[Detector Support]: EdgeTPU not detected in Frigate Docker (PCI / M.2 Coral) – "Failed to load delegate from libedgetpu.so.1.0"

[Th0mas89]

Describe the problem you are having

Sorry, to ask such stupid questions, but I don't have any ideas how to proceed after upgrading from a previous (unknown) version of frigate to the new stable version;

I’m running Frigate v14 in a Docker container with a PCI/M.2 Coral device. On the host, the Coral is properly installed and recognized (e.g., via lspci), and /dev/apex_0 is present. I have set the container to run with privileged: true, and I can see /dev/apex_0 inside the container. However, Frigate still fails to detect the EdgeTPU with the following error:

[2024-12-25 12:47:22] frigate.detectors.plugins.edgetpu_tfl ERROR : No EdgeTPU was detected. If you do not have a Coral device yet, you must configure CPU detectors. ValueError: Failed to load delegate from libedgetpu.so.1.0

What I have tried so far

1. Verified that the device appears in the container (ls /dev/ shows apex_0).
2. Checked that libedgetpu.so.1.0 exists in /usr/lib/x86_64-linux-gnu/.
3. Updated frigate.yml to explicitly use:

   detectors:
     coral:
       type: edgetpu
       device: pci

4. Ran lspci on the host – device is correctly listed as “Google Apex device.”
5. Confirmed the Docker container is privileged and that the appropriate volumes/devices are passed in.
6. I also tried to reinstall the coral drivers.

Do you have any additional hints how to proceed?
Thank you for your help!

Host OS: Ubuntu 22.04
Kernel version: 5.15.0-125-generic
Docker version: 20.10.17, build 100c701
Coral device: M.2 PCIe

Version

0.14.1-f4f3cfa

Frigate config file

mqtt:
  enabled: true
  host: 192.XXX.XXX.XX
  port: 1883
  topic_prefix: frigate
  client_id: frigate

cameras:
  Gartenkamera: # <------ Name the camera
    ffmpeg:
      inputs:
        - path: rtsp://XXXUser:[email protected]:XXXX/stream1 # <----- The stream you want to use for detection
          input_args: preset-rtsp-generic
          roles:
            - detect
            - record
    detect:
      enabled: true # <---- disable detection until you have a working camera feed
      width: 2304 # <---- update for your camera's resolution
      height: 1296 # <---- update for your camera's resolution
birdseye:
  enabled: true
  mode: continuous

detectors:
#  cpu1:
#    type: cpu
  coral:
    type: edgetpu
    device: pci
#    device: /dev/apex_0


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
  retain:
    # Required: Default retention days (default: shown below)
    default: 30
    # Optional: Per object retention days
    objects:
      person: 40

objects:
  # Optional: list of objects to track from labelmap.txt (default: shown below)
  track:
    - person
    - dog
    - cat
motion:
  # Optional: The threshold passed to cv2.threshold to determine if a pixel is different enough to be counted as motion. (default: shown below)
  # Increasing this value will make motion detection less sensitive and decreasing it will make motion detection more sensitive.
  # The value should be between 1 and 255.
  threshold: 35
  # Optional: Minimum size in pixels in the resized motion image that counts as motion (default: 30)
  # Increasing this value will prevent smaller areas of motion from being detected. Decreasing will
  # make motion detection more sensitive to smaller moving objects.
  # As a rule of thumb:
  #  - 15 - high sensitivity
  #  - 30 - medium sensitivity
  #  - 50 - low sensitivity
  contour_area: 30
  # Optional: Alpha value passed to cv2.accumulateWeighted when averaging the motion delta across multiple frames (default: shown below)
  # Higher values mean the current frame impacts the delta a lot, and a single raindrop may register as motion.
  # Too low and a fast moving person wont be detected as motion.
  delta_alpha: 0.2
  # Optional: Alpha value passed to cv2.accumulateWeighted when averaging frames to determine the background (default: shown below)
  # Higher values mean the current frame impacts the average a lot, and a new object will be averaged into the background faster.
  # Low values will cause things like moving shadows to be detected as motion for longer.
  # https://www.geeksforgeeks.org/background-subtraction-in-an-image-using-concept-of-running-average/
  frame_alpha: 0.2
  # Optional: Height of the resized motion frame  (default: 50)
  # This operates as an efficient blur alternative. Higher values will result in more granular motion detection at the expense
  # of higher CPU usage. Lower values result in less CPU, but small changes may not register as motion.
  frame_height: 50
  # Optional: motion mask
  # NOTE: see docs for more detailed info on creating masks
  # Optional: improve contrast (default: shown below)
  # Enables dynamic contrast improvement. This should help improve night detections at the cost of making motion detection more sensitive
  # for daytime.
  improve_contrast: true
  # Optional: Delay when updating camera motion through MQTT from ON -> OFF (default: shown below).
  mqtt_off_delay: 30
  mask:
    - 732,0,712,41,227,113,0,106,0,0

