Jump to content

[SUPPORT] blakeblackshear - Frigate


Recommended Posts

I finished reinstalling with the nvidia option and entered all of the required config settings, but I am still getting a lot off errors if I try enabling h/w acceleration:

 

2023-10-02 10:40:40.263887752  [2023-10-02 10:40:40] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:40:40.263918539  [2023-10-02 10:40:40] frigate.video                  ERROR   : birdcam: ffmpeg process is not running. exiting capture thread...
2023-10-02 10:40:42.371988851  [2023-10-02 10:40:39] detector.coral                 INFO    : Starting detection process: 2262
2023-10-02 10:40:42.371997559  [2023-10-02 10:40:39] frigate.detectors.plugins.edgetpu_tfl INFO    : Attempting to load TPU as usb
2023-10-02 10:40:42.382914475  [2023-10-02 10:40:42] frigate.detectors.plugins.edgetpu_tfl INFO    : TPU found
2023-10-02 10:40:47.289382996  [INFO] Starting go2rtc healthcheck service...
2023-10-02 10:40:59.420306333  [2023-10-02 10:40:59] watchdog.birdcam               ERROR   : Ffmpeg process crashed unexpectedly for birdcam.
2023-10-02 10:40:59.436311048  [2023-10-02 10:40:59] watchdog.birdcam               ERROR   : The following ffmpeg logs include the last 100 lines prior to exit.
2023-10-02 10:40:59.436328655  [2023-10-02 10:40:59] ffmpeg.birdcam.detect          ERROR   : [AVHWDeviceContext @ 0x55d8a9da1040] Cannot load libcuda.so.1
2023-10-02 10:40:59.436334006  [2023-10-02 10:40:59] ffmpeg.birdcam.detect          ERROR   : [AVHWDeviceContext @ 0x55d8a9da1040] Could not dynamically load CUDA
2023-10-02 10:40:59.436346619  [2023-10-02 10:40:59] ffmpeg.birdcam.detect          ERROR   : Device creation failed: -1.
2023-10-02 10:40:59.436399051  [2023-10-02 10:40:59] ffmpeg.birdcam.detect          ERROR   : [h264 @ 0x55d8a9d9b4c0] No device available for decoder: device type cuda needed for codec h264.
2023-10-02 10:40:59.436405127  [2023-10-02 10:40:59] ffmpeg.birdcam.detect          ERROR   : Device setup failed for decoder on input stream #0:2 : Operation not permitted
2023-10-02 10:41:00.319363449  [2023-10-02 10:41:00] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:00.319389983  [2023-10-02 10:41:00] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:00.319397193  [2023-10-02 10:41:00] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:00.319404083  [2023-10-02 10:41:00] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:00.319410810  [2023-10-02 10:41:00] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:00.319416764  [2023-10-02 10:41:00] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:00.319422951  [2023-10-02 10:41:00] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:00.335891151  [2023-10-02 10:41:00] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:00.336886460  [2023-10-02 10:41:00] frigate.video                  ERROR   : birdcam: ffmpeg process is not running. exiting capture thread...
2023-10-02 10:41:09.452729614  [2023-10-02 10:41:09] watchdog.birdcam               ERROR   : Ffmpeg process crashed unexpectedly for birdcam.
2023-10-02 10:41:09.452943020  [2023-10-02 10:41:09] watchdog.birdcam               ERROR   : The following ffmpeg logs include the last 100 lines prior to exit.
2023-10-02 10:41:09.453094972  [2023-10-02 10:41:09] ffmpeg.birdcam.detect          ERROR   : [h264 @ 0x56496a9769c0] Missing reference picture, default is 0
2023-10-02 10:41:09.453317708  [2023-10-02 10:41:09] ffmpeg.birdcam.detect          ERROR   : [h264 @ 0x56496a9769c0] decode_slice_header error
2023-10-02 10:41:09.453526268  [2023-10-02 10:41:09] ffmpeg.birdcam.detect          ERROR   : [AVHWDeviceContext @ 0x56496a9e6700] Cannot load libcuda.so.1
2023-10-02 10:41:09.453724545  [2023-10-02 10:41:09] ffmpeg.birdcam.detect          ERROR   : [AVHWDeviceContext @ 0x56496a9e6700] Could not dynamically load CUDA
2023-10-02 10:41:09.453887118  [2023-10-02 10:41:09] ffmpeg.birdcam.detect          ERROR   : Device creation failed: -1.
2023-10-02 10:41:09.454082484  [2023-10-02 10:41:09] ffmpeg.birdcam.detect          ERROR   : [h264 @ 0x56496a9e1e00] No device available for decoder: device type cuda needed for codec h264.
2023-10-02 10:41:09.454317782  [2023-10-02 10:41:09] ffmpeg.birdcam.detect          ERROR   : Device setup failed for decoder on input stream #0:2 : Operation not permitted
2023-10-02 10:41:10.380691324  [2023-10-02 10:41:10] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:10.381545593  [2023-10-02 10:41:10] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:10.382056723  [2023-10-02 10:41:10] frigate.video                  ERROR   : birdcam: ffmpeg process is not running. exiting capture thread...
2023-10-02 10:41:19.516829810  [2023-10-02 10:41:19] watchdog.birdcam               ERROR   : Ffmpeg process crashed unexpectedly for birdcam.
2023-10-02 10:41:19.516840062  [2023-10-02 10:41:19] watchdog.birdcam               ERROR   : The following ffmpeg logs include the last 100 lines prior to exit.
