Which version are you using Ender?
I think that docker-compose file is from the original cpu-only version. There’s an example docker-compose in the latest readme files. Only one container
frigate:
container_name: frigate
restart: unless-stopped
privileged: true
image: frigate:latest
volumes:
- /dev/bus/usb:/dev/bus/usb
- <path_to_config>:/config
ports:
- "5000:5000"
environment:
RTSP_PASSWORD: "password"
I did use the original one, but I added the usb passthrough volume and the privileged: true to it to make it work with the newer version. When I start up two containers simultaneously with docker compose, one of them fails to read the usb accelerator. Wondering if I need to put in some kind of time offset into starting up the containers, or if it’s a docker limitation where you can’t pass through the same usb device to two containers…
Sorry, my compose file listed above was from the CPU version… Here is one I think will work with Coral. I have moved over to my synology and am using the synology tools to manage the container now, but this compose was working on my old host, though there were other issues with compatibility with my cameras which were since solved:
version: "3"
services:
frigate:
container_name: frigate
restart: unless-stopped
privileged: true
image: frigate:latest
volumes:
- /dev/bus/usb:/dev/bus/usb
- /root/frigate/config:/config
ports:
- 5000:5000
environment:
- RTSP_PASSWORD=PASSWORD
- RTSP_PASSWORD_RIGHT=PASSWORD
- RTSP_PASSWORD_LEFT=PASSWORD
- RTSP_PASSWORD_FRONT=PASSWORD
- RTSP_PASSWORD_BACK=PASSWORD
Most of the important stuff is now in the config file rather than the docker compose/command line. If your cameras all use the same password, you can have just one password line… I had 5 passwords used for various cameras in different connection ways…
You can’t have more than 1 container with the Coral. You have to configure all cameras to be processed with a single container. Just duplicate the configuration block under cameras for additional cameras.
Thanks, @scstraus, @blakeblackshear. I thought I had seen a post above saying multiple cameras required multiple containers. I guess that was for the CPU version. I’ll merge it all together and give it a shot in one container.
these vsync options have significantly decreased the queue full messages i was getting across 6 cameras, thanks a lot for sharing
I just got my coral delivered today and have been trying to set this up, thoguht I was close but reading more of the thread it sounds like some configs are for cpu while some are for tpu. Could someone please let me know if they see what is wrong with my config? I also not sure where the label_map file is, Ive downloaded this from above ‘http://download.tensorflow.org/models/object_detection/ssdlite_mobilenet_v2_coco_2018_05_09.tar.gz’.
this is the error i get
ERROR: pull access denied for frigate, repository does not exist or may require 'docker login': denied: requested access to the resource is denied
this is my config
root@cor:/opt/frigate# cat config/config.yml
web_port: 5000
mqtt:
host: 10.1.99.113
topic_prefix: frigate
# user: username # Optional -- Uncomment for use
# password: password # Optional -- Uncomment for use
cameras:
unifi:
rtsp:
user: blah
host: 10.2.2.60
port: 7447
# values that begin with a "$" will be replaced with environment variable
password: blahblah
path: /LKNVLKSDNVKLN
################
## Optional mask. Must be the same dimensions as your video feed.
## The mask works by looking at the bottom center of the bounding box for the detected
## person in the image. If that pixel in the mask is a black pixel, it ignores it as a
## false positive. In my mask, the grass and driveway visible from my backdoor camera
## are white. The garage doors, sky, and trees (anywhere it would be impossible for a
## person to stand) are black.
################
# mask: back-mask.bmp
################
# Allows you to limit the framerate within frigate for cameras that do not support
# custom framerates. A value of 1 tells frigate to look at every frame, 2 every 2nd frame,
# 3 every 3rd frame, etc.
################
take_frame: 1
################
# Optional hardware acceleration parameters for ffmpeg. If your hardware supports it, it can
# greatly reduce the CPU power used to decode the video stream. You will need to determine which
# parameters work for your specific hardware. These may work for those with Intel hardware that
# supports QuickSync.
################
# ffmpeg_hwaccel_args:
# - -hwaccel
# - vaapi
# - -hwaccel_device
# - /dev/dri/renderD128
# - -hwaccel_output_format
# - yuv420p
regions:
- size: 1920
x_offset: 0
y_offset: 0
min_person_area: 5000
threshold: 0.8
# - size: 400
# x_offset: 350
# y_offset: 250
# min_person_area: 2000
# threshold: 0.5
# - size: 400
# x_offset: 750
# y_offset: 250
# min_person_area: 2000
# threshold: 0.5
this is my compose file
root@cor:/opt/frigate# cat docker-compose.yml
version: "3"
services:
frigate1:
container_name: tensorflow
restart: unless-stopped
image: frigate
volumes:
- /opt/frigate/config:/config:ro
- /lab/debug:/lab/debug:rw
- /opt/ssdlite_mobilenet_v2_coco_2018_05_09/frozen_inference_graph.pb:/frozen_inference_graph.pb:ro
ports:
- 5000:5000
environment:
- RTSP_URL=rtsp://10.2.2.2:7447/SVKMNSIFVHIFNSIVNS
- MQTT_HOST=10.1.99.113
- MQTT_TOPIC_PREFIX=cameras/1
- DEBUG=0
Im running this on an ubuntu 18 vm thats on a dell r720 - 16 CPUs x Intel® Xeon® CPU E5-2690 0 @ 2.90GHz
I am usiing unifi cameras…going from the above thread i just used anything for UN PW sense its not requred, and the path i filled in with random charectors for this post.
i did run docker build earlier…does it take a long time?
its sitting here, i stopped it last time thinking it was running the container i wanted but now maybe its actually installing still…
Removing intermediate container 436aedafb599
---> 1919f9873417
Step 6/24 : RUN apt-get -qq update && apt-get -qq install --no-install-recommends -y python3 ffmpeg build-essential cmake unzip pkg-config libjpeg-dev libpng-dev libtiff-dev libavcodec-dev libavformat-dev libswscale-dev libv4l-dev libxvidcore-dev libx264-dev libgtk-3-dev libatlas-base-dev gfortran python3-dev libusb-1.0-0 python3-pip python3-pil python3-numpy libc++1 libc++abi1 libunwind8 libgcc1 libva-drm2 libva2 i965-va-driver vainfo && rm -rf /var/lib/apt/lists/*
---> Running in d7799c7a759c
i hate that i fight with something for hours, finally give up and ask for help…and then see what it is im doing that is dumb. sorry about that.
Step 6/24 argh
if someone can help with where the label_map is i should be able to figure this out. thank you
Hi Darbos
If you are using the latest version (or a recent version), the model and labels are already included. The build takes approx an hour on a rpi4 from memory, so shouldn’t be taking that long. You don’t need to build it though? Blake has uploaded images to docker hub. Also, see the readme on the latest version for a docker-compose example for use with coral
Hope that helps a little bit at least
Thanks Onion. I think I might be missing a link or something, I have been following the directions at the top of this post
‘https://github.com/blakeblackshear/frigate/blob/700bd1e3efb59c1b01bebab3b8344e24aebcf657/README.md’
it looks like i need to build first then run the container according to the directions, is there a link to the pre-built containers?
Ive made some progress, it looks like im stuck on accessing the coral though. im digging into that now.
root@cor:/opt/frigate# docker logs -f 4f019bb9b091
On connect called
Traceback (most recent call last):
File "detect_objects.py", line 99, in <module>
main()
File "detect_objects.py", line 57, in main
prepped_frame_queue
File "/opt/frigate/frigate/object_detection.py", line 32, in __init__
self.engine = DetectionEngine(PATH_TO_CKPT)
File "/usr/local/lib/python3.6/dist-packages/edgetpu/detection/engine.py", line 72, in __init__
super().__init__(model_path)
File "/usr/local/lib/python3.6/dist-packages/edgetpu/basic/basic_engine.py", line 40, in __init__
self._engine = BasicEnginePythonWrapper.CreateFromFile(model_path)
RuntimeError: Error in device opening (/sys/bus/usb/devices/1-1)!