logger:
  default: debug

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: false
  # 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: Sync recordings with disk on startup and once a day (default: shown below).
  sync_recordings: false
  # 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: Recording Export Settings
  export:
    # Optional: Timelapse Output Args (default: shown below).
    # NOTE: The default args are set to fit 24 hours of recording into 1 hour playback.
    # See https://stackoverflow.com/a/58268695 for more info on how these args work.
    # As an example: if you wanted to go from 24 hours to 30 minutes that would be going
    # from 86400 seconds to 1800 seconds which would be 1800 / 86400 = 0.02.
    # The -r (framerate) dictates how smooth the output video is.
    # So the args would be -vf setpts=0.02*PTS -r 30 in that case.
    timelapse_args: -vf setpts=0.04*PTS -r 30
  # Optional: Recording Preview Settings
  preview:
    # Optional: Quality of recording preview (default: shown below).
    # Options are: very_low, low, medium, high, very_high
    quality: medium
version: 0.14

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
    image: ghcr.io/blakeblackshear/frigate:stable
    #image: ghcr.io/blakeblackshear/frigate:0.15.0-beta4-tensorrt
    shm_size: "256mb" # update for your cameras based on calculation above
    devices:
      - /dev/apex_0:/dev/apex_0
    volumes:
      - /path/to/frigate/config_folder/:/config
      - /path/to/frigate/media/:/media/frigate
      - type: tmpfs # Optional: 1GB of memory, reduces SSD/SD Card wear
        target: /tmp/cache
        tmpfs:
          size: 1000000000
    ports:
      - "5000:5000"
      - "8554:8554" # RTSP feeds
      - "8555:8555/tcp" # WebRTC over tcp
      - "8555:8555/udp" # WebRTC over udp
    environment:
      - FRIGATE_RTSP_PASSWORD="<REDACTED>"
    networks:
      - caddy_net