2023-10-02 10:41:19.516842455  [2023-10-02 10:41:19] ffmpeg.birdcam.detect          ERROR   : [AVHWDeviceContext @ 0x56157000a940] Cannot load libcuda.so.1
2023-10-02 10:41:19.516844550  [2023-10-02 10:41:19] ffmpeg.birdcam.detect          ERROR   : [AVHWDeviceContext @ 0x56157000a940] Could not dynamically load CUDA
2023-10-02 10:41:19.516846341  [2023-10-02 10:41:19] ffmpeg.birdcam.detect          ERROR   : Device creation failed: -1.
2023-10-02 10:41:19.516848379  [2023-10-02 10:41:19] ffmpeg.birdcam.detect          ERROR   : [h264 @ 0x56156ff3a440] No device available for decoder: device type cuda needed for codec h264.
2023-10-02 10:41:19.516850369  [2023-10-02 10:41:19] ffmpeg.birdcam.detect          ERROR   : Device setup failed for decoder on input stream #0:2 : Operation not permitted
2023-10-02 10:41:20.379321927  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.379331894  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387038483  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387065073  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387070413  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387075330  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387080167  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387084457  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387088741  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387093004  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387097294  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387271005  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387276949  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387282942  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387287532  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387348991  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387354264  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387358844  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387363088  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387433363  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387438750  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387442976  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387447590  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387451933  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387456273  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387460714  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387464924  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387469120  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387541206  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387546396  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387550586  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387554699  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387559336  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387563543  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387567793  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387572046  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387656249  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387661495  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.387665705  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.389857922  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:20.390326728  [2023-10-02 10:41:20] frigate.video                  ERROR   : birdcam: ffmpeg process is not running. exiting capture thread...
2023-10-02 10:41:29.537778246  [2023-10-02 10:41:29] watchdog.birdcam               ERROR   : Ffmpeg process crashed unexpectedly for birdcam.
2023-10-02 10:41:29.537801219  [2023-10-02 10:41:29] watchdog.birdcam               ERROR   : The following ffmpeg logs include the last 100 lines prior to exit.
2023-10-02 10:41:29.537803624  [2023-10-02 10:41:29] ffmpeg.birdcam.detect          ERROR   : [AVHWDeviceContext @ 0x561eb03a7840] Cannot load libcuda.so.1
2023-10-02 10:41:29.537805484  [2023-10-02 10:41:29] ffmpeg.birdcam.detect          ERROR   : [AVHWDeviceContext @ 0x561eb03a7840] Could not dynamically load CUDA
2023-10-02 10:41:29.537807339  [2023-10-02 10:41:29] ffmpeg.birdcam.detect          ERROR   : Device creation failed: -1.
2023-10-02 10:41:29.537809255  [2023-10-02 10:41:29] ffmpeg.birdcam.detect          ERROR   : [h264 @ 0x561eb03aac40] No device available for decoder: device type cuda needed for codec h264.
2023-10-02 10:41:29.537811060  [2023-10-02 10:41:29] ffmpeg.birdcam.detect          ERROR   : Device setup failed for decoder on input stream #0:2 : Operation not permitted
2023-10-02 10:41:30.413770640  [2023-10-02 10:41:30] frigate.video                  ERROR   : birdcam: Unable to read frames from ffmpeg process.
2023-10-02 10:41:30.414111185  [2023-10-02 10:41:30] frigate.video                  ERROR   : birdcam: ffmpeg process is not running. exiting capture thread...

