0.15.0 Beta 2 #15097
Replies: 32 comments 205 replies
-
|
I have an issue with Beta2 not starting. UI partially loads but no cameras and all menu items give the spinny wheel
Reverting to beta1 and it starts ok |
Beta Was this translation helpful? Give feedback.
-
|
I can confirm this also fixed the problem here. Is this something odd happening with beta 2 that can be fixed or going forward is a going to be a requirement to perform os config changes? |
Beta Was this translation helpful? Give feedback.
-
|
Not sure, if the problem truly started at the round robin then it doesn't make sense since the number of max SHM was already increased before that. The only change I can think of is that the SHM frames are initialized in the main process in that commit, so it might be as simple as redoing that |
Beta Was this translation helpful? Give feedback.
-
|
I run dev build always and didn't hit this but seems like indeed starting with that commit the openfiles count in my setup related to shared mem gets doubled: beta 1, ( average around 85k open files on startup): after that commit: (average around 170k open files on startup): |
Beta Was this translation helpful? Give feedback.
-
|
@ediguidedog @jameslivulpi can you check if this is still occurring on the latest commit? |
Beta Was this translation helpful? Give feedback.
-
|
Yes and no. but I do need to keep the following in the docker-compose.yaml or I get the same too many files message |
Beta Was this translation helpful? Give feedback.
-
|
I have a script that catches events via API, in this new beta 2 version it is giving the message: {"success":false,"message":"Event not found"}
I'm using the API to get manual events |
Beta Was this translation helpful? Give feedback.
-
|
The error message is indicating that the event ID you're passing to the API is invalid, so we'll need more information. The code to retrieve an event by event_id didn't change between beta 1 and beta 2. |
Beta Was this translation helpful? Give feedback.
-
|
The event is created perfectly and given me an ID:
I take this ID and play it in the browser and this message appears:
Manual event that I create also no longer appears in "Explore":
My camera code in the frigate settings looks like this: In the previous version of the Beta it worked and in this new version BETA 2 the problems I just described occur, I completely reinstalled the frigate deleting recordings and files and the problem also continues |
Beta Was this translation helpful? Give feedback.
-
|
I don't believe anything changed in this regard. Please show the request that you making to create the event |
Beta Was this translation helpful? Give feedback.
-
|
Nevermind I see the issue, fixed in #15100 |
Beta Was this translation helpful? Give feedback.
-
|
I create an event through a rest_command in the home assistant: |
Beta Was this translation helpful? Give feedback.
-
|
2GB. Beta 1 only uses around 22MB
…On Wed, 20 Nov 2024 at 14:52, Nicolas Mowen ***@***.***> wrote:
What is your SHM size
—
Reply to this email directly, view it on GitHub
<#15097 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ARP33XTDU5LJ5IUVZ7EJ6ID2BSO37AVCNFSM6AAAAABSEM6FT2VHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTCMZSGM4TKMY>
.
You are receiving this because you commented.Message ID:
***@***.***
com>
|
Beta Was this translation helpful? Give feedback.
-
|
Try setting it to 512MB instead |
Beta Was this translation helpful? Give feedback.
-
|
I have created a survey that we want to use to get feedback from users that have been using the GenAI feature for at least 1 week. The feedback will be very helpful in understanding users use cases and how the feature can be improved. The survey is anonymous: https://docs.google.com/forms/d/e/1FAIpQLSehVvK6oVKzikuKzwlIGz2eiaUwhYQq24ZWy9Iqe5NWSrI2Aw/viewform |
Beta Was this translation helpful? Give feedback.
-
|
I'll do the survey after a week, after an hour it works faultlessly well. But my god this is scary! |
Beta Was this translation helpful? Give feedback.
-
|
beta-2 is a no go for me, config works with 14 but not beta-1 or beta-2. Here is what I pulled from the log, I did try both with the new ffmpeg and also tried adding the path for 5.0 as noted, same result. I'll try to keep hunting down the solution but any assistance is appreciated. |
Beta Was this translation helpful? Give feedback.
-
|
Because I want to cause more problems I'm going to update my Kernel.... 🗡️ :) |
Beta Was this translation helpful? Give feedback.
-
|
And updating my kernel fixed the problem! |
Beta Was this translation helpful? Give feedback.
-
|
Now to figure out what the integration is wonky and not working even though Frigate UI is. |
Beta Was this translation helpful? Give feedback.
-
|
if you're referring to media playback, that is an HA bug |
Beta Was this translation helpful? Give feedback.
-
All good. |
Beta Was this translation helpful? Give feedback.
-
|
After updating to Beta 2 on HAOS Frigate keeps restarting after a few mins. No clear errors and/or patterns why its restarting, but everything on Beta 1 was working well. |
Beta Was this translation helpful? Give feedback.
-
|
|
Beta Was this translation helpful? Give feedback.
-
|
You can probably reduce this significantly due to the changes in beta 2. |
Beta Was this translation helpful? Give feedback.
-
|
I am running HA OS so can't reduce/change shm size. Is this likely why Frigate is eating up more memory than Beta 1? |
Beta Was this translation helpful? Give feedback.
-
Did you figure out any solution? |
Beta Was this translation helpful? Give feedback.
-
Nope.. I was checking my memory usage stats on beta 1 and I was at around 80% utilization. Beta 2 is 90+.. just gonna upgrade my memory to alleviate the issue |
Beta Was this translation helpful? Give feedback.
-
|
I have a config validation error when launching the beta2 (hailo8 related) which prevent frigate from starting: The So far I have no clue as to what's wrong 👀 |
Beta Was this translation helpful? Give feedback.
-
|
There's an issue, this will work in the latest dev images |
Beta Was this translation helpful? Give feedback.
-
|
@NickM-27 I've got the same issue here with my RPI5 with AI HAT+ (Hailo-8L). Is there a possibility to pull the latest dev image? Or is there a beta 3 scheduled soon? |
Beta Was this translation helpful? Give feedback.
-
|
yes, you can pull the latest dev image |
Beta Was this translation helpful? Give feedback.
-
|
FYI I've pulled the dev image after the fix, and everything works fine on my side so far 👍 |
Beta Was this translation helpful? Give feedback.
-
|
how to pull latest dev image? I'am not very expert on docker |
Beta Was this translation helpful? Give feedback.
-
PLEASE, just display the calculated minimum value on /system#storage where SHM use is already displayed. I feel like I have had to second guess this several times now and a clear in your face value would be nice. Maybe something like this: /dev/shm (✅ meets calculated minimum 512mb ) or /dev/shm ( |
Beta Was this translation helpful? Give feedback.
-
|
Perhaps in a later beta. We're still working out bugs and the best sizes at this point. |
Beta Was this translation helpful? Give feedback.
-
|
I would rather it be a best guess in the code than a best guess in the docs. Again all the info is already there, just update the formula with each beta. It also isn't something some of us think about much but that minimum changes every time you change your detect stream, detect resolution or add / remove cameras. It really should be an application dynamic warning. Edit: I recently had a whole thread discussion where the end result was that I should do detect on my main streams and just down scale them to get better snapshots.. I had totally forgot that would Massively change my shm calculation as for many it is something mentioned in the install section of the documentation.. When really it should probably have references in the camera and detect sections to remind you every time you add a camera or setup detect you need to pay attention to it. |
Beta Was this translation helpful? Give feedback.