networks:
  caddy_net:
    external: true

Relevant Frigate log output

2024-12-26 09:18:47.068755740  [INFO] Preparing Frigate...
2024-12-26 09:18:47.081471818  [INFO] Starting Frigate...
2024-12-26 09:18:48.419126071  [2024-12-26 09:18:48] frigate.app                    INFO    : Starting Frigate (0.14.1-f4f3cfa)
2024-12-26 09:18:48.419229807  [2024-12-26 09:18:48] frigate.util.config            INFO    : Checking if frigate config needs migration...
2024-12-26 09:18:48.443826745  [2024-12-26 09:18:48] frigate.util.config            INFO    : frigate config does not need migration...
2024-12-26 09:18:48.555327155  [2024-12-26 09:18:48] frigate.util.services          INFO    : Automatically detected vaapi hwaccel for video decoding
2024-12-26 09:18:48.588427558  [2024-12-26 09:18:48] peewee                         DEBUG   : ('CREATE TABLE IF NOT EXISTS "migratehistory" ("id" INTEGER NOT NULL PRIMARY KEY, "name" VARCHAR(255) NOT NULL, "migrated_at" DATETIME NOT NULL)', [])
2024-12-26 09:18:48.589682954  [2024-12-26 09:18:48] peewee                         DEBUG   : ('SELECT "t1"."id", "t1"."name", "t1"."migrated_at" FROM "migratehistory" AS "t1" ORDER BY "t1"."id"', [])
2024-12-26 09:18:48.592514919  [2024-12-26 09:18:48] peewee_migrate.logs            INFO    : Starting migrations
2024-12-26 09:18:48.592604158  [2024-12-26 09:18:48] peewee                         DEBUG   : ('SELECT "t1"."id", "t1"."name", "t1"."migrated_at" FROM "migratehistory" AS "t1" ORDER BY "t1"."id"', [])
2024-12-26 09:18:48.593288234  [2024-12-26 09:18:48] peewee_migrate.logs            INFO    : There is nothing to migrate
2024-12-26 09:18:48.603803118  [2024-12-26 09:18:48] frigate.app                    INFO    : Recording process started: 382
2024-12-26 09:18:48.603808387  [2024-12-26 09:18:48] frigate.app                    INFO    : Recording process started: 384
2024-12-26 09:18:48.606010596  [2024-12-26 09:18:48] asyncio                        DEBUG   : Using selector: EpollSelector
2024-12-26 09:18:48.607741755  [2024-12-26 09:18:48] frigate.app                    INFO    : go2rtc process pid: 97
2024-12-26 09:18:48.615211416  [2024-12-26 09:18:48] frigate.comms.mqtt             DEBUG   : MQTT connected
2024-12-26 09:18:48.631045716  [2024-12-26 09:18:48] detector.cpu1                  INFO    : Starting detection process: 412
2024-12-26 09:18:48.634554465  [2024-12-26 09:18:48] frigate.detectors              WARNING : CPU detectors are not recommended and should only be used for testing or for trial purposes.
2024-12-26 09:18:48.636654653  [2024-12-26 09:18:48] detector.coral                 INFO    : Starting detection process: 414
2024-12-26 09:19:01.492949360  [2024-12-26 09:18:48] frigate.app                    INFO    : Output process started: 418
2024-12-26 09:19:01.493655782  [2024-12-26 09:18:48] tzlocal                        DEBUG   : /etc/timezone found, contents:
2024-12-26 09:19:01.493661388   Etc/UTC
2024-12-26 09:19:06.223911  2024-12-26 09:19:01.493663718
2024-12-26 09:19:01.493666029  Process detector:coral:
2024-12-26 09:19:01.493668865  [2024-12-26 09:18:53] asyncio                        DEBUG   : Using selector: EpollSelector
2024-12-26 09:19:01.493676620  [2024-12-26 09:18:48] peewee.sqliteq                 DEBUG   : received query DELETE FROM "regions" WHERE NOT ("regions"."camera" IN (?))