Link to comment
51 minutes ago, yayitazale said:

Maybe you need to reboot the host...

I don't think I've misconfigured anything...

 

docker run
  -d
  --name='frigate'
  --net='br0'
  --ip='192.168.0.48'
  --privileged=true
  -e TZ="Europe/London"
  -e HOST_OS="Unraid"
  -e HOST_HOSTNAME="Unraid"
  -e HOST_CONTAINERNAME="frigate"
  -e 'TCP_PORT_5000'='5000'
  -e 'TCP_PORT_8554'='8554'
  -e 'FRIGATE_RTSP_PASSWORD'='xxxxx'
  -e 'NVIDIA_VISIBLE_DEVICES'='GPU-6b2....5a0'
  -e 'NVIDIA_DRIVER_CAPABILITIES'='compute,utility,video'
  -e 'TCP_PORT_8555'='8555'
  -e 'UDP_PORT_8555'='8555'
  -e 'TCP_PORT_1984'='1984'
  -l net.unraid.docker.managed=dockerman
  -l net.unraid.docker.webui='http://[IP]:[PORT:5000]'
  -l net.unraid.docker.icon='https://raw.githubusercontent.com/yayitazale/unraid-templates/main/frigate.png'
  -v '/mnt/cache/appdata/frigate':'/config':'rw'
  -v '/mnt/user/Media/frigate':'/media/frigate':'rw'
  -v '/mnt/user/appdata/trt-models':'/trt-models':'ro'
  -v '/mnt/user/appdata/trt-models':'/trt-models':'ro'
  -v '/etc/localtime':'/etc/localtime':'rw'
  --device='/dev/bus/usb'
  --shm-size=256mb
  --mount type=tmpfs,target=/tmp/cache,tmpfs-size=100000000
  --restart unless-stopped
  --runtime=nvidia 'ghcr.io/blakeblackshear/frigate:stable-tensorrt'

 

Link to comment
21 hours ago, BurningSky said:

I don't think I've misconfigured anything...

 

docker run
  -d
  --name='frigate'
  --net='br0'
  --ip='192.168.0.48'
  --privileged=true
  -e TZ="Europe/London"
  -e HOST_OS="Unraid"
  -e HOST_HOSTNAME="Unraid"
  -e HOST_CONTAINERNAME="frigate"
  -e 'TCP_PORT_5000'='5000'
  -e 'TCP_PORT_8554'='8554'
  -e 'FRIGATE_RTSP_PASSWORD'='xxxxx'
  -e 'NVIDIA_VISIBLE_DEVICES'='GPU-6b2....5a0'
  -e 'NVIDIA_DRIVER_CAPABILITIES'='compute,utility,video'
  -e 'TCP_PORT_8555'='8555'
  -e 'UDP_PORT_8555'='8555'
  -e 'TCP_PORT_1984'='1984'
  -l net.unraid.docker.managed=dockerman
  -l net.unraid.docker.webui='http://[IP]:[PORT:5000]'
  -l net.unraid.docker.icon='https://raw.githubusercontent.com/yayitazale/unraid-templates/main/frigate.png'
  -v '/mnt/cache/appdata/frigate':'/config':'rw'
  -v '/mnt/user/Media/frigate':'/media/frigate':'rw'
  -v '/mnt/user/appdata/trt-models':'/trt-models':'ro'
  -v '/mnt/user/appdata/trt-models':'/trt-models':'ro'
  -v '/etc/localtime':'/etc/localtime':'rw'
  --device='/dev/bus/usb'
  --shm-size=256mb
  --mount type=tmpfs,target=/tmp/cache,tmpfs-size=100000000
  --restart unless-stopped
  --runtime=nvidia 'ghcr.io/blakeblackshear/frigate:stable-tensorrt'

 

Sorry but this is out of my knowledge, I don't know why is not working, everything looks good except you had installed the driver, but theoretically with a uninstall and reboot should work again.

