coral edge tpu and a pie HAT (link). I tried using a 33W GAN power supply as well as a Mac PD 65W power brick.

As illustrated above - the host is able to see the coral, but the frigate container cannot.

Would love to get some info on how to get Frigate to recognise the coral.

Thanks!
Sefi

Version

0.13.2-6476F8A

Frigate config file

mqtt:
  enabled: True
  host: 192.168.x.x
  port: 1883

cameras:
  Front-Doorbell: # <------ Name the camera
    ffmpeg:
      inputs:
        - path: rtsp:https://admin:[email protected]:554/h264Preview_01_sub # <----- The stream you want to use for detection
          roles:
            - detect
        - path: rtsp:https://admin:[email protected]:554/h264Preview_01_main # <----- The stream you want to use for detection
          roles:
            - record
      output_args:
        record: preset-record-generic-audio-aac
    objects:
      track:
        - person
    detect:
      enabled: True # <---- disable detection until you have a working camera feed
      width: 640 # <---- update for your camera's resolution
      height: 480 # <---- update for your camera's resolution
      fps: 5
    record: # <----- Enable recording
      enabled: True
      retain:
        days: 0
        # mode: motion
      events:
        retain:
          default: 3
    snapshots: # <----- Enable snapshots
      enabled: True
      bounding_box: true
      retain:
        default: 1
        objects:
          person: 1

detectors:
  coral1:
    type: edgetpu
    device: pci:0

objects:
  track:
    - person

record:
  # Optional: Enable recording (default: shown below)
  # WARNING: If recording is disabled in the config, turning it on via
  #          the UI or MQTT later will have no effect.
  enabled: True
  # Optional: Number of minutes to wait between cleanup runs (default: shown below)
  # This can be used to reduce the frequency of deleting recording segments from disk if you want to minimize i/o
  expire_interval: 60
  # Optional: Retention settings for recording
  retain:
    # Optional: Number of days to retain recordings regardless of events (default: shown below)
    #NOTE: This should be set to 0 and retention should be defined in events section below
    #       if you only want to retain recordings of events.
    days: 0
    # Optional: Mode for retention. Available options are: all, motion, and active_objects
    #   all - save all recording segments regardless of activity
    #   motion - save all recordings segments with any detected motion
    #   active_objects - save all recording segments with active/moving objects
    #NOTE: this mode only applies when the days setting above is greater than 0
    mode: all
  # Optional: Event recording settings
  events:
    # Optional: Number of seconds before the event to include (default: shown below)
    pre_capture: 5
    # Optional: Number of seconds after the event to include (default: shown below)
    post_capture: 5
    # Optional: Objects to save recordings for. (default: all tracked objects)
    objects:
      - person
    # Optional: Restrict recordings to objects that entered any of the listed zones (default: no required zones)
    required_zones: []
    # Optional: Retention settings for recordings of events
    retain:
      # Required: Default retention days (default: shown below)
      default: 10
      # Optional: Mode for retention. (default: shown below)
      #   all - save all recording segments for events regardless of activity
      #   motion - save all recordings segments for events with any detected motion
      #   active_objects - save all recording segments for event with active/moving objects
      #
      #NOTE: If the retain mode for the camera is more restrictive than the mode configured
      #       here, the segments will already be gone by the time this mode is applied.
      #       For example, if the camera retain mode is "motion", the segments without motion are
      #       never stored, so setting the mode to "all" here won't bring them back.
      mode: motion
      # Optional: Per object retention days
      objects:
        person: 15

logger:
  # Optional: Default log verbosity (default: shown below)
  default: info
  # Optional: Component specific logger overrides
  logs:
    frigate.event: debug

### docker-compose file or Docker CLI command

```yaml
docker run -d \
  --name frigate \
  --restart=unless-stopped \
  --mount type=tmpfs,target=/tmp/cache,tmpfs-size=1000000000 \
  --device /dev/apex_0:/dev/apex_0 \
  --shm-size=512m \
  -v /frigate/storage:/media/frigate \
  -v /frigate/config.yml:/config/config.yml \
  -v /etc/localtime:/etc/localtime:ro \
  -p 5000:5000 \
  -p 8555:8555/tcp \
  -p 8555:8555/udp \
  ghcr.io/blakeblackshear/frigate:stable

### Relevant log output

```shell
I am getting this error in frigate logs:

2024-02-27 19:18:11.385091339  Process detector:coral1:
2024-02-27 19:18:11.385107635  [2024-02-27 19:18:11] frigate.detectors.plugins.edgetpu_tfl ERROR   : No EdgeTPU was detected. If you do not have a Coral device yet, you must configure CPU detectors.
2024-02-27 19:18:11.392092061  Traceback (most recent call last):
2024-02-27 19:18:11.392097469    File "/usr/lib/python3/dist-packages/tflite_runtime/interpreter.py", line 160, in load_delegate
2024-02-27 19:18:11.392098709      delegate = Delegate(library, options)
2024-02-27 19:18:11.392121913    File "/usr/lib/python3/dist-packages/tflite_runtime/interpreter.py", line 119, in __init__
2024-02-27 19:18:11.392123061      raise ValueError(capture.message)
2024-02-27 19:18:11.392123765  ValueError
2024-02-27 19:18:11.392257191  
2024-02-27 19:18:11.392260024  During handling of the above exception, another exception occurred:
2024-02-27 19:18:11.392260654  
2024-02-27 19:18:11.392261543  Traceback (most recent call last):
2024-02-27 19:18:11.392275135    File "/usr/lib/python3.9/multiprocessing/process.py", line 315, in _bootstrap
2024-02-27 19:18:11.392275931      self.run()
2024-02-27 19:18:11.392277006    File "/usr/lib/python3.9/multiprocessing/process.py", line 108, in run
2024-02-27 19:18:11.392277876      self._target(*self._args, **self._kwargs)
2024-02-27 19:18:11.392285894    File "/opt/frigate/frigate/object_detection.py", line 102, in run_detector
2024-02-27 19:18:11.392305061      object_detector = LocalObjectDetector(detector_config=detector_config)
2024-02-27 19:18:11.392307191    File "/opt/frigate/frigate/object_detection.py", line 53, in __init__
2024-02-27 19:18:11.392308654      self.detect_api = create_detector(detector_config)
2024-02-27 19:18:11.392309802    File "/opt/frigate/frigate/detectors/__init__.py", line 18, in create_detector
2024-02-27 19:18:11.392310746      return api(detector_config)
2024-02-27 19:18:11.392331506    File "/opt/frigate/frigate/detectors/plugins/edgetpu_tfl.py", line 41, in __init__
2024-02-27 19:18:11.392333987      edge_tpu_delegate = load_delegate("libedgetpu.so.1.0", device_config)
2024-02-27 19:18:11.392335228    File "/usr/lib/python3/dist-packages/tflite_runtime/interpreter.py", line 162, in load_delegate
2024-02-27 19:18:11.392336246      raise ValueError('Failed to load delegate from {}\n{}'.format(
2024-02-27 19:18:11.392337228  ValueError: Failed to load delegate from libedgetpu.so.1.0

### Operating system

Debian

### Install method

Docker CLI

### Coral version

PCIe

### Any other information that may be helpful

I chose the pcie coral version, because even though it is an M.2 coral edge tpu, it is connected to the host rpi on pcie with the HAT as linked above.

[NickM-27]

what version is the kernel

[NickM-27]

It may be an issue related to this google-coral/edgetpu#808

[sefininio]

I'm running raspberry pi official 64bit os (based on debian 12 i think), the ticket you reference is on ubuntu 23.10.
Besides, like I said, the host is able to communicate with /dev/apex_0, and it is mounted to the frigate container.

Are there any tests you would like me to run on the host or inside the container to try and pinpoint the cause?

[NickM-27]

either way, the issue is with linux kernel 6.5.0 which is the main point. Some users have had similar behavior where the coral was visible but didn't work until the driver was fixed (but again this applies to newer kernel).

Otherwise, you could try running a demo coral docker container test and see if that works

[sefininio]

$ hostnamectl
 Static hostname: frigate
       Icon name: computer
      Machine ID: d01541e808d3423490d4bc8a11c4a842
         Boot ID: 1ee1fd3a7e894dbb91e85ee0901e4240
Operating System: Debian GNU/Linux 12 (bookworm)
          Kernel: Linux 6.1.0-rpi8-rpi-v8
    Architecture: arm64
    
$ lsb_release -a
No LSB modules are available.
Distributor ID:	Debian
Description:	Debian GNU/Linux 12 (bookworm)
Release:	12
Codename:	bookworm    
    ```
    
    

[sefininio]

i'll try with pycoral to try and pinpoint

[sefininio]

OK, I found the problem and now I see in the logs:

2024-03-01 19:21:06.648956600  [2024-03-01 19:21:06] frigate.detectors.plugins.edgetpu_tfl INFO    : Attempting to load TPU as pci:0
2024-03-01 19:21:06.656609916  [2024-03-01 19:21:06] frigate.detectors.plugins.edgetpu_tfl INFO    : TPU found

However, In system, hardware acceleration is not setup:

in config, I have detectors setup

detectors:
  coral1:
    type: edgetpu
    device: pci:0

[NickM-27]

you are ocnfusing detectors and hwaccel, these are two different things