-
|
Something appears to be MASSIVLY wrong with object tracking in beta 2. Debug view on all of my cameras is rapidly filling up with objects essentially ghosts for ALL MOVING tracked objects on all cameras.
At first I saw this
and through the logo jumping issue was there but it is literally every object the camera has seen persisting. |
Beta Was this translation helpful? Give feedback.
-
|
I am also still running this build and it seems to work fine. |
Beta Was this translation helpful? Give feedback.
-
|
Same here... I am went back on beta1. Testing EDIT: Can confirm that it was "external events" that caused this issue for me. This has been resolved using the fix ff92b13 mentioned above. External events and the motion detection engine now works as expected.
|
Beta Was this translation helpful? Give feedback.
-
|
I noticed that this had something to do with the model_cache. I don't know for sure what the issue is, but when i deleted the model cache and composed up, it worked. As soon as I restarted the frigate container, this issue would start again and I saw two blob files in the model_cache. Not sure if theres some weird caching issues here. |
Beta Was this translation helpful? Give feedback.
-
|
to be clear, this issue has already been fixed and multiple users have confirmed this, it has nothing to do with the model_cache. If you are running in to this without using custom events then the issue is entirely unrelated to the one talked about here |
Beta Was this translation helpful? Give feedback.
-
|
ok ill create a new support discussion. |
Beta Was this translation helpful? Give feedback.
-
|
when i switched to beta2, i keep seeing lots of these in the logs. I switched back to beta1 and no issues. I have 12 of identical cameras, but only frontyard thows those errors. I use TPU + intel i5-10400 |
Beta Was this translation helpful? Give feedback.
-
|
I'd suggest creating a support discussion |
Beta Was this translation helpful? Give feedback.
-
|
doing it now, thanks |
Beta Was this translation helpful? Give feedback.
-
|
A quick little question, this is Beta 2: |
Beta Was this translation helpful? Give feedback.
-
|
Yes |
Beta Was this translation helpful? Give feedback.
-
|
I cannot see notifications tab under settings on my Iphone. I see it on a samsung device. Is this by design or is it just me? |
Beta Was this translation helpful? Give feedback.
-
|
Yes, same url as my samsung. I tried with both safari and chrome. |
Beta Was this translation helpful? Give feedback.
-
|
What version of iOS? |
Beta Was this translation helpful? Give feedback.
-
|
Version 18.1 |
Beta Was this translation helpful? Give feedback.
-
|
Same issue here, works fine on desktop but not on my iphone |
Beta Was this translation helpful? Give feedback.
-
|
I'm also on iOS 18.1. The notifications menu shows if frigate is saved as a PWA but not if it's open in a regular safari browsing session. I'm assuming this is intentional? Edit: Actually, the dialogue box in the notifications menu says notifications can be pushed "when it [frigate] is running in the browser or installed as a PWA", so maybe not intentional. |
Beta Was this translation helpful? Give feedback.
-
|
frigate.api.media Message No recordings found for living_room_cam during the requested time range Edit: The above was from a audio detection 28 days old showing in my explore view but I only have a 14 day retention on detections. When browsing Explore view I quite often find detections that have a Video tab for video that no longer exists (sometimes snapshots with broken images but I THINK that was recently fixed). For the most part this appears to be most often for items that would be classed as Detections not alerts.. The object showing in explore is MUCH older than any of the snapshot, alert or detection time frames. If there is no video to watch and no snapshot retained I don't think the object detection should show in explore anymore? |
Beta Was this translation helpful? Give feedback.
-
|
I'm missing the audio detections in the explore section. I guess this is the case since I'm running the beta. They do appear as an alert however. This is (all relevant) config I'm using: cameras:
back:
audio:
enabled: true
max_not_heard: 15
min_volume: 1800
listen:
- explosion
- siren
- thunder
record:
enabled: true
retain:
days: 7
mode: all
alerts:
retain:
days: 7
pre_capture: 15
post_capture: 15
detections:
retain:
days: 7
pre_capture: 15
post_capture: 15
snapshots:
enabled: true
bounding_box: true
quality: 100
retain:
default: 180
objects:
bbq_grill: 3
waste_bin: 3
bird: 7
bark: 3
siren: 7
explosion: 7
scream: 7
thunder: 7
zones:
zone_terrace:
coordinates: 1,1,1,0.613,0.595,0.672,0.62,1
inertia: 3
loitering_time: 0
zone_garden:
coordinates:
0.399,0.299,0.445,0.119,0.618,0.133,0.731,0.141,0.883,0.353,0.964,0.461,1,0.54,1,0.606,0.59,0.658,0.616,1,0.192,1,0,1,0,0.597,0.169,0.296,0.217,0.296,0.216,0.32,0.336,0.305,0.219,0.603,0.112,0.992,0.193,0.987,0.282,0.597,0.377,0.305
inertia: 3
loitering_time: 0
zone_shed:
coordinates: 0.169,0.097,0.43,0.025,0.43,0.181,0.401,0.294,0.188,0.321
inertia: 3
loitering_time: 0
zone_fence:
coordinates:
0.73,0.141,0.435,0.118,0.435,0.08,0.437,0.015,0.222,0.057,0.22,0,1,0,1,0.536,0.964,0.458
inertia: 3
loitering_time: 0
zone_garden_path:
coordinates: 0.34,0.3,0.374,0.305,0.35,0.37,0.266,0.631,0.183,1,0.111,1,0.219,0.624
loitering_time: 0
objects:
track:
- person
- cat
- dog
- face
- waste_bin
- bird
- raccoon
- umbrella
- bicycle
- squirrel
- rabbit
- bbq_grill
review:
alerts:
required_zones:
- zone_fence
labels:
- explosion
- person
detections:
required_zones:
- zone_terrace
- zone_garden
- zone_shed
- zone_fence
- zone_garden_path
labels:
- cat
- dog
- person
- scream
- siren
- thunder
- bird
- raccoon
- squirrel
- rabbit
- bicycle # to train model
- waste_bin # to train modelAny ideas? |
Beta Was this translation helpful? Give feedback.
-
|
Is it really only me having this issue? |
Beta Was this translation helpful? Give feedback.
-
|
I plan to spend some time looking in to this tomorrow |
Beta Was this translation helpful? Give feedback.
-
Audio events don't appear in explorer for me ether |
Beta Was this translation helpful? Give feedback.