Link to comment
On 10/7/2023 at 9:21 PM, sapphireone said:

Bonsoir j’ai besoin d’aide j’ai un google coral pcie mais il n’est pas reconnu dans la vm

Capture d'écran 2023-10-07 195136.png

Capture d'écran 2023-10-07 195308.png

First of all, please keep your communications in English which, as you can see, is the language we use on this site. Unfortunately I do not understand French.

 

Secondly, in this forum we deal with the problems you may have to launch the frigate docker container in unraid. Using a VM is not part of the instructions so you will have to ask in the appropriate place, which is the section related to the VM engine:
https://forums.unraid.net/forum/51-vm-engine-kvm/

 

Link to comment

I have my RTX 3070 setup in frigate to transcode and do tensor detection.  Transcoding it working and the card shows up as a detector but it isn't triggering any events.  Events do trigger correctly using a Coral TPU.  Is there something else I need to configure to get tensorrt events working?

 

Quote

tensorrt:

type: tensorrt

device: 0 #This is the default, select the first GPU

model:

path: /trt-models/yolov4-tiny-416.trt

input_tensor: nchw

input_pixel_format: rgb

width: 416

height: 416

 

Link to comment
On 10/13/2023 at 1:27 AM, Brenholder said:

I have my RTX 3070 setup in frigate to transcode and do tensor detection.  Transcoding it working and the card shows up as a detector but it isn't triggering any events.  Events do trigger correctly using a Coral TPU.  Is there something else I need to configure to get tensorrt events working?

 

 

Is there anything in the logs to give us a clue? Does your frigate docker see the models path?

Link to comment

The models are in the path

Quote

# ls /trt-models
libyolo_layer.so  tensorrt_models.sh  yolov4-tiny-288.trt  yolov4-tiny-416.trt  yolov7-tiny-416.trt

 

