[SUPPORT] blakeblackshear - Frigate

[Original_Vecna]

I've just installed this, and going to the GUI of frigate yields page cannot be found.

Reading through here, the only bit of info is that I needed to create a default config.yml file in the /mnt/user/appdata/frigate directory with the following code:

 

mqtt:
  host:
cameras:
  back:
    ffmpeg:
      inputs:
        - path: rtsp://admin:[email protected]:554/cam/realmonitor?channel=1&subtype=0&authbasic=64
          roles:
            - detect
            - rtmp
    detect:
      width: 1280
      height: 720

 

I don't have Home Assistant installed, so I've left the mqtt host blank.

I suppose I've got a couple questions. One, why isn't a config.yml included in the install to get the docker up and running after install, where you edit your preferences like ZM or Shinobi?

And two, is Frigate not a all in one NVR like ZM and Shinobi? Does it require HA to work and HA is the front end?

 

[Bruceflix]

9 hours ago, Original_Vecna said:

why isn't a config.yml included

 Because everyone's would be different.

9 hours ago, Original_Vecna said:

Does it require HA to work

No but HA certainly works well with it.

 

9 hours ago, Original_Vecna said:

and HA is the front end

not at all

[Original_Vecna]

2 hours ago, Bruceflix said:

 Because everyone's would be different.

No but HA certainly works well with it.

 

not at all

Everyone's would be different, but if you need it to even get into the GUI, kinda important to have that for a functioning product no?

[Bruceflix]

I think you need to read the documentation. The installation steps point out pretty clearly how to create the config.yml and what is needed in it.  I am not sure what configuration options you are expecting on a FOSS project.  Given the target demograph of Frigate is the HA community they would usually already be well versed in YAML.  

Hey, if I can get it to work, anyone can.  

 

[Original_Vecna]

11 hours ago, Bruceflix said:

I think you need to read the documentation. The installation steps point out pretty clearly how to create the config.yml and what is needed in it.  I am not sure what configuration options you are expecting on a FOSS project.  Given the target demograph of Frigate is the HA community they would usually already be well versed in YAML.  

Hey, if I can get it to work, anyone can.  

 

I included everything the documentation indicated to start out with the yaml file in my OP. 

So clearly the installation needs something more than just a YAML config file to access the GUI, or something in the docker image is borked.

[yayitazale]

2 hours ago, Original_Vecna said:

I included everything the documentation indicated to start out with the yaml file in my OP. 

So clearly the installation needs something more than just a YAML config file to access the GUI, or something in the docker image is borked.

First of all, this is not the forum where you complain about how the project is implemented. This forum is only to resolve any question regarding the docker container deployment under unraid.

 

Second, if you have any complain of a FOSS you are always welcome to participate on the development doing PRs on the github. You are not buying a software so please don't ask for help in the same way you will ask it to a company (don't be rude), and ALWAYS READ THE DOCS before asking for help. By the way, in this particular project the documentation very complente, with plenty of information and examples.

 

3 hours ago, Original_Vecna said:

I included everything the documentation indicated to start out with the yaml file in my OP. 

So clearly the installation needs something more than just a YAML config file to access the GUI, or something in the docker image is borked.

This is just not true. You only need a proper config file to start using the system. If you have doubts about your file, read the docs, see the logs and then if you don't see what are you doing wrong then you can come here and post a proper question indicating how is your container deployed and attaching your config file.

 

[Bruceflix]

On 12/4/2022 at 1:02 AM, Original_Vecna said:

so I've left the mqtt host blank.

Step 1 in the docs. "Frigate requires a functioning MQTT server. "

Literally the first sentence.

[Original_Vecna]

3 hours ago, Bruceflix said:

Step 1 in the docs. "Frigate requires a functioning MQTT server. "

Literally the first sentence.

Right, I don't use HA. So the end of it all is simply, Frigate requires HA. Which is fine, I'll stick with Shinobi. No worries. Thanks for clarifying.

[yayitazale]

6 hours ago, Original_Vecna said:

Right, I don't use HA. So the end of it all is simply, Frigate requires HA. Which is fine, I'll stick with Shinobi. No worries. Thanks for clarifying.

I doesn't require HA, the MQTT is required as is the way to comunicate with the system from an external software on a standar way.

 

Anyway, this is a completely different system logic than shinobi o ZM, the idea is to use it a service from another service. It can be HA, but not only.  If you wan't a complete UI like a clasical NVR this is not your software (at least yet)

[bonzi]

Wondering if someone could post a working config for an nvidia gpu. I have been working at gettin frigate running over the past few days and am really happy with my config, I would just like to make use of my GPU.

