[Config Support]: Not using OpenVino, but seeing warning/error

[Calimerorulez]

Describe the problem you are having

Using a Dual Coral PCI-E card, and having an AMD Ryzen 5 5600G CPU, I skipped configuring OpenVino, as my CPU doesn't support it.

But 'm seeing this error/warning at startup

info    | 2025-02-01 21:23:37 | detector.coral_pci                    | Starting detection process: 582
info    | 2025-02-01 21:23:37 | frigate.detectors.plugins.edgetpu_tfl | Attempting to load TPU as pci:0
info    | 2025-02-01 21:23:37 | detector.coral_pci2                   | Starting detection process: 585
info    | 2025-02-01 21:23:37 | frigate.detectors.plugins.edgetpu_tfl | TPU found
info    | 2025-02-01 21:23:38 | frigate.app                           | Embedding process started: 598
info    | 2025-02-01 21:23:37 | frigate.detectors.plugins.edgetpu_tfl | Attempting to load TPU as pci:1
info    | 2025-02-01 21:23:38 | frigate.detectors.plugins.edgetpu_tfl | TPU found
info    | 2025-02-01 21:23:38 | frigate.app                           | Output process started: 636
warning | 2025-02-01 21:23:38 | frigate.util.model                    | OpenVINO failed to build model, using CPU instead: Exception from src/inference/src/cpp/core.cpp:121:
unknown | 2025-02-01 21:23:38 | unknown                               | Check '!m_device_map.empty()' failed at src/plugins/intel_gpu/src/plugin/plugin.cpp:379:
unknown | 2025-02-01 21:23:38 | unknown                               | [GPU] Can't get OPTIMIZATION_CAPABILITIES property as no supported devices found or an error happened during devices query.
unknown | 2025-02-01 21:23:38 | unknown                               | [GPU] Please check OpenVINO documentation for GPU drivers setup guide.
unknown | 2025-02-02 09:49:35 | unknown                               | 25-02-01 21:23:38.900379158
unknown | 2025-02-02 09:49:35 | unknown                               | 25-02-01 21:23:38.900379769
unknown | 2025-02-02 09:49:35 | unknown                               | 25-02-01 21:23:38.900380250

Version

0.15.0-cea210d

Frigate config file

detectors:
  coral_pci:
    type: edgetpu
    device: pci:0
  coral_pci2:
    type: edgetpu
    device: pci:1

model:
  path: plus://*********************************

semantic_search:
  enabled: true
  reindex: false
  model_size: large

Relevant Frigate log output

warning | 2025-02-01 21:23:38 | frigate.util.model | OpenVINO failed to build model, using CPU instead: Exception from src/inference/src/cpp/core.cpp:121:
unknown | 2025-02-01 21:23:38 | unknown            | Check '!m_device_map.empty()' failed at src/plugins/intel_gpu/src/plugin/plugin.cpp:379:
unknown | 2025-02-01 21:23:38 | unknown            | [GPU] Can't get OPTIMIZATION_CAPABILITIES property as no supported devices found or an error happened during devices query.
unknown | 2025-02-01 21:23:38 | unknown            | [GPU] Please check OpenVINO documentation for GPU drivers setup guide.
unknown | 2025-02-02 09:49:35 | unknown            | 25-02-01 21:23:38.900379158
unknown | 2025-02-02 09:49:35 | unknown            | 25-02-01 21:23:38.900379769
unknown | 2025-02-02 09:49:35 | unknown            | 25-02-01 21:23:38.900380250

Relevant go2rtc log output

n/a

Frigate stats

No response

Operating system

Proxmox

Install method

Docker Compose

docker-compose file or Docker CLI command

n/a

Object Detector

Coral

Screenshots of the Frigate UI's System metrics pages

No response

Any other information that may be helpful

No response

[hawkeye217]

You're seeing this because semantic search using the large model will run on GPU. You should set it to small so that it runs on CPU. That should make the error go away.

[Calimerorulez]

The large one. I'm passing my 5600G GPU through to my LXC.

[FredHags]

I'm seeing the same error using a Nvidia GPU.
Also using the large model.

[hawkeye217]

I'm seeing the same error using a Nvidia GPU.

Also using the large model.

See the solution above. 👆🏻

[simondsmason]

@hawkeye217 I am seeing the same error using a Nvidia GPU. According to the previous response "You're seeing this because semantic search using the large model will run on GPU" and the solution is to set it to small model to have it run on the CPU. But don't I want it to run on the GPU since I have one? Or is it that the large option is not compatible with the Nvidia?

[NickM-27]

Are you sure you're using the tensortt docker image variant