I don't see anything in the log that tripped me off 

Quote

2023-10-14 09:32:18.736043978  [INFO] Preparing go2rtc config...
2023-10-14 09:32:18.736045908  [INFO] Starting NGINX...
2023-10-14 09:32:18.736158460  [INFO] Starting Frigate...
2023-10-14 09:32:19.734493709  [INFO] Starting go2rtc...
2023-10-14 09:32:19.812796501  09:32:19.812 INF go2rtc version 1.2.0 linux/amd64
2023-10-14 09:32:19.812964273  09:32:19.812 INF [api] listen addr=:1984
2023-10-14 09:32:19.813096617  09:32:19.813 INF [rtsp] listen addr=:8554
2023-10-14 09:32:19.813250926  09:32:19.813 INF [srtp] listen addr=:8443
2023-10-14 09:32:19.813384121  09:32:19.813 INF [webrtc] listen addr=:8555
2023-10-14 09:32:23.721252445  2023/10/14 09:32:23 [error] 134#134: *1 connect() failed (111: Connection refused) while connecting to upstream, client: 192.168.86.4, server: , request: "GET /api/stats HTTP/1.1", upstream: "http://127.0.0.1:5001/stats", host: "192.168.86.5:5055"
2023-10-14 09:32:23.721311825  192.168.86.4 - - [14/Oct/2023:09:32:23 -0500] "GET /api/stats HTTP/1.1" 502 157 "-" "HomeAssistant/2023.10.1 aiohttp/3.8.5 Python/3.11" "-"
2023-10-14 09:32:24.226518758  [2023-10-14 09:32:24] frigate.app                    INFO    : Starting Frigate (0.12.1-367d724)
2023-10-14 09:32:24.266512431  [2023-10-14 09:32:24] frigate.config                 WARNING : Customizing more than a detector model path is unsupported.
2023-10-14 09:32:24.273917114  [2023-10-14 09:32:24] peewee_migrate                 INFO    : Starting migrations
2023-10-14 09:32:24.282660613  [2023-10-14 09:32:24] peewee_migrate                 INFO    : There is nothing to migrate
2023-10-14 09:32:24.302542076  [2023-10-14 09:32:24] detector.tensorrt              INFO    : Starting detection process: 1009
2023-10-14 09:32:24.336196918  [2023-10-14 09:32:24] frigate.app                    INFO    : Output process started: 1011
2023-10-14 09:32:24.336205560  [2023-10-14 09:32:24] frigate.app                    INFO    : Camera processor started for basement_camera: 1015
2023-10-14 09:32:24.336214851  [2023-10-14 09:32:24] frigate.app                    INFO    : Camera processor started for kitchen_camera: 1018
2023-10-14 09:32:24.354986466  [2023-10-14 09:32:24] frigate.app                    INFO    : Capture process started for basement_camera: 1019
2023-10-14 09:32:24.356919617  [2023-10-14 09:32:24] frigate.app                    INFO    : Capture process started for kitchen_camera: 1022
2023-10-14 09:32:27.234669882  [2023-10-14 09:32:27] frigate.detectors.plugins.tensorrt INFO    : [MemUsageChange] Init CUDA: CPU +564, GPU +0, now: CPU 622, GPU 859 (MiB)
2023-10-14 09:32:27.354484989  [2023-10-14 09:32:27] frigate.detectors.plugins.tensorrt INFO    : Loaded engine size: 12 MiB
2023-10-14 09:32:27.387129417  [2023-10-14 09:32:27] frigate.detectors.plugins.tensorrt INFO    : [MemUsageChange] Init cuBLAS/cuBLASLt: CPU +6, GPU +8, now: CPU 664, GPU 881 (MiB)
2023-10-14 09:32:27.404870682  [2023-10-14 09:32:27] frigate.detectors.plugins.tensorrt INFO    : [MemUsageChange] Init cuDNN: CPU +1, GPU +10, now: CPU 665, GPU 891 (MiB)
2023-10-14 09:32:27.406150773  [2023-10-14 09:32:27] frigate.detectors.plugins.tensorrt INFO    : [MemUsageChange] TensorRT-managed allocation in engine deserialization: CPU +0, GPU +11, now: CPU 0, GPU 11 (MiB)
2023-10-14 09:32:27.412294459  [2023-10-14 09:32:27] frigate.detectors.plugins.tensorrt INFO    : [MemUsageChange] Init cuBLAS/cuBLASLt: CPU +0, GPU +8, now: CPU 653, GPU 883 (MiB)
2023-10-14 09:32:27.412299640  [2023-10-14 09:32:27] frigate.detectors.plugins.tensorrt INFO    : [MemUsageChange] Init cuDNN: CPU +0, GPU +8, now: CPU 653, GPU 891 (MiB)
2023-10-14 09:32:27.412484778  [2023-10-14 09:32:27] frigate.detectors.plugins.tensorrt INFO    : [MemUsageChange] TensorRT-managed allocation in IExecutionContext creation: CPU +0, GPU +7, now: CPU 0, GPU 18 (MiB)

 

Link to comment
32 minutes ago, yayitazale said:

And in the debug page do you see detections in any of the cameras?

No detections.  Watching the debug view I can see the bounding box for motion but no detections.  The GPU shows up on the system page as a detector.

 

Using the same config file if enable a TPU detector the detection works and I can see the bonding boxes in the debug view. So I think I have the camera and object sections setup correctly.

Link to comment
1 hour ago, Brenholder said:

No detections.  Watching the debug view I can see the bounding box for motion but no detections.  The GPU shows up on the system page as a detector.

 

Using the same config file if enable a TPU detector the detection works and I can see the bonding boxes in the debug view. So I think I have the camera and object sections setup correctly.

Can you test lowering the min_score of the objects?

Link to comment
2 hours ago, yayitazale said:

Can you test lowering the min_score of the objects?

I turned person down to 10% and still nothing.

 

I went back to the logs from the docker that generates the .trt files there were a bunch of lines

 

Quote

[10/13/2023-18:00:43] [TRT] [W] Weights [name=030_convolutional.weight] had the following issues when converted to FP16:
[10/13/2023-18:00:43] [TRT] [W]  - Subnormal FP16 values detected. 
[10/13/2023-18:00:43] [TRT] [W]  - Values less than smallest positive FP16 Subnormal value detected. Converting to FP16 minimum subnormalized value.

