Frigate doesn‘t detect my Coral USB TPU?!?

mime270 · September 8, 2022, 10:58am

Hello,

a few days ago i installed Home Assistant OS with frigate on a RasPi 4B 4GB. Everything works great but the Coral USB TPU could not be detected. Here is what i‘ve tried:

reboot
changing the usb-port

There is no other device connected to the raspi, the white led on the TPU is on.

I use the same system in my house allready. Here everything works fine.

Thanks for some help.

mime270

crzynik · September 8, 2022, 12:45pm

We always recommend using a powered USB hub as the coral is very picky about power.

In any case, which addon are you using? Some users have found that they need to run the full access addon with protection mode disabled for the coral to get picked up.

Greg.o · October 24, 2022, 1:38pm

Hi,

I’m jumping in the discussion because I have same problem. I’ve got an USB Coral edgetpu. I used to use it on HA on a NUC, but it went down and I had to replace it.

I now try to reinstall Frigate and coral on a NUC11ATKC2.

I run HA Core 2022.10.5 and OS version 9.2

I tried to load the last backup I made from preceding NUC (frigate and coral were perfectly operationnals). I can’t get Frigate detect Coral on USB.

I tried different USB ports, with or without a powered USB hub, with different USB cables… Nothing works.

I wonder if there would be a hardware failure in the Coral. Do you know how I could sort it out ?

crzynik · October 24, 2022, 3:16pm

See [Support]: Updated to 11.0 - Now crashes on trying to get TPU · Issue #3890 · blakeblackshear/frigate · GitHub

Greg.o · October 25, 2022, 5:45pm

Hi,

thx a lot, my coral is now detected. I still have to understand why my CPU load is over 60%, even with the edgeTPU.

crzynik · October 25, 2022, 6:08pm

github.com/blakeblackshear/frigate

⚡I have a Coral, but my CPU usage is still high

opened 11:14PM - 16 Sep 22 UTC

closed 11:15PM - 16 Sep 22 UTC

NickM-27

documentation

## Common config issues causing higher than expected CPU load: We get this qu…estion pretty often. While object detection is a considerable load on Frigate, there are other loads as well: - Decoding the video stream to run motion & object detection - Other modifications like resizing or filtering the stream (which runs on the CPU) - Running motion detection (which happens on the CPU) ## Using hwaccel to decode the stream is always highly recommended: As is explained [in the docs](https://docs.frigate.video/hardware#do-hwaccel-args-help-if-i-am-using-a-coral) decompressing video streams takes a significant amount of CPU power. Video compression uses key frames (also known as I-frames) to send a full frame in the video stream. The following frames only include the difference from the key frame, and the CPU has to compile each frame by merging the differences with the key frame. [More detailed explanation](https://blog.video.ibm.com/streaming-video-tips/keyframes-interframe-video-compression/). Higher resolutions and frame rates mean more processing power is needed to decode the video stream, so try and set them on the camera to avoid unnecessary decoding work. Please see [the hardware acceleration docs](https://docs.frigate.video/configuration/hardware_acceleration) for how to setup hardware acceleration for your GPU. ## Poorly configured camera detect config: [It is important to properly set your cameras detect config](https://docs.frigate.video/guides/camera_setup#choosing-a-detect-resolution) Frigate will resize the frames from the decoded camera stream to whatever is set in `detect -> width / height` unless it is the same size as the actual stream. This means if you camera is `1280 x 960` and your detect config is: ```yaml detect: width: 1080 height: 720 ``` then frigate will resize the frames to `1080 x 720` which will use a non-negligible amount of CPU to do. This is why it is recommended to run detect on the actual size of your stream. **NOTE:** The default detect config is 1080 x 720 so you always need to set it to exactly what it is. ## Motion detection: Motion detection is run on the CPU. The higher the resolution of the stream, the more work it is to detect motion frame to frame. [This is one of the reasons why using high resolutions is discouraged](https://docs.frigate.video/guides/camera_setup#choosing-a-detect-resolution). It is also important to add [motion masks](https://docs.frigate.video/configuration/masks) to places likely to not have objects like trees, skylines, etc. **NOTE:** Motion masks are not meant to block out actual objects, do not use them for this.