-
|
latest dev with audio fixes shows this in my logs: docker logs frigate | grep -A2 audio
--
2024-12-01 19:47:08.832614559 [2024-12-01 19:47:08] audio.crib WARNING : Failed to create audio event with status code 500
2024-12-01 19:47:21.480054088 [2024-12-01 19:47:21] frigate.api.event ERROR : The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()
2024-12-01 19:47:21.482088304 127.0.0.1 - - [01/Dec/2024:19:47:21 -0500] "POST /api/events/crib/crying/create HTTP/1.1" 500 55 "-" "python-requests/2.32.3" "-"
2024-12-01 19:47:21.484325971 [2024-12-01 19:47:21] audio.crib WARNING : Failed to create audio event with status code 500
2024-12-01 19:47:21.498257244 [2024-12-01 19:47:21] frigate.api.event ERROR : The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()
2024-12-01 19:47:21.499582765 127.0.0.1 - - [01/Dec/2024:19:47:21 -0500] "POST /api/events/crib/crying/create HTTP/1.1" 500 55 "-" "python-requests/2.32.3" "-"
2024-12-01 19:47:21.500668425 [2024-12-01 19:47:21] audio.crib WARNING : Failed to create audio event with status code 500Trying to debug it further. |
Beta Was this translation helpful? Give feedback.
-
|
submitted pr for it #15299 |
Beta Was this translation helpful? Give feedback.
-
|
Trying to dip my toes into openvino and onnx on Intel and got openvino working without issues but hitting some problems with onnx. Here is the detectors section: Here is how I got the model for onnx in 640x640 format: The detector loads but eventually spits out this error: Any thoughts or nudges in the right direction would be very welcomed and thank you! |
Beta Was this translation helpful? Give feedback.
-
|
Yes, ONNX has some issue with internal libraries so using OpenVINO is recommended in that case |
Beta Was this translation helpful? Give feedback.
-
So I have the onnx model, and trying to use that with openvino detector and getting the following error: I used the following python in a virtual environment using super-gradients 3.7.1 to generate the onnx model: Here is the frigate configuration at the moment: Any thoughts? |
Beta Was this translation helpful? Give feedback.
-
|
I think it is the convert_to_onnx that might be the problem, the file should already be in onnx format and that step should not be necessary |
Beta Was this translation helpful? Give feedback.
-
Yup! So using openvino now with onnx model. Waiting on my frigate+ model generation to finish and will play with it more. One last question... I noticed the documentation states multiple GPUs but if I have an iGPU i7-13700 do I configure more than one or just one uses the whole thing? This server is bare metal Debian 12 running just frigate and nothing else. |
Beta Was this translation helpful? Give feedback.
-
|
You can configure more than one if you are seeing skipped fps. It does not mean "more than one GPU" it simply means more than one detector running on the same GPU |
Beta Was this translation helpful? Give feedback.
-
|
Because new features like Semantic Search will not be able to run remotely, should low-power always-on devices like Raspberry Pi be considered unsupported by Frigate as of 0.15 ? |
Beta Was this translation helpful? Give feedback.
-
|
OK, in this case I think we can say that there will be new features added starting from 0.15 which will be unavailable for lower-operating-cost devices like the Raspberry Pi. |
Beta Was this translation helpful? Give feedback.
-
|
They are not unavailable. You can certainly enable Semantic Search and try running it, but the experience and performance will likely be very poor. This isn't something we can improve, which is why the docs and release notes indicate that we don't recommend low power devices like the Pi. |
Beta Was this translation helpful? Give feedback.
-
|
Thanks for clarifying. Out of curiosity, are you aware of any analytics or poll showing the hardware that people run Frigate on? I would suspect that for always-on self-hosted applications like NVR, people would gravitate toward devices that minimize operating cost. |
Beta Was this translation helpful? Give feedback.
-
|
We have a good impression based on providing support to a large amount of users on here, reddit, discord, etc as well as the docker pulls by system architecture. It is clear that a majority of users are at least running on semi modern x86 systems |
Beta Was this translation helpful? Give feedback.
-
NUC clones or surplus Dell/Lenovo/HP mini/micro PCs work well for this, they are quite low power. |
Beta Was this translation helpful? Give feedback.
-
|
Hi services: |
Beta Was this translation helpful? Give feedback.
-
|
In what way is it not working? |
Beta Was this translation helpful? Give feedback.
-
|
In webui i'm not geting frigate+ benefits like frigate 14 and show noting to confirm object or send them to frigate+ |
Beta Was this translation helpful? Give feedback.
-
|
Where are you looking? Have you seen this from the release notes?
|
Beta Was this translation helpful? Give feedback.
-
0.15 doesn't have a dedicated frigate+ tab, submitting to frigate plus is now an option on the details of the review screen (you can submit frames from video) or on the explore screen, you can filter and submit from all of the object detections |
Beta Was this translation helpful? Give feedback.
-
|
I got it |
Beta Was this translation helpful? Give feedback.
-
|
Is there a specific reason why the filter list in the Explore section contains all labels enabled in the configuration? The filter in the Review section only contains the ones that make sense to filter on. |
Beta Was this translation helpful? Give feedback.
-
|
No, I'm referring to the "save search" feature in the search input (the star icon). But it would also apply to bookmarking the page. |
Beta Was this translation helpful? Give feedback.
-
|
Where can I see the star icron? Is this maybe related to semantic search? In that case it explains why I don't see that icon. |
Beta Was this translation helpful? Give feedback.
-
|
Yes, Semantic Search must be enabled to view the search input and save searches. |
Beta Was this translation helpful? Give feedback.
-
|
In that case there might be a use case how it is designed now. Maybe we can make a distinction between both situations. Leave as is with semantic search enabled, and present a clean list when it is not enabled? Again, I'm not arguing. At least now I understand better why it was desnigned in this way. |
Beta Was this translation helpful? Give feedback.
-
|
All good, I don't feel like you're arguing. I'm not sure there's an overly compelling reason to have the behavior be different between enabled/disabled Semantic Search, but maybe the other maintainers have a different opinion. |
Beta Was this translation helpful? Give feedback.
-
|
Not sure if this is related to my storage running low but all my cameras have a red box around them and then eventually frigate restarts and everything is back to normal. Logs attached |
Beta Was this translation helpful? Give feedback.
-
|
Hi! Yes, there is a bug in beta2 that crashes the detection. Use this docker image instead or patch the python file. It is beta2 but with this fix. |
Beta Was this translation helpful? Give feedback.
-
|
Thanks! I'm on HA OS so I'll wait for Beta 3 :) |
Beta Was this translation helpful? Give feedback.
-
|
Im just getting to test the capabilities in beta 2, since I had some issues with deployment. Before I go down a rabbit hole of working out what I might have broken I thought I would ask on here first. are there any known issues with being able to play back media from the "explore" section? For example if I go into Tracked Object Details> Video (for an event that happened 30mins ago), its fine and plays back. If I select something from 2 or 3 days ago, it just hangs? |
Beta Was this translation helpful? Give feedback.
