ONNX detector specify GPU#

[numminorih-git]

Dear community,

please advise, can I specify which GPU (I have multiple nvidia in server) onnx detector should use. For example, in tensorrt detector i can set "device: 1" or "device: 0" and detector's process will run on indicated videocard.

[NickM-27]

Currently no, I will add this

[NickM-27]

implemented in #14898

[numminorih-git]

ghcr.io/blakeblackshear/frigate:0.15.0-beta2-tensorrt
specify device: 1 with onnx detector doesn't work, and cause error in config:

Config Error:

Traceback (most recent call last):
  File "/opt/frigate/frigate/api/app.py", line 187, in config_save
    FrigateConfig.parse_yaml(new_config)
  File "/opt/frigate/frigate/config/config.py", line 696, in parse_yaml
    return cls.parse(config_yaml, is_json=False, **context)
  File "/opt/frigate/frigate/config/config.py", line 692, in parse
    return cls.parse_object(config, **context)
  File "/opt/frigate/frigate/config/config.py", line 702, in parse_object
    return cls.model_validate(
  File "/usr/local/lib/python3.9/dist-packages/pydantic/main.py", line 568, in model_validate
    return cls.__pydantic_validator__.validate_python(
pydantic_core._pydantic_core.ValidationError: 1 validation error for FrigateConfig
onnx.device
  Input should be a valid string [type=string_type, input_value=1, input_type=int]
    For further information visit https://errors.pydantic.dev/2.8/v/string_type

detectors:
  onnx:
    type: onnx
    device: 1   #This is the default, select the first GPU
model:
 model_type: yolonas
 width: 320 # <--- should match whatever was set in notebook
 height: 320 # <--- should match whatever was set in notebook
 input_pixel_format: bgr
 path: /config/yolo_nas_s.onnx
 input_tensor: nchw
 labelmap_path: /labelmap/coco-80.txt

[NickM-27]

You need quotes around it

[numminorih-git]

Yes, thx!
device: "1" - work
but still absent in docs (https://deploy-preview-13787--frigate-docs.netlify.app/)