[SUPPORT] blakeblackshear - Frigate


Recommended Posts

4 hours ago, yayitazale said:

Can you attach your config file and a screenshot of your deployment settings in advanced mode?

No config yet, as I'm just trying to install it. And the only changes I made to the settings are the ones listed here:

 

 

Link to comment
22 hours ago, flyize said:

No config yet, as I'm just trying to install it. And the only changes I made to the settings are the ones listed here:

 

 

Please Read the instructions before posting.

 

You have to create a config file before starting the container otherwise it will not work.

 

Link to comment

I hadn't seen anywhere that the Docker container would be trashed if the config wasn't there, but I did find a couple of posts in this thread that indicate something similar. That said, I copied in a working .yml file from Home Assistant and still got the same orphaned image.

 

Can you suggest a vanilla config I can try? I can build it out from there.

Edited by flyize
Link to comment

Having issues getting the my pci coral to show up correctly, /dev/apex_0 doesnt show up. I keep getting this in the logs and am lost. Any help would be greatly appreciated!

 

Jul 12 16:31:03 unraid kernel: apex 0000:0d:00.0: Page table init timed out
Jul 12 16:31:03 unraid kernel: apex 0000:0d:00.0: MSI-X table init timed out
Jul 12 16:31:03 unraid kernel: apex: probe of 0000:0d:00.0 failed with error -110

 

Link to comment
11 hours ago, flyize said:

I hadn't seen anywhere that the Docker container would be trashed if the config wasn't there, but I did find a couple of posts in this thread that indicate something similar. That said, I copied in a working .yml file from Home Assistant and still got the same orphaned image.

 

Can you suggest a vanilla config I can try? I can build it out from there.

Ok as I see you are changing the extra parameters of the container with the ones listed here:

The actual template doesn't need to change anything, your issue is caused by the "--rm" parameter that is deleting the container as soon as it stops by a faulty start becouse of the lack of the config file.

 

Delete that and you will not get a orphan container.

 

About the config file, please follow the steps of the github documentation to create your personal one adjusted to your needs:

 

https://blakeblackshear.github.io/frigate/

 

 

Link to comment

Hello.

I've been having issues with exit 139 errors after setting up the docker. Is there a parameter to set to make it auto restart if it happens?

I'm running with a CPU instead of TPU to test it out. Only 1 camera so far, but want to make sure it suits my need before purchasing a TPU.

Link to comment
15 minutes ago, tushn said:

Hello.

I've been having issues with exit 139 errors after setting up the docker. Is there a parameter to set to make it auto restart if it happens?

I'm running with a CPU instead of TPU to test it out. Only 1 camera so far, but want to make sure it suits my need before purchasing a TPU.

Please attach your config file and error output log

Link to comment
8 minutes ago, yayitazale said:

Please attach your config file and error output log

I've attached the config file. 

Docker is pretty standard, only changed paths for storage.

No error output. There's no indication in the logs, it just just says on the docker page "Exit 139"

 

I've just tried changing to privileged mode to see if it changes anything. Normally it can run for a few hours, but'll follow up with a complete log on the next exit.

 

config.yml

Edited by tushn
Link to comment

With the nvidia version of the container, should I be seeing much cpu usage? With 8 cameras and a pci TPU i'm getting ~50% cpu usage on a 24 core machine, which seems odd.

 

Also I cant seem to get anything other than person to show up as far as object detection goes? I set each camera to detect person, cat, dog, and car but it doesn't seem to take.

config.yml

Link to comment
3 hours ago, tushn said:

I've attached the config file. 

Docker is pretty standard, only changed paths for storage.

No error output. There's no indication in the logs, it just just says on the docker page "Exit 139"

 

I've just tried changing to privileged mode to see if it changes anything. Normally it can run for a few hours, but'll follow up with a complete log on the next exit.

 

config.yml 3.49 kB · 0 downloads

Your config looks fine (but you may add ffmpeg output args). Exit 139 code is a segmentation fault so maybe you have a corrupted donwload of the image or some enviromental issue. You can try to delete the image and reinstall the container. Anyway, if the problem persist please ask for help on the github forum of the project as this is beyond my knowledge.

Edited by yayitazale
Link to comment
2 hours ago, Lygris said:

With the nvidia version of the container, should I be seeing much cpu usage? With 8 cameras and a pci TPU i'm getting ~50% cpu usage on a 24 core machine, which seems odd.

 

Also I cant seem to get anything other than person to show up as far as object detection goes? I set each camera to detect person, cat, dog, and car but it doesn't seem to take.

config.yml 4.51 kB · 0 downloads

I'm almost sure that the high cpu usage is caused by improper configuration of the ffmpeg args. I'm not an ffmpeg expert so if you need help please refer to the github forum of the project.

 

About the detection, you are only filtering the clips to that objects like this:

 

clips:

      objects:

        - person

        - dog

        - cat

        - car

 

but actually are not tracking them doing so. You need to add the specific lines on the config to define witch objects you want to track at camera level or at global level as so:

 

objects:

      track:

        - person

        - car

        - motorcycle

 

Please follow the steps of the docs.

Link to comment
  • 2 weeks later...