-
mqtt:
host: 192.168.150.4
port: 1883
topic_prefix: frigate
client_id: frigate
user: frigate
password: '123456789'
stats_interval: 300
model:
path: plus://blah
record:
enabled: True
retain:
days: 7
mode: all
alerts:
retain:
days: 30
mode: active_objects
detections:
retain:
days: 30
mode: active_objects
timestamp_style:
format: '%m/%d/%Y %H:%M:%S'
color:
red: 255
green: 255
blue: 255
thickness: 2
effect: shadow
audio:
enabled: true
listen:
- speech
- yell
- howl
- meow
objects:
filters:
dog:
min_score: .7
threshold: .9
cat:
min_score: .65
threshold: .8
face:
min_score: .7
package:
min_score: .65
threshold: .9
license_plate:
min_score: .6
amazon:
min_score: .75
ups:
min_score: .75
fedex:
min_score: .75
person:
min_score: .65
threshold: .85
car:
min_score: .65
threshold: .85
go2rtc:
log:
exec: trace
streams:
ANPR:
- ffmpeg:http://192.168.150.18/flv?port=1935&app=bcs&stream=channel0_main.bcs&user=admin&password=123456789#video=copy#audio=copy#audio=opus
ANPR_sub:
- ffmpeg:http://192.168.150.18/flv?port=1935&app=bcs&stream=channel0_ext.bcs&user=admin&password=123456789#video=copy#audio=copy#audio=opus
backdoor_view:
- rtsp://admin:[email protected]/h264Preview_01_main
- ffmpeg:back#video=h264#audio=aac
backdoor_view_sub:
- rtsp://admin:[email protected]/h264Preview_01_sub
- ffmpeg:back#video=h264#audio=aac
side:
- ffmpeg:http://192.168.150.20/flv?port=1935&app=bcs&stream=channel0_main.bcs&user=admin&password=123456789!#video=copy#audio=copy#audio=opus
side_sub:
- ffmpeg:http://192.168.150.20/flv?port=1935&app=bcs&stream=channel0_sub.bcs&user=admin&password=123456789!#video=copy#audio=copy#audio=opus
backdoor:
- ffmpeg:http://192.168.150.10/flv?port=1935&app=bcs&stream=channel0_main.bcs&user=admin&password=123456789#video=copy#audio=copy#audio=opus
backdoor_sub:
- ffmpeg:http://192.168.150.10/flv?port=1935&app=bcs&stream=channel0_ext.bcs&user=admin&password=123456789#video=copy#audio=copy#audio=opus
office:
- ffmpeg:http://192.168.150.11/flv?port=1935&app=bcs&stream=channel0_main.bcs&user=admin&password=!123456789#video=copy#audio=copy#audio=opus
office_sub:
- ffmpeg:http://192.168.150.11/flv?port=1935&app=bcs&stream=channel0_ext.bcs&user=admin&password=!123456789#video=copy#audio=copy#audio=opus
bell:
- ffmpeg:http://192.168.150.15/flv?port=1935&app=bcs&stream=channel0_main.bcs&user=admin&password=123456789#video=copy#audio=copy#audio=opus
bell_sub:
- ffmpeg:http://192.168.150.15/flv?port=1935&app=bcs&stream=channel0_ext.bcs&user=admin&password=123456789#video=copy#audio=copy#audio=opus
shed:
- ffmpeg:http://192.168.150.12/flv?port=1935&app=bcs&stream=channel0_main.bcs&user=admin&password=123456789#video=copy#audio=copy#audio=opus
shed_sub:
- ffmpeg:http://192.168.150.12/flv?port=1935&app=bcs&stream=channel0_ext.bcs&user=admin&password=123456789#video=copy#audio=copy#audio=opus
steps:
- ffmpeg:http://192.168.150.19/flv?port=1935&app=bcs&stream=channel0_sub.bcs&user=admin&password=123456789#video=copy#audio=copy#audio=opus
steps_sub:
- rtsp://admin:[email protected]/h264Preview_01_sub
spare:
- rtsp://admin:[email protected]/h264Preview_01_main
- ffmpeg:back#video=h264#audio=aac
spare_sub:
- rtsp://admin:[email protected]/h264Preview_01_sub
- ffmpeg:back#video=h264#audio=aac
frontdrive:
- ffmpeg:http://192.168.150.22/flv?port=1935&app=bcs&stream=channel0_main.bcs&user=admin&password=123456789!#video=copy#audio=copy#audio=opus
frontdrive_sub:
- rtsp://admin:[email protected]/h264Preview_01_sub
cameras:
ANPR:
ffmpeg:
hwaccel_args: preset-nvidia-h264
inputs:
- path: rtsp://127.0.0.1:8554/ANPR_sub
input_args: preset-rtsp-restream
roles:
- detect
- audio
- path: rtsp://127.0.0.1:8554/ANPR
input_args: preset-rtsp-restream
roles:
- record
output_args:
record: preset-record-generic-audio-aac
detect:
width: 640
height: 360
fps: 15
objects:
track:
- person
- face
- license_plate
- car
- motorcycle
snapshots:
enabled: true
timestamp: true
bounding_box: true
crop: true
height: 360
retain:
default: 14
zones:
ANPR_Drive:
coordinates: 0,0.957,0.111,0.825,0.731,0.646,1,0.652,1,1,0,1
loitering_time: 0
ANPR_snap:
coordinates: 0.226,0.487,0.237,0.783,0.659,0.667,0.34,0.429
loitering_time: 0
objects: car
motion:
mask: 0.295,0.069,0.703,0.066,0.704,0.023,0.293,0.021
record:
enabled: true
retain:
days: 3
mode: all
alerts:
retain:
days: 30
mode: active_objects
pre_capture: 5
post_capture: 15
detections:
retain:
days: 5
pre_capture: 5
post_capture: 15
review:
alerts:
labels:
- car
- cat
- person
- speech
backdoor:
ffmpeg:
hwaccel_args: preset-nvidia-h264
inputs:
- path: rtsp://127.0.0.1:8554/backdoor_sub
input_args: preset-rtsp-restream
roles:
- detect
- audio
- path: rtsp://127.0.0.1:8554/backdoor
input_args: preset-rtsp-restream
roles:
- record
output_args:
record: preset-record-generic-audio-aac
detect:
width: 640
height: 360
fps: 15
objects:
track:
- person
- face
- license_plate
- dog
- cat
- car
- amazon
- fedex
- ups
- package
- bird
- bicycle
- motorcycle
snapshots:
enabled: true
timestamp: true
bounding_box: true
required_zones:
- backdoor_patio
- backdoor_whole_area
crop: true
height: 360
retain:
default: 14
zones:
backdoor_patio:
coordinates: 1,1,0.737,0.382,0.095,0.652,0,1,0,1
objects:
- person
- face
- license_plate
- dog
- cat
- car
- amazon
- fedex
- ups
- package
- bird
- bicycle
- motorcycle
inertia: 3
backdoor_whole_area:
coordinates: 0,1,1,1,1,0.65,1,0.333,0.565,0.116,0.3,0.095,0.17,0.145,0,0.497
objects:
- person
- face
- license_plate
- dog
- cat
- car
- amazon
- fedex
- ups
- package
- bird
- bicycle
- motorcycle
inertia: 3
motion:
mask: 0.295,0.069,0.703,0.066,0.704,0.023,0.293,0.021
record:
enabled: true
retain:
days: 3
mode: all
alerts:
retain:
days: 30
mode: active_objects
pre_capture: 5
post_capture: 15
detections:
retain:
days: 5
pre_capture: 5
post_capture: 15
review:
alerts:
labels:
- car
- cat
- person
- speech
required_zones:
- backdoor_patio
- backdoor_whole_area
detections:
required_zones:
- backdoor_whole_area
- backdoor_patio
backdoor_view:
ffmpeg:
hwaccel_args: preset-nvidia-h264
inputs:
- path: rtsp://127.0.0.1:8554/backdoor_view_sub
input_args: preset-rtsp-restream