Does that mean I should disable FP16?

Link to comment
On 10/14/2023 at 10:25 PM, Brenholder said:

I turned person down to 10% and still nothing.

 

I went back to the logs from the docker that generates the .trt files there were a bunch of lines

 

Does that mean I should disable FP16?

Not sure but I think this is a normal log output but you can test it with USE_FP16=False

Link to comment
13 hours ago, yayitazale said:

Not sure but I think this is a normal log output but you can test it with USE_FP16=False

OK, I've made some progress.

 

detectors: 
#  coral:
#    type: edgetpu
#    device: pci:0
  tensorrt:
    type: tensorrt
    device: 0 #This is the default, select the first GPU
    model:
      path: /trt-models/yolov4-tiny-416.trt
      input_tensor: nchw
      input_pixel_format: rgb
      width: 416
      height: 416

I had the config file like this.  The detector doesn't work, but there are no errors in the log.  Uncommenting the coral lines and the coral detector with start working. 

detectors: 
#  coral:
#    type: edgetpu
#    device: pci:0
  tensorrt:
    type: tensorrt
    device: 0 #This is the default, select the first GPU
model:
  path: /trt-models/yolov4-tiny-416.trt
  input_tensor: nchw
  input_pixel_format: rgb
  width: 416
  height: 416

This config does work, the tensorrt detector events start showing up.  But the there is an issue when trying to use the coral detector.  Uncommenting those lines will cause the coral to not be detected

 

Quote

2023-10-16 15:35:48.281712703 [2023-10-16 15:35:48] frigate.detectors.plugins.edgetpu_tfl INFO : TPU found 2023-10-16 15:35:48.281819586 [2023-10-16 15:35:48] frigate.detectors.plugins.edgetpu_tfl ERROR : No EdgeTPU was detected. If you do not have a Coral device yet, you must configure CPU detectors.

At this point I can have the coral tpu or the GPU detector running, but not both.

 

Link to comment
9 hours ago, Brenholder said:

OK, I've made some progress.

 

detectors: 
#  coral:
#    type: edgetpu
#    device: pci:0
  tensorrt:
    type: tensorrt
    device: 0 #This is the default, select the first GPU
    model:
      path: /trt-models/yolov4-tiny-416.trt
      input_tensor: nchw
      input_pixel_format: rgb
      width: 416
      height: 416

I had the config file like this.  The detector doesn't work, but there are no errors in the log.  Uncommenting the coral lines and the coral detector with start working. 

detectors: 
#  coral:
#    type: edgetpu
#    device: pci:0
  tensorrt:
    type: tensorrt
    device: 0 #This is the default, select the first GPU
model:
  path: /trt-models/yolov4-tiny-416.trt
  input_tensor: nchw
  input_pixel_format: rgb
  width: 416
  height: 416

This config does work, the tensorrt detector events start showing up.  But the there is an issue when trying to use the coral detector.  Uncommenting those lines will cause the coral to not be detected

 

At this point I can have the coral tpu or the GPU detector running, but not both.

 

You posted the config code without respecting the tabulation son I wasn't able to see that, indeed, you must follow the exact tabulation of the examples: 

https://docs.frigate.video/configuration/detectors#configuration-parameters

 

You can't have multiple type detectors running at the same time (at least for now). You can only have multiple Coral detectors at the same time:

https://docs.frigate.video/configuration/detectors#multiple-usb-corals

 

Link to comment
  • 2 weeks later...

So I had a power outage, and some data corruption. I ended up having to reinstall Frigate, MQTT, and go2rtc. Frigate appears to be working, but I am notincing MQTT errors in both the MQTT and Frigate logs:

 

MQTT

Saving in-memory database to /config/data/mosquitto.db.
Error saving in-memory database, unable to open /config/data/mosquitto.db.new for writing.
Error: Permission denied.