Greg.o · October 26, 2022, 4:56pm

Once again : thx !!

kilopter · August 8, 2023, 3:33pm

how you fix it?

andrejsvet · August 13, 2023, 12:33pm

I tried to use Coral USB Accelerator in HA for Frigate (installed in HA as add-on) but it’s not detected by HA/Frigate (see log file bellow). The latest version of HA is installed in Virtualbox 7.0 on Mini PC (Intel Celeron N5100, Ubuntu 23.04). I also added in configration file of Frigate the following:
detectors:
coral:
type: edgetpu
device: usb

Any idea what I’m doing wrong here?

2023-08-12 13:29:30.041231439  Process detector:coral:
2023-08-12 13:29:30.043694778  [2023-08-12 15:29:30] frigate.detectors.plugins.edgetpu_tfl ERROR   : No EdgeTPU was detected. If you do not have a Coral device yet, you must configure CPU detectors.
2023-08-12 13:29:30.045000356  [2023-08-12 15:29:30] frigate.app                    INFO    : Camera processor started for camzunaj: 307
2023-08-12 13:29:30.050800853  Traceback (most recent call last):
2023-08-12 13:29:30.050850595    File "/usr/lib/python3/dist-packages/tflite_runtime/interpreter.py", line 160, in load_delegate
2023-08-12 13:29:30.050852218      delegate = Delegate(library, options)
2023-08-12 13:29:30.050853671    File "/usr/lib/python3/dist-packages/tflite_runtime/interpreter.py", line 119, in __init__
2023-08-12 13:29:30.050859027      raise ValueError(capture.message)
2023-08-12 13:29:30.050860497  ValueError
2023-08-12 13:29:30.050874935  
2023-08-12 13:29:30.050876366  During handling of the above exception, another exception occurred:
2023-08-12 13:29:30.050877273  
2023-08-12 13:29:30.050878358  Traceback (most recent call last):
2023-08-12 13:29:30.050948803    File "/usr/lib/python3.9/multiprocessing/process.py", line 315, in _bootstrap
2023-08-12 13:29:30.050950345      self.run()
2023-08-12 13:29:30.050951662    File "/usr/lib/python3.9/multiprocessing/process.py", line 108, in run
2023-08-12 13:29:30.050952817      self._target(*self._args, **self._kwargs)
2023-08-12 13:29:30.050954114    File "/opt/frigate/frigate/object_detection.py", line 98, in run_detector
2023-08-12 13:29:30.050955418      object_detector = LocalObjectDetector(detector_config=detector_config)
2023-08-12 13:29:30.051019488    File "/opt/frigate/frigate/object_detection.py", line 52, in __init__
2023-08-12 13:29:30.051021493      self.detect_api = create_detector(detector_config)
2023-08-12 13:29:30.051022745    File "/opt/frigate/frigate/detectors/__init__.py", line 24, in create_detector
2023-08-12 13:29:30.051023832      return api(detector_config)
2023-08-12 13:29:30.051025132    File "/opt/frigate/frigate/detectors/plugins/edgetpu_tfl.py", line 37, in __init__
2023-08-12 13:29:30.051026445      edge_tpu_delegate = load_delegate("libedgetpu.so.1.0", device_config)
2023-08-12 13:29:30.051040431    File "/usr/lib/python3/dist-packages/tflite_runtime/interpreter.py", line 162, in load_delegate
2023-08-12 13:29:30.051041859      raise ValueError('Failed to load delegate from {}\n{}'.format(
2023-08-12 13:29:30.051077776  ValueError: Failed to load delegate from libedgetpu.so.1.0

Ijeamaka · October 2, 2023, 11:16pm

Disabled Protection mode on the Full access worked for me. Thanks.