roles:
- detect
- audio
- path: rtsp://127.0.0.1:8554/backdoor_view
input_args: preset-rtsp-restream
roles:
- record
output_args:
record: preset-record-generic-audio-aac
detect:
width: 640
height: 360
fps: 15
objects:
track:
- person
- face
- license_plate
- dog
- cat
- car
- amazon
- fedex
- ups
- package
- bird
- bicycle
- motorcycle
snapshots:
enabled: true
timestamp: true
bounding_box: true
required_zones:
- backdoor_view_patio
- backdoor_view_whole_area
crop: true
height: 360
retain:
default: 14
zones:
backdoor_view_door:
coordinates:
0.572,0.5,0.952,0.714,0.987,0.518,1,0.339,1,0.199,0.959,0.167,0.867,0.267,0.699,0.336,0.685,0.469
inertia: 3
loitering_time: 0
objects:
- person
- face
- license_plate
- amazon
- fedex
- ups
backdoor_view_garden:
coordinates:
0,0.995,0.008,0.439,0.378,0.127,0.574,0.149,0.598,0.075,0.732,0.099,0.894,0.19,0.923,0.48,0.962,0.992
loitering_time: 0
objects:
- person
- cat
- car
motion:
mask: 0.295,0.069,0.703,0.066,0.704,0.023,0.293,0.021
record:
enabled: true
retain:
days: 3
mode: all
alerts:
retain:
days: 30
mode: active_objects
pre_capture: 5
post_capture: 15
detections:
retain:
days: 5
pre_capture: 5
post_capture: 15
review:
alerts:
labels:
- car
- cat
- person
- speech
required_zones: backdoor_view_door
detections:
required_zones: backdoor_view_door
bell:
ffmpeg:
hwaccel_args: preset-nvidia-h264
inputs:
- path: rtsp://127.0.0.1:8554/bell_sub
input_args: preset-rtsp-restream
roles:
- detect
- audio
- path: rtsp://127.0.0.1:8554/bell
input_args: preset-rtsp-restream
roles:
- record
output_args:
record: preset-record-generic-audio-aac
detect:
width: 640
height: 360
fps: 15
objects:
track:
- person
- package
- amazon
- fedex
- ups
snapshots:
enabled: true
timestamp: true
bounding_box: true
crop: true
height: 360
retain:
default: 14
zones:
bell_front:
coordinates: 1,1,1,0.866,0.767,0.684,0.394,0.767,0.031,0.911,0.088,1
inertia: 3
loitering_time: 0
objects:
- package
- person
bell_detection:
coordinates:
0.823,0.599,0.815,0.482,0.355,0.558,0.124,0.555,0.005,0.702,0.002,0.924,0.389,0.76,0.769,0.68,0.811,0.707
loitering_time: 0
objects:
- person
- package
motion:
mask: 0.295,0.069,0.703,0.066,0.704,0.023,0.293,0.021
record:
enabled: true
retain:
days: 3
mode: all
alerts:
retain:
days: 30
mode: active_objects
pre_capture: 5
post_capture: 15
detections:
retain:
days: 5
pre_capture: 5
post_capture: 15
review:
alerts:
labels:
- car
- cat
- person
- speech
required_zones:
- bell_front
detections:
required_zones:
- bell_front
frontdrive:
ffmpeg:
hwaccel_args: preset-nvidia-h264
inputs:
- path: rtsp://127.0.0.1:8554/frontdrive_sub
input_args: preset-rtsp-restream
roles:
- detect
- audio
- path: rtsp://127.0.0.1:8554/frontdrive
input_args: preset-rtsp-restream
roles:
- record
output_args:
record: preset-record-generic-audio-aac
detect:
width: 640
height: 360
fps: 15
objects:
track:
- person
- face
- license_plate
- car
- package
snapshots:
enabled: true
timestamp: true
bounding_box: true
crop: true
height: 360
retain:
default: 14
zones:
frontdrive_area:
coordinates: 1,1,0.737,0.382,0.095,0.652,0,1,0,1
objects:
- person
- face
- license_plate
- car
- package
inertia: 3
motion:
mask: 0.295,0.069,0.703,0.066,0.704,0.023,0.293,0.021
record:
enabled: true
retain:
days: 3
mode: all
alerts:
retain:
days: 30
mode: active_objects
pre_capture: 5
post_capture: 15
detections:
retain:
days: 5
pre_capture: 5
post_capture: 15
review:
alerts:
labels:
- car
- cat
- person
- speech
required_zones:
- frontdrive_area
detections:
required_zones:
- frontdrive_area
shed:
ffmpeg:
hwaccel_args: preset-nvidia-h264
inputs:
- path: rtsp://127.0.0.1:8554/shed_sub
input_args: preset-rtsp-restream
roles:
- detect
- audio
- path: rtsp://127.0.0.1:8554/shed
input_args: preset-rtsp-restream
roles:
- record
output_args:
record: preset-record-generic-audio-aac
detect:
width: 640
height: 360
fps: 15
objects:
track:
- person
- face
- car
- package
snapshots:
enabled: true
timestamp: true
bounding_box: true
crop: true
height: 360
retain:
default: 14
zones:
shed_area:
coordinates: 1,1,0.737,0.382,0.095,0.652,0,1,0,1
objects:
- person
- face
- car
- package
inertia: 3
motion:
mask: 0.295,0.069,0.703,0.066,0.704,0.023,0.293,0.021
record:
enabled: true
retain:
days: 3
mode: all
alerts:
retain:
days: 30
mode: active_objects
pre_capture: 5
post_capture: 15
detections:
retain:
days: 5
pre_capture: 5
post_capture: 15
review:
alerts:
labels:
- car
- cat
- person
- speech
required_zones:
- shed_area
detections:
required_zones:
- shed_area
steps:
ffmpeg:
hwaccel_args: preset-nvidia-h264
inputs:
- path: rtsp://127.0.0.1:8554/steps_sub
input_args: preset-rtsp-restream
roles:
- detect
- audio
- path: rtsp://127.0.0.1:8554/steps
input_args: preset-rtsp-restream
roles:
- record
output_args:
record: preset-record-generic-audio-aac
detect:
width: 640
height: 360
fps: 15
objects:
track:
- person
- package
snapshots:
enabled: true
timestamp: true
bounding_box: true
crop: true
height: 360
retain:
default: 14
zones:
steps_garden:
coordinates: 1,1,0.985,0.858,0.921,0.601,0.843,0.371,0.529,0.373,0.285,0.387,0,0.425,0,1,0,1
inertia: 3
loitering_time: 0
motion:
mask: 0.295,0.069,0.703,0.066,0.704,0.023,0.293,0.021
record:
enabled: true
retain:
days: 3
mode: all
alerts:
retain:
days: 30
mode: active_objects
pre_capture: 5
post_capture: 15
detections:
retain:
days: 5
pre_capture: 5
post_capture: 15
review:
alerts:
labels:
- car
- cat
- person
- speech
required_zones: steps_garden
detections:
required_zones: steps_garden
spare:
ffmpeg:
hwaccel_args: preset-nvidia-h264
inputs:
- path: rtsp://127.0.0.1:8554/spare_sub
input_args: preset-rtsp-restream
roles:
- detect
- audio
- path: rtsp://127.0.0.1:8554/spare
input_args: preset-rtsp-restream
roles:
- record
output_args:
record: preset-record-generic-audio-aac
detect:
width: 640
height: 360
fps: 15
objects:
track:
- person
- package
snapshots:
enabled: true
timestamp: true
bounding_box: true
crop: true
height: 360
retain:
default: 14
zones:
spare_path:
coordinates: 1,1,1,0.69,0.683,0.53,0.391,0.437,0.127,0.422,0.152,0.752,0.187,1
inertia: 3
loitering_time: 0
objects: person
