Never investigated why my automation fires to often. The one from the blueprint is working perfectly.
Decided not to need a critical notification when someone is at the front door.
Will retry the critical notification when I add my indoor cam.
This thread is nuts itās so popular!
Cannot find anything or track stuff so easy
So I created a discord for anyone who whatās to discuss
Totally unofficial and unconnected!
Iām getting a lot of melting vids, I donāt see this happen in my DVR software.
Like this:
I receive regular error messages like these in the logs.
[h264 @ 0x55d9ea237d00] error while decoding MB 38 2, bytestream -31
Last message repeated 2 times
[h264 @ 0x55d9ea237d00] cabac decode of qscale diff failed at 16 5
[h264 @ 0x55d9ea237d00] error while decoding MB 16 5, bytestream 20315
[h264 @ 0x55d9ea237d00] left block unavailable for requested intra4x4 mode -1
[h264 @ 0x55d9ea237d00] error while decoding MB 0 3, bytestream 243949
[h264 @ 0x5644b7580d00] error while decoding MB 22 0, bytestream -32
[h264 @ 0x5644b7580d00] error while decoding MB 22 0, bytestream -29
[h264 @ 0x5644b7580d00] left block unavailable for requested intra mode
[h264 @ 0x5644b7580d00] error while decoding MB 0 2, bytestream 25945
[h264 @ 0x5644b7580d00] top block unavailable for requested intra mode -1
[h264 @ 0x5644b7580d00] error while decoding MB 25 0, bytestream 276116
[h264 @ 0x55e600204d00] error while decoding MB 11 8, bytestream -36
[h264 @ 0x55e600204d00] top block unavailable for requested intra mode -1
[h264 @ 0x55e600204d00] error while decoding MB 23 0, bytestream 39859
Using Coral:
coral:
type: edgetpu
Configuration is as follows for this particular camera:
drive:
ffmpeg:
inputs:
- path: rtsp://REDACTED
roles:
- clips
- record
- detect
width: 640
height: 480
fps: 5
motion:
mask:
- 391,480,640,480,640,422,385,427
snapshots:
enabled: True
timestamp: False
bounding_box: False
crop: True
height: 270
retain:
default: 10
clips:
enabled: True
pre_capture: 5
post_capture: 5
retain:
default: 10
Brand of camera?
Had same issue when using RTSP streams from REOINK camsā¦ Switched to RTMP and it went away.
1 Like
Is anyone successful in using the USB Coral with Home Assistant OS running in Virtualbox on macOS? I just got mine today and canāt get Frigate to find it. It always shows this:
frigate.edgetpu INFO : Attempting to load TPU as usb
frigate.edgetpu INFO : No EdgeTPU detected.
Iām running a 2013 Mac Pro and followed the instructions here to set up the Coral but itās still showing in System Information as Composite Device from 0x1a6e (Global Unichip Corp.).
I set up Virtual Box with the USB Filter as well as found this article that talks about setting up a filter for a Google device. I tried that also but it didnāt seem to make a difference.
I can tell that Home Assistant sees it by using the SSH add-on and typing lsusb I get Bus 002 Device 003: ID 1a6e:089a and that disappears if I unplug the Coral.
Lastly, the Frigate addon works fine if I switch to a CPU detector instead of the EdgeTPU detector. Iāve tried with Protection Mode enabled and disabled, doesnāt seem to make a difference. Anyone have any ideas of other things to check, either macOS or in Home Assistant?
Relevant frigate.yml:
detectors:
coral:
type: edgetpu
device: usb
Logs:
* Starting nginx nginx
...done.
frigate.app INFO : Creating directory: /tmp/cache
frigate.app WARNING : Camera front_yard_1 has rtmp enabled, but rtmp is not assigned to an input.
Starting migrations
peewee_migrate INFO : Starting migrations
There is nothing to migrate
peewee_migrate INFO : There is nothing to migrate
frigate.app INFO : Camera processor started for front_yard_1: 41
frigate.app INFO : Capture process started for front_yard_1: 43
frigate.mqtt INFO : MQTT connected
detector.coral INFO : Starting detection process: 36
frigate.edgetpu INFO : Attempting to load TPU as usb
Process detector:coral:
frigate.edgetpu INFO : No EdgeTPU detected.