 

Frigate

2023-10-25 09:02:43.544108757 [2023-10-25 09:02:43] frigate.comms.mqtt

ERROR : Unable to publish to birdcam/all: client is not connected 2023-10-25 09:02:49.293686075 [2023-10-25 09:02:49] frigate.comms.mqtt

ERROR : Unable to publish to birdcam/motion: client is not connected 2023-10-25 09:03:53.810229650 [2023-10-25 09:03:53] frigate.comms.mqtt

ERROR : Unable to publish to stats: client is not connected 2023-10-25 09:04:57.212057130 [2023-10-25 09:04:57] frigate.comms.mqtt

ERROR : Unable to publish to stats: client is not connected 2023-10-25 09:06:00.600421374 [2023-10-25 09:06:00] frigate.comms.mqtt

ERROR : Unable to publish to stats: client is not connected

 

I checked the permissions for /config/data/mosquitto.db.new and they are 666, so anyone should be able to write to it.

 

Again, Frigate appears to be working fine, but I would like to get rid of these errors, if possible.

Link to comment
31 minutes ago, irishjd said:

So I had a power outage, and some data corruption. I ended up having to reinstall Frigate, MQTT, and go2rtc. Frigate appears to be working, but I am notincing MQTT errors in both the MQTT and Frigate logs:

 

MQTT

Saving in-memory database to /config/data/mosquitto.db.
Error saving in-memory database, unable to open /config/data/mosquitto.db.new for writing.
Error: Permission denied.

 

Frigate

2023-10-25 09:02:43.544108757 [2023-10-25 09:02:43] frigate.comms.mqtt

ERROR : Unable to publish to birdcam/all: client is not connected 2023-10-25 09:02:49.293686075 [2023-10-25 09:02:49] frigate.comms.mqtt

ERROR : Unable to publish to birdcam/motion: client is not connected 2023-10-25 09:03:53.810229650 [2023-10-25 09:03:53] frigate.comms.mqtt

ERROR : Unable to publish to stats: client is not connected 2023-10-25 09:04:57.212057130 [2023-10-25 09:04:57] frigate.comms.mqtt

ERROR : Unable to publish to stats: client is not connected 2023-10-25 09:06:00.600421374 [2023-10-25 09:06:00] frigate.comms.mqtt

ERROR : Unable to publish to stats: client is not connected

 

I checked the permissions for /config/data/mosquitto.db.new and they are 666, so anyone should be able to write to it.

 

Again, Frigate appears to be working fine, but I would like to get rid of these errors, if possible.

So the error is not about frigate, is that your MQTT server is not starting. You should check the permissions and ownership on unraid of the mapped folder to MQTT docker "/config/data/" path.

Link to comment

@yayitazale yea, I was thinking that it is a MQTT issue, but was hoping that another Frigate user might have run across it and found a solution. The permissions appear to be fine: RW RW RW, so I'm not sure why it is having write issues. Anyway, I have posted in the MQTT forum as well. If I can get it resolved, I'll put the solution here in case someone else has the issue too.

  • Like 1
Link to comment

I expect you are running in to the config issue. See my post below.

 

NOTE that frigates url has a port but mqqt SHOULD NOT have a port as it uses the default mqqt port.

 

On 9/9/2023 at 5:46 PM, dopeytree said:

Took me a few days but finally got Whosinmyfeeder working. 

 

838556974_Screenshot2023-09-09at17_44_36.thumb.png.25feb6085c90ff9596629e04c1f5b690.png

 

Frigate

1749627485_Screenshot2023-09-09at17_44_55.thumb.png.94e74b51b1a4b514ea54429c52a55445.png

 

The problem was I was putting the MQQT port on the end of the IP.

For anyone else looking this is what the config should look like for Whosinmyfeeder

 

NOTE: Only frigate needs a port no on the end of the IP

 

Whosinmyfeeder config.yml

frigate:
  frigate_url: http://192.168.22.2:5000
  mqtt_server: 192.168.22.2
  mqtt_auth: false
  mqtt_username: 
  mqtt_password: 
  main_topic: frigatea
  camera:
    - barn_inside
  object: bird
classification:
  model: model.tflite
  threshold: 0.6
webui:
  port: 7766
  host: 0.0.0.0

 

 

 

Edited by dopeytree
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.

×
×
  • Create New...