2024-12-26 09:19:01.493929232  [2024-12-26 09:18:58] asyncio                        DEBUG   : Using selector: EpollSelector
2024-12-26 09:19:01.493934233  [2024-12-26 09:18:48] peewee                         DEBUG   : ('DELETE FROM "regions" WHERE NOT ("regions"."camera" IN (?))', ['Gartenkamera'])
2024-12-26 09:19:01.494120416  [2024-12-26 09:18:48] frigate.detectors.plugins.edgetpu_tfl INFO    : Attempting to load TPU as pci
2024-12-26 09:19:01.494255113  [2024-12-26 09:18:48] peewee                         DEBUG   : ('SELECT "t1"."camera", "t1"."grid", "t1"."last_update" FROM "regions" AS "t1" WHERE ("t1"."camera" = ?) LIMIT ? OFFSET ?', ['Gartenkamera', 1, 0])
2024-12-26 09:19:01.494418919  [2024-12-26 09:19:01] frigate.detectors.plugins.edgetpu_tfl ERROR   : No EdgeTPU was detected. If you do not have a Coral device yet, you must configure CPU detectors.
2024-12-26 09:19:01.494423151  [2024-12-26 09:18:48] tzlocal                        DEBUG   : /etc/localtime found
2024-12-26 09:19:01.494538745  [2024-12-26 09:18:48] tzlocal                        DEBUG   : 2 found:
2024-12-26 09:19:01.494542674   {'/etc/timezone': 'Etc/UTC', '/etc/localtime is a symlink to': 'Etc/UTC'}
2024-12-26 09:19:01.494654464  [2024-12-26 09:18:48] peewee                         DEBUG   : ('SELECT "t1"."id" FROM "event" AS "t1" WHERE ((("t1"."camera" = ?) AND (("t1"."false_positive" IS NULL) OR ("t1"."false_positive" = ?))) AND ("t1"."start_time" > ?))', ['Gartenkamera', False, 0])
2024-12-26 09:19:01.494808917  [2024-12-26 09:18:48] frigate.util.object            DEBUG   : Found 0 new events for Gartenkamera
2024-12-26 09:19:01.494899939  [2024-12-26 09:18:48] frigate.app                    INFO    : Camera processor started for Gartenkamera: 433
2024-12-26 09:19:01.495050701  [2024-12-26 09:18:48] frigate.app                    INFO    : Capture process started for Gartenkamera: 435
2024-12-26 09:19:01.495172647  [2024-12-26 09:18:48] peewee                         DEBUG   : ('SELECT COUNT(?) FROM "recordings" AS "t1" WHERE (("t1"."camera" = ?) AND ("t1"."segment_size" > ?))', ['*', 'Gartenkamera', 0.0])
2024-12-26 09:19:01.495308046  [2024-12-26 09:18:48] urllib3.connectionpool         DEBUG   : Starting new HTTPS connection (1): api.github.com:443
2024-12-26 09:19:01.495334160  Traceback (most recent call last):
2024-12-26 09:19:01.495365592    File "/usr/lib/python3/dist-packages/tflite_runtime/interpreter.py", line 160, in load_delegate
2024-12-26 09:19:01.495368451      delegate = Delegate(library, options)
2024-12-26 09:19:01.495371378    File "/usr/lib/python3/dist-packages/tflite_runtime/interpreter.py", line 119, in __init__
2024-12-26 09:19:01.495373929      raise ValueError(capture.message)
2024-12-26 09:19:01.495390875  ValueError
2024-12-26 09:19:06.223962  2024-12-26 09:19:01.495402504
2024-12-26 09:19:01.495405463  During handling of the above exception, another exception occurred:
2024-12-26 09:19:06.223969  2024-12-26 09:19:01.495407463
2024-12-26 09:19:01.495418783  Traceback (most recent call last):
2024-12-26 09:19:01.495432004  [2024-12-26 09:18:48] peewee                         DEBUG   : ('SELECT AVG("t1"."segment_size" / ("t1"."end_time" - "t1"."start_time")) FROM "recordings" AS "t1" WHERE (("t1"."camera" = ?) AND ("t1"."segment_size" > ?)) LIMIT ?', ['Gartenkamera', 0.0, 100])
2024-12-26 09:19:01.495452275    File "/usr/lib/python3.9/multiprocessing/process.py", line 315, in _bootstrap