back_garden:
coordinates:
0.125,0.412,0.124,0.005,0.797,0.008,0.995,0.182,0.997,0.682,0.623,0.503,0.413,0.435
loitering_time: 0
inertia: 3
motion:
mask: 0.295,0.069,0.703,0.066,0.704,0.023,0.293,0.021
record:
enabled: true
retain:
days: 3
mode: all
alerts:
retain:
days: 30
mode: active_objects
pre_capture: 5
post_capture: 15
detections:
retain:
days: 5
pre_capture: 5
post_capture: 15
review:
alerts:
labels:
- car
- cat
- person
- speech
required_zones: spare_path
detections:
required_zones: spare_path
side:
ffmpeg:
hwaccel_args: preset-nvidia-h264
inputs:
- path: rtsp://127.0.0.1:8554/side_sub
input_args: preset-rtsp-restream
roles:
- detect
- audio
- path: rtsp://127.0.0.1:8554/side
input_args: preset-rtsp-restream
roles:
- record
output_args:
record: preset-record-generic-audio-aac
detect:
width: 640
height: 360
fps: 15
objects:
track:
- person
- face
- license_plate
- dog
- cat
- car
- amazon
- fedex
- ups
- package
- bird
- bicycle
- motorcycle
snapshots:
enabled: true
timestamp: true
bounding_box: true
crop: true
height: 360
retain:
default: 14
zones:
garden:
coordinates: 1,0.589,0.677,0.535,0.697,0.384,0.297,0.336,0,0.384,0,1,0,1,1,1
inertia: 3
loitering_time: 0
objects:
- person
- face
- license_plate
- dog
- cat
- car
- amazon
- fedex
- ups
- package
- bird
- bicycle
- motorcycle
field:
coordinates: 0,0.382,0,0.123,0.108,0.121,1,0.114,0.993,0.382,0.703,0.382,0.292,0.336
loitering_time: 0
objects: person
motion:
mask: 0.295,0.069,0.703,0.066,0.704,0.023,0.293,0.021
record:
enabled: true
retain:
days: 3
mode: all
alerts:
retain:
days: 30
mode: active_objects
pre_capture: 5
post_capture: 15
detections:
retain:
days: 5
pre_capture: 5
post_capture: 15
review:
alerts:
labels:
- car
- cat
- person
- speech
required_zones: garden
detections:
required_zones: garden
detectors:
coral1:
type: edgetpu
device: usb
semantic_search:
enabled: True
model_size: large
genai:
enabled: True
provider: openai
api_key: "blah"
model: gpt-4o
birdseye:
enabled: true
width: 1280
height: 1080
quality: 1
mode: objects
version: 0.15-0 |
Beta Was this translation helpful? Give feedback.
-
|
I mean logs after Frigate has been running for some time |
Beta Was this translation helpful? Give feedback.
-
|
Here. |
Beta Was this translation helpful? Give feedback.
-
|
@NickM-27 what do you think? |
Beta Was this translation helpful? Give feedback.
-
This seems problematic but not causing major failure. Looks like there are some camera timeout issues and then this occurs which I have not seen before
|
Beta Was this translation helpful? Give feedback.
-
|
Do you want me to share here or elsewhere?
…On Wed, 4 Dec 2024 at 17:59, Nicolas Mowen ***@***.***> wrote:
we would really need to see a full copy of logs, but the error you pasted
indicates that some camera's streams are not stable and this would explain
missing recordings
—
Reply to this email directly, view it on GitHub
<#15097 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AHUULJBDQW3LHCNW5AISMUL2D47IJAVCNFSM6AAAAABSEM6FT2VHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTCNBWGQ2TAMI>
.
You are receiving this because you commented.Message ID:
***@***.***
com>
|
Beta Was this translation helpful? Give feedback.
-
|
Just to clarify genai description generation - does it use the lifetime of the detect stream (ie multiple images) or just a snapshot?
The reason I ask, is that using the setting below, the response is always based on a single image. Is there a config I am missing? I dont define use_snapshot: True at the camera level. |
Beta Was this translation helpful? Give feedback.
-
|
Well - that could do it 😄 An alternate question - which may be simple or may require more explanation, the system uses frigate.yml as the config, yet the editable config shown in settings, is frigate.yaml |
Beta Was this translation helpful? Give feedback.
-
|
Both filenames are valid. Where do you see the ".yaml" extension exactly? We edited the docs a while ago to make it consistent as ".yml", so if there's something we've missed, we'd like to fix it. |
Beta Was this translation helpful? Give feedback.
-
|
This was the information I was referring to:
It seems yaml is ignored and yml is used |
Beta Was this translation helpful? Give feedback.
-
|
Ah yes, this is specific to the Home Assistant addon. So you're saying that the statement in the docs is backwards? |
Beta Was this translation helpful? Give feedback.
-
|
Yes, for my setup - any changes made in the settings > configuration editor , they are reflected in the frigate.yaml file (when opened in Studio Code Server). However, these are not the actual setting used by the system (changes have no effect) When I manually edit frigate.yml in Studio Code Server and restart frigate, those settings in yml are used. I had a lot of confusing hours wondering what was going on :) |
Beta Was this translation helpful? Give feedback.
-
|
re: GenAI usefulness - can it be used to somehow further enhance identification of tracked objects? Example - right now I have a Christmas inflatable in view of my camera. The camera detects the little Santa on the inflatable as a person. GenAI has come back with "There is no person in the image sequence provided. The image shows a Christmas decoration on a porch." It would be great if "there is no person" could be fed back into Frigate to cancel / remove the false identification of a person. |
Beta Was this translation helpful? Give feedback.
-
|
it's also worth mentioning that Frigate doesn't need to implement this feature. If you really want to do this, you can just setup an MQTT listener to listen on the |
Beta Was this translation helpful? Give feedback.
-
|
Thanks guys. I don't have specific expectation of behavior or what happens if it is at night or whatever else - I just see some really accurate descriptions coming from GenAI on my images where object detection has got it wrong. @hawkeye217 asked for other ideas for GenAI (#14654). This seemed to be really useful from an end user perspective so I thought I'd put it out there. |
Beta Was this translation helpful? Give feedback.