Hello, i like to ask a question about the Coral USB, because the Frigate doc's send me here i give it a try. It might be a bit off-topic, hope you dont mind...

 

I bought a Coral USB to run Frigate on a NUC. Working toghetter with Home Assistant. I followed the instructions to set it up, log keeps telling me that the edgeTPU is not detected. Also do not see it with lsusb command (Linux Debian 10 / 64, USB 3 port). Plugged it into my Windows 10 and that also does not see a device. Is there a way to find out if the stick is healthy? There are no LED's or something ?

 

thanks!

Link to comment
6 hours ago, Hendrik6073 said:

Hello, i like to ask a question about the Coral USB, because the Frigate doc's send me here i give it a try. It might be a bit off-topic, hope you dont mind...

 

I bought a Coral USB to run Frigate on a NUC. Working toghetter with Home Assistant. I followed the instructions to set it up, log keeps telling me that the edgeTPU is not detected. Also do not see it with lsusb command (Linux Debian 10 / 64, USB 3 port). Plugged it into my Windows 10 and that also does not see a device. Is there a way to find out if the stick is healthy? There are no LED's or something ?

 

thanks!

If you don't even see an unknown device in Windows device manager, I'd assume the stick is dead.

Link to comment
4 minutes ago, Hendrik6073 said:

Okay! Where is it located and when does it lit? Always when connected to a USB (3.x) plug ? Or only when the drivers are loaded?. Yes i use the cable that did come with the package. Thanks!

The LED is on the translucid part of the coral. I only used it on USB 3 ports.

 

https://coral.ai/docs/accelerator/datasheet/

 

Quote

USB Accelerator datasheet v1.4
4.2 LED behavior
The USB Accelerator includes an LED that indicates the current system state. The following table describes the behavior of
the LED.
Edge TPU initialized On (solid)
Edge TPU running Pulse (breathe)

 

If you connect it to the PC it is solid blue.

 

You have to follow this steps (I think you only need to install the Windows drivers if you are going to use it natively as frigate docker it already comes with the drivers, but give a try):

 

https://coral.ai/docs/accelerator/get-started/#runtime-on-windows

https://coral.ai/docs/accelerator/get-started/#pycoral-on-mac-win

Link to comment
14 minutes ago, yayitazale said:

The LED is on the translucid part of the coral. I only used it on USB 3 ports.

 

 

14 minutes ago, yayitazale said:

The LED is on the translucid part of the coral. I only used it on USB 3 ports.

 

https://coral.ai/docs/accelerator/datasheet/

 

 

If you connect it to the PC it is solid blue.

 

You have to follow this steps (I think you only need to install the Windows drivers if you are going to use it natively as frigate docker it already comes with the drivers, but give a try):

 

https://coral.ai/docs/accelerator/get-started/#runtime-on-windows

https://coral.ai/docs/accelerator/get-started/#pycoral-on-mac-win

 

 

 

Thanks! i could have thought of that (the datasheet) myself 🥴.

 

Anyway, reading this:

 

4.2 LED behavior
The USB Accelerator includes an LED that indicates the current system state. The following table describes the behavior of
the LED.
Table 2. USB Accelerator LED behavior and meaning
System state LED behavior
Edge TPU initialized - Edge TPU initialized On (solid)

Edge TPU running - Edge TPU running Pulse (breathe)

 

Gives me the impression it might not lit at all when not loaded/initialized. The Windows machine i have is my tool for work, so have to be carefull installing this just for testing. have to find me another to test it on Windows

Edited by Hendrik6073
fallen off text
Link to comment

Hey all,

Does anyone know how I can prevent Frigate from making multiple overlapping clips when detecting multiple objects? It's really annoying having to constantly open several clips to get the entire event just because Frigate decides each object deserves its own clip.

Surely there's a way to make it keep recording so long as it detects ANY motion?

Link to comment
22 minutes ago, Corvus said:

Hey all,

Does anyone know how I can prevent Frigate from making multiple overlapping clips when detecting multiple objects? It's really annoying having to constantly open several clips to get the entire event just because Frigate decides each object deserves its own clip.

Surely there's a way to make it keep recording so long as it detects ANY motion?

https://github.com/blakeblackshear/frigate/issues/1313#issuecomment-872174685

Link to comment
2 hours ago, Hendrik6073 said:

 

 

Gives me the impression it might not lit at all when not loaded/initialized. The Windows machine i have is my tool for work, so have to be carefull installing this just for testing. have to find me another to test it on Windows

 

installed de drivers on a Ubuntu (only USB2) also not detected, nothing in LSUSB.

 I think it is dead

Link to comment

Fyi I've got a usb coral plugged into the back of my HP server which I'm reasonably sure is USB2.0 and I do indeed get a LED lit with a frigate docker. 

 

My inference speed is around 37 in debug which i have no idea is good or not!

 

Dave

Edited by Flubster
Spelling
Link to comment
1 hour ago, Flubster said:

Fyi I've got a usb coral plugged into the back of my HP server which I'm reasonably sure is USB2.0 and I do indeed get a LED lit with a frigate docker. 

 

My inference speed is around 37 in debug which i have no idea is good or not!

 

Dave

thanks for checking, going to send it back, hope next oe works!

 

Link to comment

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.