But definitely some docker is running inside HAOS, I thought you have run it with HAOS since I’ve found this topic on this forum, what I have available is a Portrainer, however I’m unsure how to manually add your package.
I’m really interested to set up a good NVR, unfortunately Frigate (the one with nice features and HAOS integration) has horrible performance. MontionEye has good performance but very limited features, thus I have high hopes in yours
Unfortunately Object detection with Coral Edge TPU (USB) doesn’t work.
During container deployment I used the recommended parameter: -v /dev/bus/usb:/dev/bus/usb --privileged \
and in Object detection config:
object_detection:
type: edgetpu
Motion gets detected, but objects - not.
In case I remove
type: edgetpu
Objects are detected as usually.
I must say that Coral device is functioning properly, as I use Frigate too. But Viseron is much better due to lower resources usage & more accurate object detection, thus I wanted to optimize it with EdgeTPU (but didn’t succeed). Any advice is appreciated.
Thanks!
Hi, I’m a newbie developer trying to set up NVR environment for the new project.
I’m wondering if it’s possible to connect to a camera on Android phone or Android robot using Viseron from Ubuntu OS, or if it only supports certain type of CCTV cameras? I’m trying to use another app called DroidCam on Android phone to connect to Viseron. (The Android and Viseron are on the same local network)
below is the config.yaml file. I tried testing with and without leaving the ‘width’, ‘height’, ‘fps’ section empty.
# See the README for the full list of configuration options.
cameras:
- name: samsung galaxy note8
host: 192.168.0.183
port: 4747
username: myusername
password: mypassword
path: http://192.168.0.183:4747/
width: 1920
height: 1080
fps: 30
# MQTT is optional
#mqtt:
# broker: <ip address or hostname of broker>
# port: <port the broker listens on>
# username: <if auth is enabled>
# password: <if auth is enabled>
my run logs
[s6-init] making user provided files available at /var/run/s6/etc...exited 0.
[s6-init] ensuring user provided files have correct perms...exited 0.
[fix-attrs.d] applying ownership & permissions fixes...
[fix-attrs.d] done.
[cont-init.d] executing container initialization scripts...
[cont-init.d] 10-adduser: executing...
************************ UID/GID *************************
User uid: 0
User gid: 0
************************** Done **************************
[cont-init.d] 10-adduser: exited 0.
[cont-init.d] 20-gid-video-device: executing...
[cont-init.d] 20-gid-video-device: exited 0.
[cont-init.d] 30-edgetpu-permission: executing...
************** Setting EdgeTPU permissions ***************
Coral Vendor IDs:
"1a6e"
"18d1"
No EdgeTPU USB device was found
************************** Done **************************
[cont-init.d] 30-edgetpu-permission: exited 0.
[cont-init.d] 40-set-env-vars: executing...
****** Checking for hardware acceleration platforms ******
OpenCL cannot be used
VA-API cannot be used
CUDA cannot be used
*********************** Done *****************************
[cont-init.d] 40-set-env-vars: exited 0.
[cont-init.d] 50-check-if-rpi: executing...
********** Checking if we are running on an RPi **********
Not running on any supported RPi
*********************** Done *****************************
[cont-init.d] 50-check-if-rpi: exited 0.
[cont-init.d] 55-check-if-jetson: executing...
****** Checking if we are running on a Jetson Board ******
Not running on any supported Jetson board
*********************** Done *****************************
[cont-init.d] 55-check-if-jetson: exited 0.
[cont-init.d] 60-ffmpeg-path: executing...
****************** Getting FFmpeg path *******************
FFmpeg path: /usr/local/bin/ffmpeg
*********************** Done *****************************
[cont-init.d] 60-ffmpeg-path: exited 0.
[cont-init.d] done.
[services.d] starting services
[services.d] done.
[2021-09-13 14:15:57] [INFO ] [viseron ] - -------------------------------------------
[2021-09-13 14:15:57] [INFO ] [viseron ] - Initializing...
[2021-09-13 14:15:57] [INFO ] [viseron ] - Initializing NVR threads
[2021-09-13 14:15:57] [ERROR ] [viseron.camera.stream.samsung_galaxy_note8] - rtsp://*****:*****@192.168.0.183:4747http://192.168.0.183:4747/: Invalid data found when processing input
[2021-09-13 14:15:57] [ERROR ] [viseron ] - Failed to initialize camera samsung galaxy note8: FFprobe could not connect to stream. Output: {'error': {'code': -1094995529, 'string': 'Invalid data found when processing input'}}, Command: ffprobe -hide_banner -loglevel error -print_format json -show_error -show_streams rtsp://peter:[email protected]:4747http://192.168.0.183:4747/
[2021-09-13 14:15:57] [INFO ] [viseron ] - Initialization complete
[cont-finish.d] executing container finish scripts...