HI, Just for info I first tried with ubuntu 18 VM ESXI host and couldn’t get the coral to work. Moved to Ubuntu 18 on a standalone machine and worked straight away both with the Google Coral Examples and the frigate docker images.
You can try this google coral example to test if your setup will work with your Google Coral if you haven’t already:
https://coral.withgoogle.com/docs/accelerator/get-started/
Cool, that looks like the latest readme. The latest release is tagged v0.2.0 https://github.com/blakeblackshear/frigate/tree/v0.2.0
Is that the version you built?
Prebuilt images are here https://hub.docker.com/r/blakeblackshear/frigate
Looks like you are getting there though. Not had that error myself. I just used the /dev/bus/usb
Thanks man. That first link is the correct one, what I have been working with. Anychance any one has seen this error? It looks like its running now but after a few minutes it spits out this error, and looks like it goes back to runnin? but i also cant access to web url
root@cor:/opt/frigate# docker logs -f frigate
On connect called
Traceback (most recent call last):
File "detect_objects.py", line 99, in <module>
main()
File "detect_objects.py", line 57, in main
prepped_frame_queue
File "/opt/frigate/frigate/object_detection.py", line 32, in __init__
self.engine = DetectionEngine(PATH_TO_CKPT)
File "/usr/local/lib/python3.6/dist-packages/edgetpu/detection/engine.py", line 72, in __init__
super().__init__(model_path)
File "/usr/local/lib/python3.6/dist-packages/edgetpu/basic/basic_engine.py", line 40, in __init__
self._engine = BasicEnginePythonWrapper.CreateFromFile(model_path)
RuntimeError: Error in device opening (/sys/bus/usb/devices/1-2)!
On connect called
root@cor:/opt/frigate# cat docker-compose.yml
frigate:
container_name: frigate
restart: unless-stopped
privileged: true
image: frigate:latest
volumes:
- /dev/bus/usb:/dev/bus/usb
- /opt/frigate/config:/config
ports:
- "5000:5000"
root@cor:/opt/frigate# cat config/config.yml
web_port: 5000
mqtt:
host: 10.1.99.113
topic_prefix: frigate
cameras:
unifi:
rtsp:
host: 10.x.x.x
port: 7447
path: /longunifistring
user: this
password: could
regions:
- size: 1920
x_offset: 0
y_offset: 0
min_person_area: 5000
threshold: 0.8
Your size can’t be 1920 because your image is only 1080 pixels tall. Your region is outside the bounds of your image.
Also, you will be much better off using a substream that is a lower resolution where your region size is closer to 300. There is no improvement in accuracy for using a 1080p stream if your region size is greater than 300. All regions are resized to 300 before detection is run, so you are just using extra CPU cycles to resize the image down. Rule of thumb is to choose a resolution that gets your smallest region as close to 300 as possible.
OK thank you. I just tried to change it to 1080 and got the same error. I actually just copied that from a person above that is using unifi as well. It looks to me like it cant connect to the usb, based on the last line in the error. however i can see it fine device (see below).
Just seen the second post, thats good info i assumed 1080, I will use the unifi stream 640x360 for this then, i assume adjust the size in config to 360 as well.
I havent even really dont anything to the region section yert, i was just trying to get her working first, so anything i need to change there is no issue.
ill make those changes and report back. I do have to go eat dinner but ill be back an a hour or two im not giving up! 
root@cor:/opt/frigate# lsusb
Bus 002 Device 001: ID 1d6b:0003 Linux Foundation 3.0 root hub
Bus 001 Device 005: ID 0e0f:0002 VMware, Inc. Virtual USB Hub
Bus 001 Device 004: ID 0e0f:0002 VMware, Inc. Virtual USB Hub
Bus 001 Device 003: ID 1a6e:089a Global Unichip Corp. <–this is the same name/how it looks in vmware when i add it to the vm.