2024-12-26 09:19:01.495454838      self.run()
2024-12-26 09:19:01.495457620    File "/usr/lib/python3.9/multiprocessing/process.py", line 108, in run
2024-12-26 09:19:01.495460233      self._target(*self._args, **self._kwargs)
2024-12-26 09:19:01.495476553    File "/opt/frigate/frigate/object_detection.py", line 102, in run_detector
2024-12-26 09:19:01.495479502      object_detector = LocalObjectDetector(detector_config=detector_config)
2024-12-26 09:19:01.495482268    File "/opt/frigate/frigate/object_detection.py", line 53, in __init__
2024-12-26 09:19:01.495484900      self.detect_api = create_detector(detector_config)
2024-12-26 09:19:01.495487671    File "/opt/frigate/frigate/detectors/__init__.py", line 18, in create_detector
2024-12-26 09:19:01.495490016      return api(detector_config)
2024-12-26 09:19:01.495492812    File "/opt/frigate/frigate/detectors/plugins/edgetpu_tfl.py", line 41, in __init__
2024-12-26 09:19:01.495516223      edge_tpu_delegate = load_delegate("libedgetpu.so.1.0", device_config)
2024-12-26 09:19:01.495519809    File "/usr/lib/python3/dist-packages/tflite_runtime/interpreter.py", line 162, in load_delegate
2024-12-26 09:19:01.495547008      raise ValueError('Failed to load delegate from {}\n{}'.format(
2024-12-26 09:19:01.495562151  ValueError: Failed to load delegate from libedgetpu.so.1.0
2024-12-26 09:19:06.223984  2024-12-26 09:19:01.495564463
2024-12-26 09:19:01.495567380  [2024-12-26 09:18:48] frigate.storage                DEBUG   : Gartenkamera has a bandwidth of 0 MiB/hr.
2024-12-26 09:19:01.496873619  [2024-12-26 09:18:48] urllib3.connectionpool         DEBUG   : https://api.github.com:443 "GET /repos/blakeblackshear/frigate/releases/latest HTTP/11" 200 1667
2024-12-26 09:19:01.496880184  [2024-12-26 09:18:48] frigate.api.auth               DEBUG   : Using jwt secret from .jwt_secret file in config directory.
2024-12-26 09:19:01.496884015  [2024-12-26 09:18:48] peewee.sqliteq                 DEBUG   : received query UPDATE "reviewsegment" SET "end_time" = ? WHERE ("reviewsegment"."end_time" IS NULL)
2024-12-26 09:19:01.496887881  [2024-12-26 09:18:48] peewee                         DEBUG   : ('UPDATE "reviewsegment" SET "end_time" = ? WHERE ("reviewsegment"."end_time" IS NULL)', [1735204728.739798])
2024-12-26 09:19:01.496891657  [2024-12-26 09:18:48] peewee.sqliteq                 DEBUG   : received query UPDATE "event" SET "end_time" = ("event"."start_time" + ?) WHERE ("event"."end_time" IS NULL)
2024-12-26 09:19:01.496895456  [2024-12-26 09:18:48] peewee                         DEBUG   : ('UPDATE "event" SET "end_time" = ("event"."start_time" + ?) WHERE ("event"."end_time" IS NULL)', [30])
2024-12-26 09:19:01.496901698  [2024-12-26 09:18:48] peewee                         DEBUG   : ('SELECT COUNT(1) FROM (SELECT 1 FROM "user" AS "t1") AS "_wrapped"', [])
2024-12-26 09:19:01.496939416  [2024-12-26 09:18:51] asyncio                        DEBUG   : Using selector: EpollSelector
2024-12-26 09:19:01.496943170  [2024-12-26 09:18:53] frigate.comms.ws               DEBUG   : Publishing mqtt message from websockets at onConnect.
2024-12-26 09:19:01.496946828  [2024-12-26 09:18:53] frigate.comms.ws               DEBUG   : Publishing mqtt message from websockets at onConnect.
2024-12-26 09:19:03.615607294  [2024-12-26 09:19:03] asyncio                        DEBUG   : Using selector: EpollSelector