[cont-finish.d] done.
[s6-finish] waiting for services.
[s6-finish] sending all processes the TERM signal.
[s6-finish] sending all processes the KILL signal and exiting.
that seems ok. so im currently using zoneminder with mlapi (machine learning api). I use a pcie based coral and i can see it in the mlapi docker as such:
crwxrwxrwx 1 root root 120, 0 Aug 28 15:46 apex_0
try in the viseron docker >> chmod 777 /dev/bus/usb to give full permission then restart viseron
It seems that generally, Coral works with Viseron, though Object detection accuracy is very poor.
I managed to make it work using the same config (confidence, sensitivity, other parameters) only in 1 camera out of 4, the rest shows motion detection boundaries where objects (“Person” is the subject of interest) are not detected at all.
I wonder if it’s configurable.
Frigate uses the same technique, but detection is better.
I remember the same story with Viseron 1.7.x, when author @roflcoopter recommended me to use 1.8.0 beta version that was optimized for particular YOLO detector that was able to work with smaller objects. Maybe that is the reason…
Are you still experiencing problems?
I believe frigate employs a little different technique for object detection which might make it more accurate, will look into it a bit more
Yes, I tried using Darknet vs. EdgeTPU vs. Deepstack and here’s what I came up with:
Darknet detects objects with best sensitivity, however sometimes trash bags are detected as humans (is it possible to load custom models in Darknet within Viseron container?)
Deepstack works, but detection accuracy & speed is much worse compared to Darknet, though GPU version of Deepstack loads CPU less and thus looks more attractive in terms of resource-intensiveness.
EdgeTPU literally sucks… It detects objects (in my case - People) when they pose & stand still in order to look as an ideal model.
In my case CCTV cameras are installed on 3rd floor of the building, and it seems to make harder for AI to detect humans especially when they approach closer to the building. Maybe there is some special trained dataset for such scenes (when camera is installed above human’s heads).
Please advise how to improve accuracy and possibly utilize hardware for object detection.
P.S. For your reference Frigate works much worse in case of EdgeTPU compared to your product, if I use Darknet engine (I mean sensitivity of detection, some clearly visible persons may not be detected by Frigate, when Viseron detects with 80% confidence).
If you have custom Darknet models you can just mount them in the container and point towards them instead of the default ones.
I have plans on implementing some kind of transfer learning functionality, which would allow you to improve the accuracy of the model by supplying training data that is specific to each camera, but its hard to find the time to do everything!
Hi, thanks for the reply coopter.
I removed the ‘path’ attribute like you suggested, but still this python error occurs.
Traceback (most recent call last):
File "/usr/lib/python3.8/runpy.py", line 185, in _run_module_as_main
mod_name, mod_spec, code = _get_module_details(mod_name, _Error)
File "/usr/lib/python3.8/runpy.py", line 144, in _get_module_details
return _get_module_details(pkg_main_name, error)
File "/usr/lib/python3.8/runpy.py", line 111, in _get_module_details
__import__(pkg_name)
File "/src/viseron/__init__.py", line 8, in <module>
from viseron.config import CONFIG, NVRConfig, ViseronConfig
File "/src/viseron/config/__init__.py", line 114, in <module>
CONFIG = VISERON_CONFIG_SCHEMA(raw_config)
File "/usr/local/lib/python3.8/dist-packages/voluptuous/schema_builder.py", line 272, in __call__
return self._compiled([], data)
File "/usr/local/lib/python3.8/dist-packages/voluptuous/validators.py", line 205, in _run
return self._exec(self._compiled, value, path)
File "/usr/local/lib/python3.8/dist-packages/voluptuous/validators.py", line 287, in _exec
raise e if self.msg is None else AllInvalid(self.msg, path=path)
File "/usr/local/lib/python3.8/dist-packages/voluptuous/validators.py", line 285, in _exec
v = func(path, v)
File "/usr/local/lib/python3.8/dist-packages/voluptuous/schema_builder.py", line 594, in validate_dict
return base_validate(path, iteritems(data), out)
File "/usr/local/lib/python3.8/dist-packages/voluptuous/schema_builder.py", line 432, in validate_mapping
raise er.MultipleInvalid(errors)
voluptuous.error.MultipleInvalid: required key not provided @ data['cameras'][0]['path']
Could you discover what is wrong from this traceback? Thanks
Hello, I’ve still not given up on trying to connect my Android phone and record video.
Finally I’ve succeeded in initializing & pairing my phone with Viseron! I used RTSP protocol app(IP webcam) to do this. Then I’m trying to record video that is being streamed. I tuned yaml to save recording as mp4, but viseron keeps saving only jpg file on /recordings. Is there a way to record mp4 video?