-
|
right, my point was to dig deeper into the request and see how you'd expect it to work. For example, just deleting something is likely not a good idea because the risk involved is too great. Perhaps there could be a separate "GenAI challenged" flag or something, but not sure if that would really be helpful or not. |
Beta Was this translation helpful? Give feedback.
-
|
One suggestion on how GenAI may be helpful in this situation would be to change your prompt to include a line or phrase to say something like "if this object is not found in these images, add the phrase 'GenAI challenged' to the end of your description." Then, you can create a saved search for "GenAI challenged" (make sure you include the description as a source by flipping the switch in the Search Settings dropdown) and always have quick access to any of your tracked objects that GenAI has challenged. Then as you're there in Explore, you can quickly submit them to Frigate+ as false positives if necessary. |
Beta Was this translation helpful? Give feedback.
-
Great idea! I'll give that a shot. Thank you! |
Beta Was this translation helpful? Give feedback.
-
|
When the detect object is clipped to the bottom of the image, the label and score is not shown:
I'm running dev version 0.15.0-c0ba98e. I'm sure when this was introduced, but my feeling is that this was not the case with beta 1 (or even 2). |
Beta Was this translation helpful? Give feedback.
-
|
You're right. Strange thing that sometimes things can look different, even though you've looked at he pictures thousands of times.... |
Beta Was this translation helpful? Give feedback.
-
|
To be honest, I thought the same thing after implementing the fix, and I had to go back to old snapshots to make sure as well :) |
Beta Was this translation helpful? Give feedback.
-
|
Got these warnings after running latest dev based on your PR: {
"totalLines": 31,
"lines": [
"2024-12-06 16:56:05.829068012 [INFO] Preparing Frigate...",
"2024-12-06 16:56:05.886680664 [INFO] Starting Frigate...",
"2024-12-06 16:56:08.789772810 /usr/local/lib/python3.9/dist-packages/pydantic/_internal/_fields.py:161: UserWarning: Field \"model_hash\" has conflict with protected namespace \"model_\".",
"2024-12-06 22:42:15.416099 2024-12-06 16:56:08.789780259",
"2024-12-06 16:56:08.789783280 You may be able to resolve this warning by setting `model_config['protected_namespaces'] = ()`.",
"2024-12-06 16:56:08.789785434 warnings.warn(",
"2024-12-06 16:56:08.789880675 /usr/local/lib/python3.9/dist-packages/pydantic/_internal/_fields.py:161: UserWarning: Field \"model_type\" has conflict with protected namespace \"model_\".",
"2024-12-06 22:42:15.416134 2024-12-06 16:56:08.789883233",
"2024-12-06 16:56:08.789885447 You may be able to resolve this warning by setting `model_config['protected_namespaces'] = ()`.",
"2024-12-06 16:56:08.789886960 warnings.warn(",
"2024-12-06 16:56:12.009671823 [2024-12-06 16:56:12] frigate.util.config INFO : Checking if frigate config needs migration...",
"2024-12-06 16:56:12.241412756 [2024-12-06 16:56:12] frigate.util.config INFO : frigate config does not need migration...",
"2024-12-06 16:56:12.522989652 [2024-12-06 16:56:12] frigate.app INFO : Starting Frigate (0.15.0-8aa6297)",
"2024-12-06 16:56:12.543668376 [2024-12-06 16:56:12] peewee_migrate.logs INFO : Starting migrations",
"2024-12-06 16:56:12.544119246 [2024-12-06 16:56:12] peewee_migrate.logs INFO : There is nothing to migrate",
"2024-12-06 16:56:12.568436202 [2024-12-06 16:56:12] frigate.app INFO : Recording process started: 337",
"2024-12-06 16:56:12.570367983 [2024-12-06 16:56:12] frigate.app INFO : Review process started: 346",
"2024-12-06 16:56:12.575621877 [2024-12-06 16:56:12] frigate.app INFO : go2rtc process pid: 100",
"2024-12-06 16:56:12.600937577 [2024-12-06 16:56:12] detector.coral INFO : Starting detection process: 353",
"2024-12-06 16:56:12.606879383 [2024-12-06 16:56:12] frigate.detectors.plugins.edgetpu_tfl INFO : Attempting to load TPU as pci:0",
"2024-12-06 16:56:12.624413304 [2024-12-06 16:56:12] frigate.app INFO : Output process started: 380",
"2024-12-06 16:56:12.627658563 [2024-12-06 16:56:12] frigate.detectors.plugins.edgetpu_tfl INFO : TPU found",
"2024-12-06 16:56:13.375853003 [2024-12-06 16:56:13] frigate.app INFO : Camera processor started for voordeur: 417",
"2024-12-06 16:56:13.391610006 [2024-12-06 16:56:13] frigate.app INFO : Camera processor started for achtertuin: 420",
"2024-12-06 16:56:13.408060862 [2024-12-06 16:56:13] frigate.app INFO : Camera processor started for oprit: 426",
"2024-12-06 16:56:13.429651328 [2024-12-06 16:56:13] frigate.app INFO : Capture process started for voordeur: 439",
"2024-12-06 16:56:13.468287794 [2024-12-06 16:56:13] frigate.app INFO : Capture process started for achtertuin: 447",
"2024-12-06 16:56:13.488126977 [2024-12-06 16:56:13] frigate.app INFO : Capture process started for oprit: 451",
"2024-12-06 16:56:13.577413271 [2024-12-06 16:56:13] frigate.audio_manager INFO : Audio processor started (pid: 455)",
"2024-12-06 16:56:13.743710609 [2024-12-06 16:56:13] frigate.api.fastapi_app INFO : Starting FastAPI app",
"2024-12-06 16:56:14.582559298 [2024-12-06 16:56:14] frigate.api.fastapi_app INFO : FastAPI started"
]
}Something went wrong building the dev image? Those 2 in future time stamps don't make sense? |
Beta Was this translation helpful? Give feedback.
-
|
These messagse are due to #15327, which was also committed today. |
Beta Was this translation helpful? Give feedback.
-
|
Thanks. All good now, fixed with Nick's last PR. |
Beta Was this translation helpful? Give feedback.
-
|
Ugg, I know there is a feature request for this already but could we PLEASE get some form of semi real time AUDIO detection debugging like we do for motion and object? Like a graph of audio intensity vs the current threshold set? Like a streaming list of audio labels that fired with their % or something? Just missed an important delivery that required signing (which now needs pickup after end of day across town) because the delivery company doesn't know how to use a door bell and just knocks and then runs. |
Beta Was this translation helpful? Give feedback.
-
|
It's a planned feature that will be coming in a future version. |
Beta Was this translation helpful? Give feedback.
-
|
So I've been putting my new setup through all it's paces after completely resetting up my HA setup to now be HAOS exclusive. With that, I'm having none of the "weird" issues I previously had with shared network storage. HAOS seems to just work for the most part. Now that I'm digging into the GenAI stuff I'm trying to figure out how I actually can use it. From what I can tell all I can do with it for now is go and look at it and the process to do that is a 5-6 button clicks or am I missing something? Process I have been following: Is that really the process? Is there a better way for me to access these descriptions? |
Beta Was this translation helpful? Give feedback.