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

[NickM-27]

Can you confirm that no other services are using the coral?

[Th0mas89]

Dear Nick, I can confirm that by using sudo lsof /dev/apex_0, sudo fuser /dev/apex_0.

But i also tried this , that gave me a non interpretable output.

Thank you for your help!

journalctl -k | grep apex

Dez 26 14:18:21 homeserver kernel: apex 0000:02:00.0: enabling device (0000 -> 0002)
Dez 26 14:18:34 homeserver kernel: apex 0000:02:00.0: RAM did not enable within timeout (12000 ms)
Dez 26 14:18:39 homeserver kernel: apex 0000:02:00.0: Apex performance not throttled due to temperature
Dez 26 14:18:44 homeserver kernel: apex 0000:02:00.0: Apex performance not throttled due to temperature
Dez 26 14:18:49 homeserver kernel: apex 0000:02:00.0: Apex performance not throttled due to temperature
Dez 26 14:18:54 homeserver kernel: apex 0000:02:00.0: Apex performance not throttled due to temperature
Dez 26 14:18:59 homeserver kernel: apex 0000:02:00.0: Apex performance not throttled due to temperature
Dez 26 14:19:04 homeserver kernel: apex 0000:02:00.0: Apex performance not throttled due to temperature
Dez 26 14:19:10 homeserver kernel: apex 0000:02:00.0: Apex performance not throttled due to temperature
Dez 26 14:19:15 homeserver kernel: apex 0000:02:00.0: Apex performance not throttled due to temperature
Dez 26 14:19:20 homeserver kernel: apex 0000:02:00.0: Apex performance not throttled due to temperature
Dez 26 14:19:25 homeserver kernel: apex 0000:02:00.0: Apex performance not throttled due to temperature
Dez 26 14:19:30 homeserver kernel: apex 0000:02:00.0: Apex performance not throttled due to temperature
Dez 26 14:19:35 homeserver kernel: apex 0000:02:00.0: Apex performance not throttled due to temperature
Dez 26 14:19:40 homeserver kernel: apex 0000:02:00.0: Apex performance not throttled due to temperature
Dez 26 14:19:45 homeserver kernel: apex 0000:02:00.0: Apex performance not throttled due to temperature

[jmtatsch]

From a quick look on my working docker compose setup i found the following differences:
devices for me doesnt need the apex device
however under volumes the device is passed through
volumes:
- "/dev/bus/usb:/dev/bus/usb"
- "/etc/localtime:/etc/localtime:ro"

[Th0mas89]

I think the device is mounted properly, but at this point I am not able to debug any further, as I have no knowledge in this area.

I tried to run the pycoral tests in the docker environment to test, wether the recommended test procedures worked. I can see, that the edgetpu is also connected

root@2a3d68db7707:/opt/frigate# python3
Python 3.9.2 (default, Feb 28 2021, 17:03:44) 
[GCC 10.2.1 20210110] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> from pycoral.utils import edgetpu
>>> from pycoral.pybind._pywrap_coral import SetVerbosity as set_verbosity
>>> from pycoral.pybind._pywrap_coral import ListEdgeTpus as list_edge_tpus
>>> list_edge_tpus()
[{'type': 'pci', 'path': '/dev/apex_0'}]
>>> from pycoral.utils.dataset import read_label_file
>>> from pycoral.utils.edgetpu import make_interpreter
>>> from pycoral.adapters import common
>>> from pycoral.adapters import classify
>>> from PIL import Image
>>> import numpy as np
>>> import time
>>> import argparse
>>> labels = read_label_file("/media/frigate/coral/pycoral/test_data/inat_bird_labels.txt")
>>> interpreter = make_interpreter(["/media/frigate/coral/pycoral/test_data/mobilenet_v2_1.0_224_inat_bird_quant_edgetpu.tflite"])
Traceback (most recent call last):
  File "/usr/lib/python3/dist-packages/tflite_runtime/interpreter.py", line 160, in load_delegate
    delegate = Delegate(library, options)
  File "/usr/lib/python3/dist-packages/tflite_runtime/interpreter.py", line 119, in __init__
    raise ValueError(capture.message)
ValueError

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/lib/python3/dist-packages/pycoral/utils/edgetpu.py", line 87, in make_interpreter
    delegates = [load_edgetpu_delegate({'device': device} if device else {})]
  File "/usr/lib/python3/dist-packages/pycoral/utils/edgetpu.py", line 52, in load_edgetpu_delegate
    return tflite.load_delegate(_EDGETPU_SHARED_LIB, options or {})
  File "/usr/lib/python3/dist-packages/tflite_runtime/interpreter.py", line 162, in load_delegate
    raise ValueError('Failed to load delegate from {}\n{}'.format(
ValueError: Failed to load delegate from libedgetpu.so.1

