[Detector Support]: 0.14 - OVNO Documentation?

[HomeAssistantNewbie]

Describe the problem you are having

Hi, so updated to 0.14 Frigate (Running as Addon on HAOS) and it kept crashing. Managed to narrow it down to a model failure; looked around on release notes but didn't see anything beyond "updated to new Openvino".

Looking at the detector documentation (https://docs.frigate.video/configuration/object_detectors) it's the same so probably deprecated. Looking at the error in the logs... is a GPU mandatory perhaps?

Version

0.14.0

Frigate config file

detectors:
  ov:
    type: openvino
    device: AUTO

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

docker-compose file or Docker CLI command

-

Relevant log output

2024-08-08 21:28:27.365304486  Process detector:ov:
2024-08-08 21:28:27.366622023  Traceback (most recent call last):
2024-08-08 21:28:27.366639235    File "/usr/lib/python3.9/multiprocessing/process.py", line 315, in _bootstrap
2024-08-08 21:28:27.366640295      self.run()
2024-08-08 21:28:27.369538490    File "/usr/lib/python3.9/multiprocessing/process.py", line 108, in run
2024-08-08 21:28:27.369541581      self._target(*self._args, **self._kwargs)
2024-08-08 21:28:27.369543001    File "/opt/frigate/frigate/object_detection.py", line 102, in run_detector
2024-08-08 21:28:27.369544401      object_detector = LocalObjectDetector(detector_config=detector_config)
2024-08-08 21:28:27.369547963    File "/opt/frigate/frigate/object_detection.py", line 53, in __init__
2024-08-08 21:28:27.369558781      self.detect_api = create_detector(detector_config)
2024-08-08 21:28:27.369560559    File "/opt/frigate/frigate/detectors/__init__.py", line 18, in create_detector
2024-08-08 21:28:27.369562145      return api(detector_config)
2024-08-08 21:28:27.369563533    File "/opt/frigate/frigate/detectors/plugins/openvino.py", line 43, in __init__
2024-08-08 21:28:27.369565076      self.interpreter = self.ov_core.compile_model(
2024-08-08 21:28:27.369566592    File "/usr/local/lib/python3.9/dist-packages/openvino/runtime/ie_api.py", line 521, in compile_model
2024-08-08 21:28:27.369580669      super().compile_model(model, device_name, {} if config is None else config),
2024-08-08 21:28:27.369582043  RuntimeError: Exception from src/inference/src/cpp/core.cpp:123:
2024-08-08 21:28:27.369591473  Exception from src/inference/src/dev/plugin.cpp:59:
2024-08-08 21:28:27.369592988  Check 'contexts.count(device_id)' failed at src/plugins/intel_gpu/src/plugin/plugin.cpp:204:
2024-08-08 21:28:27.369594524  [GPU] Context was not initialized for 0 device

Operating system

HassOS

Install method

HassOS Addon

Object Detector

OpenVino

Any other information that may be helpful

No response

[NickM-27]

Change auto to cpu

[HomeAssistantNewbie]

Pretty obvious retrospectively... thanks Nick!

[rykr]

if it's using quicksync should it not be GPU? Mine seems to be failing as well and I have mine set to GPU

[NickM-27]

That usually means the GPU is not mapped in correctly to the docker container

[rykr]

I think I fixed it. had my renderD128 not mapped in right. my container is using gpus=all so it is reporting against the 1060 in there as well but I'm not using in trt-models.