 

Here is what I have done. I added --runtime=nvidia to my extra parameters, I added a key for 'NVIDIA_DRIVER_CAPABILITIES' and set to 'all'. Then I added my GPU ID to another key called 'NVIDIA_VISIBLE_DEVICES' with value 'GPU-41cb87cb-9647-0b97-bd8e-2921d00c4844'

 

I then deleted /dev/dri/renderD128 from Intel GPU/AMD mapping. This didn't start, so I put it back, it did start but doesn't use the GPU.

 

I am sure it is something obvious I have missed but I am a little stumped and hopefully someone can point me in the right direction.

 

Here is my config.yml

mqtt:

  host: 192.168.1.205

  port: 1883



detectors:

  coral:

    type: edgetpu

    device: usb



birdseye:

  enabled: True

  mode: continuous



cameras:

  frontdoor:

    ffmpeg:

      inputs:

        - path: rtsp://admin:[email protected]:554//h264Preview_01_main

          roles:

            - detect

            - record

  side:

    ffmpeg:

      inputs:

        - path: rtsp://admin:[email protected]:554//h264Preview_01_main

          roles:

            - detect

            - record

  front:

    ffmpeg:

      inputs:

        - path: rtsp://admin:[email protected]:554//h264Preview_01_main

          roles:

            - detect

            - record

detect:

  width:  2560

  height: 1920

record:

      enabled: True        

      retain:

days: 30

        mode: all

      events: 

retain:

          default: 30

          mode: motion

 

[bonzi]

Got it, I needed to add hwaccel_args: -c:v h264_cuvid after the ffmpeg lines for my inputs.

 

One last question is what should I put for the /dev/dri/renderD128, should I just leave this? It seems to work..

[yayitazale]

8 hours ago, bonzi said:

Got it, I needed to add hwaccel_args: -c:v h264_cuvid after the ffmpeg lines for my inputs.

 

One last question is what should I put for the /dev/dri/renderD128, should I just leave this? It seems to work..

If you going to use the Nvidia GPU with your actual configuration, you can just delete this mapping entry.

 

If you want to switch to integrated GPU instead of your Nvidia GPU, you must change the ffmpeg hwaccel_args acording to what CPU you have following the instructions of the docs.

 

 

[Original_Vecna]

Quick Question. So I went ahead and installed HA with Mosquito, and got Frigate running on latest version of Ubuntu minimal desktop.

Frigate installed, connected to HA no problems, even managed to sort out the cameras. System looks great. My only issue is with installing pycoral library.  It repeatedly fails saying my version of python is too high etc. How do you get around this?

[Bruceflix]

9 minutes ago, Original_Vecna said:

got Frigate running on latest version of Ubuntu minimal desktop.

 

The hint is in the domain name of this forum. 

[Original_Vecna]

2 hours ago, Bruceflix said:

The hint is in the domain name of this forum. 

Running frigate in unraid isn't the recommended configuration.

[Original_Vecna]

2 hours ago, Bruceflix said:

The hint is in the domain name of this forum. 

Though I did take your advice, and try installing the frigate unraid docker. It worked, and the coral TPU is working. So...screw it! Unraid it is! LOL

[Xxharry]

I want to update to frigate 0.11.0. I understand that I have to delete the existing app and recreate it with the new image. Do I lose all the old footage by doing so? Thanks

[yayitazale]

On 12/10/2022 at 9:44 PM, Xxharry said:

I want to update to frigate 0.11.0. I understand that I have to delete the existing app and recreate it with the new image. Do I lose all the old footage by doing so? Thanks

If you create a new container but using the same mappings for the recors and the config folders, and you also mantain the database in the same path you will not lose anything.

 

In case that you want to roll back from v11 to v10 you should do a database restore because this migration to v11 will modify the structure of the db, so just in case, make a copy of the database.db file.

 

 

[ec911]

Hi,

 

I'm having problems with the Frigate Docker container. I've tried to troubleshoot as much as I can but I can't fix it.

Everytime I start a fresh install it seems to work fine (detection with Coral and recording) but after a couple hours I get this storage error in the log : Errno 28 - No space left om

device.

I am passing a freshly formated 2 TB hard drive that is clearly not full.

I also tried tweaking the tmpfs size from 1gb to 3gb for the /tmp/cache folder and it isn't working.

I don't know what to do next

 

 

[yayitazale]

6 minutes ago, ec911 said:

Hi,

 

I'm having problems with the Frigate Docker container. I've tried to troubleshoot as much as I can but I can't fix it.

Everytime I start a fresh install it seems to work fine (detection with Coral and recording) but after a couple hours I get this storage error in the log : Errno 28 - No space left om

device.