Traceback (most recent call last):
File "/usr/local/lib/python3.8/dist-packages/tflite_runtime/interpreter.py", line 152, in load_delegate
delegate = Delegate(library, options)
File "/usr/local/lib/python3.8/dist-packages/tflite_runtime/interpreter.py", line 111, in __init__
raise ValueError(capture.message)
ValueError
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/usr/lib/python3.8/multiprocessing/process.py", line 315, in _bootstrap
self.run()
File "/usr/lib/python3.8/multiprocessing/process.py", line 108, in run
self._target(*self._args, **self._kwargs)
File "/opt/frigate/frigate/edgetpu.py", line 124, in run_detector
object_detector = LocalObjectDetector(tf_device=tf_device, num_threads=num_threads)
File "/opt/frigate/frigate/edgetpu.py", line 63, in __init__
edge_tpu_delegate = load_delegate('libedgetpu.so.1.0', device_config)
File "/usr/local/lib/python3.8/dist-packages/tflite_runtime/interpreter.py", line 154, in load_delegate
raise ValueError('Failed to load delegate from {}\n{}'.format(
ValueError: Failed to load delegate from libedgetpu.so.1.0
frigate.watchdog INFO : Detection appears to have stopped. Exiting frigate...
frigate.app INFO : Stopping...
frigate.record INFO : Exiting recording maintenance...
frigate.object_processing INFO : Exiting object processor...
frigate.events INFO : Exiting event processor...
frigate.events INFO : Exiting event cleanup...
frigate.watchdog INFO : Exiting watchdog...
2 Likes
Reolink. How did you switch to RTMP?
Try this url:
rtmp://ip adress/bcs/channel0_main.bcs?channel=0&stream=0&user=usenam&password=password
For substream use channel0_sub
Works in VLC, but Frigate throws a fit:
frigate.video INFO : drive: ffmpeg sent a broken frame. memoryview assignment: lvalue and rvalue have different structures
frigate.video INFO : drive: ffmpeg process is not running. exiting capture thread...
drive:
ffmpeg:
inputs:
- path: rtmp://REDACTED/bcs/channel0_sub.bcs?channel=0&stream=0&user=REDACTED&password=REDACTED
roles:
- clips
- record
- detect
width: 640
height: 480
fps: 5
motion:
mask:
- 391,480,640,480,640,422,385,427
snapshots:
enabled: True
timestamp: False
bounding_box: False
crop: True
height: 270
retain:
default: 10
clips:
enabled: True
pre_capture: 5
post_capture: 5
retain:
default: 10
Try adding this to ffmpeg
ffmpeg:
....
....
input_args:
- '-avoid_negative_ts'
- make_zero
- '-fflags'
- nobuffer
- '-flags'
- low_delay
- '-strict'
- experimental
- '-fflags'
- +genpts+discardcorrupt
- '-use_wallclock_as_timestamps'
- '1'
1 Like
you sir are a saint.
Thanks
Glad to helpā¦ 
Recently changed to a google coral, but the inference speed is still at 30. Are there any ways to improve it?
Picked one up, itāll be here today. Iāll be attempting to load HassOS+Frigate on it and run it that way for a bit to see how well it works. My last venture with an Atom x5 was not well so curious to see how well this Celeron can handle both. Coral will be later this month after I run it for a bit to evaluate performance. What a great find for $39, thank you!!
@unrly you bought t430 TC for 39$? Where , I will buy few.
This device is awesome, runs 4W and HA Supervised is using ~5% of the CPU.
1 Like
Picked it up on eBay, normal price is $45. Refurbished, so Iāll be curious to see the condition it comes in.
Glad to hear youāre still liking it! That is impressive!
Edit: HA up and running on the emmc, works fantastic. Just have Frigate running for now. Iāll restore my VM snapshot soon. Inference speeds are similar to a 6th gen i3 in VirtualBox, about 350ms for 1 cam at 1080p.
2 Likes
if I use Frigate in a docker container, is there a way to retrieve the following image from within hassio? Like a button on Lovelace that will open a pop-up of the image:
Add it as camera
1 Like
Im not sure, but i think nuc wifi adapter is using m.2 E key? Why dont u just remove wifi adapthee and use coral m.2 E? Would it work with rc2 hassos?
Anyone tried chuwi larkbox? It has 128gb emmc with m.2 2242 expansion slot. Im not sure intel UHD in the celeron is good enuff for h254 decoding. If the m.2 expansion does work with coral m.2 in upcoming HassOs it could be a perfect cheap smallbox for frigate and hass. Any idea?
On an ODROID-N2+ with just 4 camera feeds I get 50% cpu usage even when using a Coral USB device.
Is there any capability of hardware accelerated h264 decoding on this platform?