[Detector Support]: OpenVino supporto of Intel Celeron J4125

[pilot1981]

Describe the problem you are having

Hi,
I know this isn't fully supported from OpenVino, but I read many users are able to use Intel Celeron J4125 as detector...can you sahre correct configuration to use as OpenVino detector?

Version

0.13.2

Frigate config file

detectors:
  ov:
    type: openvino
    device: AUTO
    model:
      path: /openvino-model/ssdlite_mobilenet_v2.xml

docker-compose file or Docker CLI command

NA

Relevant log output

2024-02-29 08:47:42.992443994  [2024-02-29 09:47:42] frigate.detectors.plugins.openvino INFO    : Model Input Shape: [1,300,300,3]
2024-02-29 08:47:42.992450751  [2024-02-29 09:47:42] frigate.detectors.plugins.openvino INFO    : Model Output-0 Shape: [1,1,100,7]
2024-02-29 08:47:42.992452983  [2024-02-29 09:47:42] frigate.detectors.plugins.openvino INFO    : Model has 1 Output Tensors
2024-02-29 08:47:43.304936187  Process detector:ov:
2024-02-29 08:47:43.364493107  Traceback (most recent call last):
2024-02-29 08:47:43.364539622    File "/usr/lib/python3.9/multiprocessing/process.py", line 315, in _bootstrap
2024-02-29 08:47:43.364541846      self.run()
2024-02-29 08:47:43.364562753    File "/usr/lib/python3.9/multiprocessing/process.py", line 108, in run
2024-02-29 08:47:43.364564799      self._target(*self._args, **self._kwargs)
2024-02-29 08:47:43.364566554    File "/opt/frigate/frigate/object_detection.py", line 125, in run_detector
2024-02-29 08:47:43.364568214      detections = object_detector.detect_raw(input_frame)
2024-02-29 08:47:43.364588000    File "/opt/frigate/frigate/object_detection.py", line 75, in detect_raw
2024-02-29 08:47:43.364590178      return self.detect_api.detect_raw(tensor_input=tensor_input)
2024-02-29 08:47:43.364591941    File "/opt/frigate/frigate/detectors/plugins/openvino.py", line 84, in detect_raw
2024-02-29 08:47:43.364593322      infer_request.infer([tensor_input])
2024-02-29 08:47:43.364614160    File "/usr/local/lib/python3.9/dist-packages/openvino/runtime/ie_api.py", line 152, in infer
2024-02-29 08:47:43.364616420      return super().infer(normalize_inputs(self, dict(enumerate(inputs))))
2024-02-29 08:47:43.364636417  RuntimeError: The input blob size is not equal to the network input size: got 307200 expecting 270000
2024-02-29 08:48:01.908728107  [2024-02-29 09:48:01] frigate.watchdog               INFO    : Detection appears to be stuck. Restarting detection process...
2024-02-29 08:48:01.938298601  [2024-02-29 09:48:01] detector.ov                    INFO    : Starting detection process: 308
2024-02-29 08:48:02.782798144  [2024-02-29 09:48:02] frigate.detectors.plugins.openvino INFO    : Model Input Shape: [1,300,300,3]
2024-02-29 08:48:02.782804997  [2024-02-29 09:48:02] frigate.detectors.plugins.openvino INFO    : Model Output-0 Shape: [1,1,100,7]
2024-02-29 08:48:02.782807327  [2024-02-29 09:48:02] frigate.detectors.plugins.openvino INFO    : Model has 1 Output Tensors
2024-02-29 08:48:02.790809887  Process detector:ov:
2024-02-29 08:48:02.793788032  Traceback (most recent call last):
2024-02-29 08:48:02.794030922    File "/usr/lib/python3.9/multiprocessing/process.py", line 315, in _bootstrap
2024-02-29 08:48:02.794033810      self.run()
2024-02-29 08:48:02.794092219    File "/usr/lib/python3.9/multiprocessing/process.py", line 108, in run
2024-02-29 08:48:02.794094634      self._target(*self._args, **self._kwargs)
2024-02-29 08:48:02.794743239    File "/opt/frigate/frigate/object_detection.py", line 125, in run_detector
2024-02-29 08:48:02.794747446      detections = object_detector.detect_raw(input_frame)
2024-02-29 08:48:02.794749338    File "/opt/frigate/frigate/object_detection.py", line 75, in detect_raw
2024-02-29 08:48:02.794751037      return self.detect_api.detect_raw(tensor_input=tensor_input)
2024-02-29 08:48:02.794752897    File "/opt/frigate/frigate/detectors/plugins/openvino.py", line 84, in detect_raw
2024-02-29 08:48:02.794811066      infer_request.infer([tensor_input])
2024-02-29 08:48:02.794813540    File "/usr/local/lib/python3.9/dist-packages/openvino/runtime/ie_api.py", line 152, in infer
2024-02-29 08:48:02.794816027      return super().infer(normalize_inputs(self, dict(enumerate(inputs))))
2024-02-29 08:48:02.794818372  RuntimeError: The input blob size is not equal to the network input size: got 307200 expecting 270000
2024-02-29 08:48:11.924904917  [2024-02-29 09:48:11] frigate.watchdog               INFO    : Detection appears to have stopped. Exiting Frigate...

Operating system

HassOS

Install method

HassOS Addon

Coral version

Other

Any other information that may be helpful

I'm trying to use OpenVINO detector for only 1 webcam using CPU Intel Celeron J4125

[NickM-27]

you need to follow the full openvino config https://docs.frigate.video/configuration/object_detectors#openvino-detector

[pilot1981]

not compatible with Intel Celeron J4125 + Intel HID Graphics 600

[tonio757]

OpenVino works fine on J4125 in GPU mode - Just need to use a linux kernel image 5.x. on the host system.
Will not work on 6.x kernels.
Currently running Debian bookworm / ha supervised / frigate 0.15 with a 5.15 kernel without issues - inference 26-28ms