>>> set_verbosity(10)
True
>>> interpreter = make_interpreter(["/media/frigate/coral/pycoral/test_data/mobilenet_v2_1.0_224_inat_bird_quant_edgetpu.tflite"])
I tflite/edgetpu_manager_direct.cc:453] No matching device is already opened for shared ownership.
I driver/usb/local_usb_device.cc:944] EnumerateDevices: vendor:0x1a6e, product:0x89a
I driver/usb/local_usb_device.cc:979] EnumerateDevices: checking bus[1] port[4]
I driver/usb/local_usb_device.cc:979] EnumerateDevices: checking bus[4] port[0]
I driver/usb/local_usb_device.cc:979] EnumerateDevices: checking bus[1] port[1]
I driver/usb/local_usb_device.cc:979] EnumerateDevices: checking bus[3] port[4]
I driver/usb/local_usb_device.cc:979] EnumerateDevices: checking bus[1] port[0]
I driver/usb/local_usb_device.cc:979] EnumerateDevices: checking bus[2] port[1]
I driver/usb/local_usb_device.cc:979] EnumerateDevices: checking bus[2] port[0]
I driver/usb/local_usb_device.cc:979] EnumerateDevices: checking bus[3] port[0]
I driver/usb/local_usb_device.cc:944] EnumerateDevices: vendor:0x18d1, product:0x9302
I driver/usb/local_usb_device.cc:979] EnumerateDevices: checking bus[1] port[4]
I driver/usb/local_usb_device.cc:979] EnumerateDevices: checking bus[4] port[0]
I driver/usb/local_usb_device.cc:979] EnumerateDevices: checking bus[1] port[1]
I driver/usb/local_usb_device.cc:979] EnumerateDevices: checking bus[3] port[4]
I driver/usb/local_usb_device.cc:979] EnumerateDevices: checking bus[1] port[0]
I driver/usb/local_usb_device.cc:979] EnumerateDevices: checking bus[2] port[1]
I driver/usb/local_usb_device.cc:979] EnumerateDevices: checking bus[2] port[0]
I driver/usb/local_usb_device.cc:979] EnumerateDevices: checking bus[3] port[0]
I tflite/edgetpu_context_direct.cc:106] USB always DFU: False (default)
I tflite/edgetpu_context_direct.cc:128] USB bulk-in queue capacity: default
I tflite/edgetpu_context_direct.cc:67] Performance expectation: Max (default)
I ./driver/mmio/host_queue.h:266] Starting in normal mode
I driver/kernel/kernel_registers.cc:83] Opening /dev/apex_0. read_only=0
I tflite/edgetpu_context_direct.cc:401] Failed to open device [Apex (PCIe)] at [/dev/apex_0]: Failed precondition: Device open failed : -1 (Connection timed out)
Traceback (most recent call last):
  File "/usr/lib/python3/dist-packages/tflite_runtime/interpreter.py", line 160, in load_delegate
    delegate = Delegate(library, options)
  File "/usr/lib/python3/dist-packages/tflite_runtime/interpreter.py", line 119, in __init__
    raise ValueError(capture.message)
ValueError

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/lib/python3/dist-packages/pycoral/utils/edgetpu.py", line 87, in make_interpreter
    delegates = [load_edgetpu_delegate({'device': device} if device else {})]
  File "/usr/lib/python3/dist-packages/pycoral/utils/edgetpu.py", line 52, in load_edgetpu_delegate
    return tflite.load_delegate(_EDGETPU_SHARED_LIB, options or {})
  File "/usr/lib/python3/dist-packages/tflite_runtime/interpreter.py", line 162, in load_delegate
    raise ValueError('Failed to load delegate from {}\n{}'.format(
ValueError: Failed to load delegate from libedgetpu.so.1

Also adding

options vfio-pci ids=1ac1:089a disable_idle_d3=1

to /etc/modprobe.d/tpu.conf on the host, as mentioned in this reddit post did not help https://www.reddit.com/r/VFIO/comments/l5awg0/using_google_coral_mpcie_tpu_in_qemu_vm/

Best Thomas