[Detector Support]: OpenVINO multiprocessing crash TypeError model_path NoneType on Intel N150 with Frigate 0.17.0 (config parsed OK, models present, iGPU exposed)

[loloviolo]

Describe the problem you are having

Frigate 0.17.0 crashes systematically when starting the OpenVINO multiprocessing detector:

Process frigate.detector:ov:
Traceback (most recent call last):
File "/usr/lib/python3.11/multiprocessing/process.py", line 314, in _bootstrap
...
File "/opt/frigate/detectors/detection_runners.py", line 299, in init
if not os.path.isfile(model_path):
TypeError: stat: path should be string, bytes, os.PathLike or integer, not NoneType

The YAML config is parsed correctly (python -c "import yaml; print(yaml.safe_load(open('/config/config.yml'))['detectors']['ov'])"), the model files are present, and the GPU is correctly exposed via /dev/dri, but model_path becomes None in the child process.

Frigate becomes unresponsive (UI returns 502 errors), and the watchdog kills the FFmpeg processes for the cameras after ~20 seconds. As a workaround, using a CPU detector (cpu2 with num_threads: 2) keeps the system stable at around 20% total CPU usage for 4 cameras.

Hardware

• Intel N150 mini‑PC (4c/4t, UHD Graphics iGPU)
• RAM: 12 GB, SSD: 512 GB NVMe
• Host OS: Debian 12 (bare metal or Proxmox LXC)
• Docker: device /dev/dri:/dev/dri:ro exposed to the container

vainfo | grep -i intel:
libva info: VA-API version 1.22.0
...
Intel iHD driver for Intel(R) Gen Graphics - 24.3.4 ()

ls -la /dev/dri/card* /dev/dri/renderD* (inside container):
crw-rw---- 1 root video 226, 0 ... /dev/dri/card0
crw-rw---- 1 root 992 226,128 ... /dev/dri/renderD128

The iGPU is detected and working, but the crash still occurs whether device: is set to AUTO, GPU, or CPU.

Software

• Frigate: ghcr.io/blakeblackshear/frigate:0.17.0-f0d69f7
• Migration from Frigate 0.16 to 0.17 completed successfully (new tables such as trigger, password_changed_at, etc. created).
• same error before migration

Version

frigate:0.17.0-f0d69f7

Frigate config file

detectors:
  ov:
    type: openvino
    device: AUTO  # I also tried GPU and CPU
    model:
      path: /openvino-model/ssdlite_mobilenet_v2.xml
      labelmap_path: /openvino-model/coco_91cl_bkgr.txt
      width: 300
      height: 300
      input_tensor: nhwc
      input_pixel_format: bgr

docker-compose file or Docker CLI command

services:
  frigate:
    container_name: frigate
    privileged: true
    restart: unless-stopped
    image: ghcr.io/blakeblackshear/frigate:stable
    shm_size: "512mb" 
    devices:
      - /dev/dri/renderD128:/dev/dri/renderD128
    volumes:
      - /etc/localtime:/etc/localtime:ro
      - ./config:/config
      - /tmp/cache:/tmp/cache
      - ./storage:/media/frigate
    ports:
      - "5000:5000"
      - "8971:8971"
      - "8554:8554"
      - "8555:8555/tcp"
      - "8555:8555/udp"

Relevant Frigate log output

2026-03-06 12:49:13.205656120  Process frigate.detector:ov:
2026-03-06 12:49:13.205659839  Traceback (most recent call last):
2026-03-06 12:49:13.205661284    File "/usr/lib/python3.11/multiprocessing/process.py", line 314, in _bootstrap
2026-03-06 12:49:13.205665128      self.run()
2026-03-06 12:49:13.205667136    File "/opt/frigate/frigate/object_detection/base.py", line 143, in run
2026-03-06 12:49:13.205668568      object_detector = LocalObjectDetector(detector_config=self.detector_config)
2026-03-06 12:49:13.205686438                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2026-03-06 12:49:13.205688399    File "/opt/frigate/frigate/object_detection/base.py", line 62, in __init__
2026-03-06 12:49:13.205689571      self.detect_api = create_detector(detector_config)
2026-03-06 12:49:13.205690511                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2026-03-06 12:49:13.205711770    File "/opt/frigate/frigate/detectors/__init__.py", line 18, in create_detector
2026-03-06 12:49:13.205712834      return api(detector_config)
2026-03-06 12:49:13.205713827             ^^^^^^^^^^^^^^^^^^^^
2026-03-06 12:49:13.205715045    File "/opt/frigate/frigate/detectors/plugins/openvino.py", line 45, in __init__
2026-03-06 12:49:13.205732623      self.runner = OpenVINOModelRunner(
2026-03-06 12:49:13.205733657                    ^^^^^^^^^^^^^^^^^^^^
2026-03-06 12:49:13.205734817    File "/opt/frigate/frigate/detectors/detection_runners.py", line 299, in __init__
2026-03-06 12:49:13.205735913      if not os.path.isfile(model_path):
2026-03-06 12:49:13.205736813             ^^^^^^^^^^^^^^^^^^^^^^^^^^
2026-03-06 12:49:13.205737809    File "<frozen genericpath>", line 30, in isfile
2026-03-06 12:49:13.205739032  TypeError: stat: path should be string, bytes, os.PathLike or integer, not NoneType

Install method

Docker Compose

Object Detector

OpenVino

Screenshots of the Frigate UI's System metrics pages

None (system crashed)

Any other information that may be helpful

everything working fine with default CPU detector (not openvino)

detectors:
cpu2:
type: cpu

[inkeep-ai-assistant[bot]]

The error you're encountering (TypeError: stat: path should be string, bytes, os.PathLike or integer, not NoneType) indicates that the model_path is becoming None when the OpenVINO detector process starts(1).

Based on the Frigate documentation, your configuration structure is incorrect. The model section should be at the top level of your config, not nested under the detectors.ov section(1)(2).

Here's the corrected configuration:

detectors:
  ov:
    type: openvino
    device: GPU

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

(1)(2)

The model configuration should be a separate top-level section, not nested inside the detector configuration(1)(2). When you nest it incorrectly, the detector process cannot access the model path, resulting in None being passed to os.path.isfile().

Additionally, for your Intel N150 iGPU setup, ensure you're using device: GPU rather than AUTO for best compatibility with OpenVINO on Intel integrated graphics(1).

If you continue experiencing issues after this configuration change, you may need to install the latest Intel Compute Runtime inside the Frigate container, as documented in similar cases with Intel iGPUs(3).

📚 Sources:

1. Supported Hardware
2. Full Reference Config
3. [Bug]: [OpenVino:] Intel compute runtime required for Frigate

---

Have another question? Just tag @inkeep.