Bus 001 Device 002: ID 0e0f:0003 VMware, Inc. Virtual Mouse
Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
root@cor:/opt/frigate#
Doing this through VMWare is tricky… the Coral likes to change Hardware IDs. There is a website out there that walks you through it, and will give you both hardware IDs to use, but don’t be surprised if you have to keep changing the USB device filters for passing into the VM to get it to recognize and work.
Honestly, I gave up on it after awhile (thankfully the RP4 came in.)
Also side note, I just read some article saying Hikvision and Dahua cams are now banned for import in the US.
https://ipvm.com/reports/aug-13-2019
Alright, I really need some help PLEASE. Ive spent the entire day on this now. I cant figure out the pre-built or manual install. after i do docker pull what do i need to do for the pre-built docker?
for the manual install here is my configs and output:
(venv) root@ai:/opt/frigate# cat docker-compose.yml
frigate:
container_name: frigate:latest
restart: unless-stopped
privileged: true
image: frigate:latest
volumes:
- /dev/bus/usb:/dev/bus/usb
- /opt/frigate:/config
ports:
- "5000:5000"
(venv) root@ai:/opt/frigate# cat config/config.yml
web_port: 5000
mqtt:
host: 10.1.99.113
topic_prefix: frigate
# user: username # Optional -- Uncomment for use
# password: password # Optional -- Uncomment for use
cameras:
unifi:
rtsp:
user: viewer
host: 10.2.2.2
port: 7447
# values that begin with a "$" will be replaced with environment variable
# password: $RTSP_PASSWORD
path: /longunifistring
################
## Optional mask. Must be the same dimensions as your video feed.
## The mask works by looking at the bottom center of the bounding box for the detected
## person in the image. If that pixel in the mask is a black pixel, it ignores it as a
## false positive. In my mask, the grass and driveway visible from my backdoor camera
## are white. The garage doors, sky, and trees (anywhere it would be impossible for a
## person to stand) are black.
################
# mask: back-mask.bmp
################
# Allows you to limit the framerate within frigate for cameras that do not support
# custom framerates. A value of 1 tells frigate to look at every frame, 2 every 2nd frame,
# 3 every 3rd frame, etc.
################
# take_frame: 1
################
# Optional hardware acceleration parameters for ffmpeg. If your hardware supports it, it can
# greatly reduce the CPU power used to decode the video stream. You will need to determine which
# parameters work for your specific hardware. These may work for those with Intel hardware that
# supports QuickSync.
################
# ffmpeg_hwaccel_args:
# - -hwaccel
# - vaapi
# - -hwaccel_device
# - /dev/dri/renderD128
# - -hwaccel_output_format
# - yuv420p
regions:
- size: 360
x_offset: 0
y_offset: 300
min_person_area: 5000
threshold: 0.5
# - size: 400
# x_offset: 350
# y_offset: 250
# min_person_area: 2000
# threshold: 0.5
# - size: 400
# x_offset: 750
# y_offset: 250
# min_person_area: 2000
# threshold: 0.5
(venv) root@ai:/opt/frigate# sudo docker run --rm \
> --privileged \
> -v /dev/bus/usb:/dev/bus/usb \
> -v /opt/frigate:/config:ro \
> -p 5000:5000 \
> blakeblackshear/frigate
standard_init_linux.go:211: exec user process caused "exec format error"
(venv) root@ai:/opt/frigate# docker logs frigate
(venv) root@ai:/opt/frigate# docker-compose up -d
Pulling frigate (frigate:latest)...
ERROR: The image for the service you're trying to recreate has been removed. If you continue, volume data could be lost. Consider backing up your data before continuing.