I am passing a freshly formated 2 TB hard drive that is clearly not full.

I also tried tweaking the tmpfs size from 1gb to 3gb for the /tmp/cache folder and it isn't working.

102.29 kB · 0 downloads

I don't know what to do next

 

 

I'll asume that you know what are you doing writting directly to disks istead of using shares under unraid. As you can see there is no space on the media path folder that you are mapping to /media/frigate

[ec911]

10 hours ago, yayitazale said:

I'll asume that you know what are you doing writting directly to disks istead of using shares under unraid. As you can see there is no space on the media path folder that you are mapping to /media/frigate

Ok what I did tonight is just click the mount button next to the drive in unassigned devices (I am still a noob) and I've changed the access mode of the /media/frigate folder in the docker configuration to read/write slave and everything booted up correctly!

 

Finger crossed and I'll see if everything is still ok tomorrow

 

 

[Original_Vecna]

Got my system dialed in pretty well now using the coral TPU, and I must admit, this thing is amazing. Fantastic Job. I have one issue however, and that is frigate isn't recording events reliably. I suspect it's ENTIRELY my settings that I'm currently using. On one camera, I lowered the person threshold to 0.7 from 0.8, and it started working better, but that kind of defeats the purpose of reducing false positives.

 

My question is, all my substreams are 640x480, and I've read that is well enough for detection, but is that in fact true? Or should I be detecting at the native 1080p or 4k of the camera?

This is what I have for code, which is the same for every camera.

 

cameras:
  Driveway:
    ffmpeg:
      inputs:
       ### 1080P Mainstream ###
        - path: rtsp://path
          roles:
            - record
       ### 640x480 Substream ###
        - path: rtsp://path
          roles:
            - detect
            - rtmp
    detect:
      width: 640
      height: 480
      fps: 29.970030
    objects:
      track:
        - person
        - car
        - dog
      filters:
        person:
          threshold: 0.8
          min_area: 2500
          max_area: 100000
    snapshots:
      enabled: True
    record:
      enabled: True
      retain:
        days: 7
        mode: motion
      events:
        retain:
          default: 30
          mode: active_objects
          objects:
            dog: 2
    rtmp:
      enabled: False
    motion:
      mask:
        - 128,79,142,82,159,72,179,63,200,51,217,45,235,36,250,30,271,27,297,29,349,28,388,21,413,25,436,29,462,32,508,44,540,56,570,70,599,83,640,105,640,0,0,0,0,138
        - 640,43,640,0,428,0,429,42

 

Edited December 27, 2022 by Original_Vecna
Grammar

[Original_Vecna]

My major issues are detection, not being accurate (not triggering). I literally have to be standing in front of the camera, and at night, the recording (detection) drops to near zero. 

 

Is this more an issue with poor cameras, or are my posted settings just plum wrong?

Edited December 29, 2022 by Original_Vecna

[yayitazale]

19 hours ago, Original_Vecna said:

My major issues are detection, not being accurate (not triggering). I literally have to be standing in front of the camera, and at night, the recording (detection) drops to near zero. 

 

Is this more an issue with poor cameras, or are my posted settings just plum wrong?

Sorry, I don't know what happened but it seems that the answer I wrote to you has not been published correctly.

 

What I was saying:

Using a higher resolution can help only in cases where the camera plane is very wide and the objects to be identified are far away and/or very small. You can read about this here: https://docs.frigate.video/guides/camera_setup#choosing-a-detect-resolution

 

In any case, you could lower the FPS to 5 to reduce the load of Coral and your CPU/GPU when processing the incoming video.

 

For the trigger values, when you say you have dropped from 0.7 to 0.8 it is not quite correct. That number is the confidence of the model on the identified object. You should start setting lower values (0.3 or 0.5) in the min_score and a value of 0.5 to 0.6 in threshold to see if the number of detections increases. You can track the events or watch the camera on the debugging screen and tinker with the values until you reach a point where the number of false positives is minimal and the number of detections is acceptable. https://docs.frigate.video/guides/false_positives

 

You can accompany this tinkering steps by introducing masks, regions etc. to try to minimize false positives.

 

As for night detection, in my case at least with IR cameras the number of detections is reduced but I get decent results, I think it all depends on the lighting, camera plane and the configuration of the values that I mention.

 

In any case, the purpose of this forum thread is to help with the setup of the docker container on unraid, for questions purely related to the application configuration you can open a discussion or an issue on github.

https://github.com/blakeblackshear/frigate/discussions

[cpu]

Anone has tips how to run go2rtc to use webrtc along frigate 0.12b1 ? I'm struggling with that.