-
|
There is no reason to go through reviews, just use the explore page where you can search for objects (the genai descriptions are used for searching along with semantic search) You can also just apply any filters you want and then click on any object to see its description |
Beta Was this translation helpful? Give feedback.
-
|
Duh... I just realized I could just go to explore... ;) |
Beta Was this translation helpful? Give feedback.
-
Beta Was this translation helpful? Give feedback.
-
|
I've successfully set up Frigate with OpenVINO using TTech's script on Proxmox LXC (without Docker). Now, I want to run Frigate beta on Proxmox LXC with Debian 12, Docker, and Portainer. Frigate works with hardware acceleration, but OpenVINO isn't working because I can't find the /openvino_model or /opt/frigate folders. I copied the /openvino_model folder from the working setup, but still no luck. I haven't found a guide for getting OpenVINO to work with Proxmox + Docker + Portainer. Any hints or guidance would be greatly appreciated! |
Beta Was this translation helpful? Give feedback.
-
|
Proxmox is technically an unsupported installation method, and none of the Frigate maintainers use it. Perhaps someone else in the community can help? |
Beta Was this translation helpful? Give feedback.
-
|
I can find the guide to get Frigate worked on Proxmox+Portainer but with Coral. I don't have Coral so I would like to use OpenVINO. Hope someone with the same setup and help. Thanks. |
Beta Was this translation helpful? Give feedback.
-
|
I've been trying to get 0.15.0-beta2 working however on startup I get the following error: Commenting out the plus model reference in my configuration file allows it to start without issue. Looking at the latest code in main, it should be attempting to download the plus model if it's not there, so that to means the subdirectory of When I exec into the container with the models commented out I can see the |
Beta Was this translation helpful? Give feedback.
-
|
I run Frigate in a Kubernetes cluster. This is what changed between it working and not working: adampetrovic/home-ops@3d7c48e Literally just updating the Docker image tag and adding in the auth directive to the Frigate configuration. Everything else either runs in the ephemeral disk space of the container (deleted after every redeploy), in this case
yes.
yes.
no.
no.
|
Beta Was this translation helpful? Give feedback.
-
|
/config needs to be a mounted volume not just the config file. Go check the required volumes in the current docker install directions Frigate has required mounting more than just the config file for some time now. |
Beta Was this translation helpful? Give feedback.
-
0.14 has been working just fine without a mounted /config In any case, mounting my volume to /config seems to have worked, so I guess it's just something specific with how 0.15 is interacting with the file system. No big deal, problem solved :) |
Beta Was this translation helpful? Give feedback.
-
I didn't look through your whole conifg but I am surprised your DB even works. this changed in 0.13.0 https://github.com/blakeblackshear/frigate/releases/tag/v0.13.0 This was under breaking changes for docker users For Docker Container Users: |
Beta Was this translation helpful? Give feedback.
-
|
@bagobones that's because my db was explicitly mounted to |
Beta Was this translation helpful? Give feedback.
-
|
Is there a way to increase thumbnail size or resolution for passing to Gen AI? I’m experimenting with local LLM’s and it often says the image is blurry or too small. Alternatively can I pass something larger like a snapshot? |
Beta Was this translation helpful? Give feedback.
-
|
There is no quality difference between a snapshot and the sequence of images passed to GenAI during the object's lifecycle. They are both taken from the uncompressed detect stream. There is a quality difference between the thumbnail and snapshot if you regenerate the description after the object lifecycle has ended, however. The thumbnail that is saved is both compressed and cropped. This is why the regenerate button offers the ability to use a snapshot. |
Beta Was this translation helpful? Give feedback.
-
|
In general the recommendation would be to improve the sub stream quality or use the main stream scaled down by defining the detect resolution |
Beta Was this translation helpful? Give feedback.
-
|
Interesting, I hadn’t noticed the regenerate based on snapshot. I have some objects that appear stuck in the lifecycle and only show video options and no snapshot. The snapshot does work better. I guess it’s not really an issue then. I’ve just added it and been regenerating but missed the snapshot option. Thanks! |
Beta Was this translation helpful? Give feedback.
-
|
Whoa, the 0.15 UI is fantastic! The review page is awesome. The #1 thing I miss as I'm trying to dial things in is some form of debug replay to tune missed detections and alerts. Basically in Configuration>Settings in the Motion Tuner and Debug tabs I'd love to be able to feed a recording fragment instead of seeing the live video. An example: I get a DHL delivery maybe once a week and when the van arrived today it didn't get picked up correctly. I'd love to be able to replay that segment in debug to see what frigate did and figure out how to tune it. I know this is non-trivial 'cause you effectively have to construct a temporary new camera in the system to run this through... |
Beta Was this translation helpful? Give feedback.
-
|
Doesn't help for MISSED objects but object on the /explore page have an "Object Lifecycle" which will show you the progression of the object over time as its zone changed or detection changed during the event. |
Beta Was this translation helpful? Give feedback.
-
|
Yes, this is not a trivial task. There is a pinned feature request for it though, so it will come in a future version. In the meantime, you can do it manually by grabbing a clip of footage and running it through Frigate: https://docs.frigate.video/development/contributing#2-modify-your-local-config-file-for-testing |
Beta Was this translation helpful? Give feedback.
and others
-
Beta Documentation: https://deploy-preview-13787--frigate-docs.netlify.app/
Images
ghcr.io/blakeblackshear/frigate:0.15.0-beta2ghcr.io/blakeblackshear/frigate:0.15.0-beta2-standard-arm64ghcr.io/blakeblackshear/frigate:0.15.0-beta2-tensorrtghcr.io/blakeblackshear/frigate:0.15.0-beta2-tensorrt-jp4ghcr.io/blakeblackshear/frigate:0.15.0-beta2-tensorrt-jp5ghcr.io/blakeblackshear/frigate:0.15.0-beta2-rkghcr.io/blakeblackshear/frigate:0.15.0-beta2-rocmghcr.io/blakeblackshear/frigate:0.15.0-beta2-h8lChanges since Beta 1
Major Changes for 0.15.0
Breaking Changes
There are several breaking changes in this release, Frigate will attempt to update the configuration automatically. In some cases manual changes may be required. It is always recommended to back up your current config and database before upgrading:
frigate.dbfileshm_sizeis too low then a warning will be printed in the log stating that it needs to be increased.recordconfig has been refactored to allow for direct control of how longalertsanddetectionsare retained. These values will be automatically populated from your current config, but you may want to adjust the values that are set after updating. See the updated docs here and ensure your updated config retains the footage you want it to.hwaccelpreset they are using (preset-vaapimay need to now bepreset-intel-qsv-h264orpreset-intel-qsv-h265) if camera feeds are not functioning correctly after upgrading. This may need to be adjusted on a per-camera basis. If aqsvpreset is not working properly, you may still need to use apreset-vaapior revert to the previous ffmpeg version as described below.path: "5.0"in yourffmpeg:config entry. For example:ffmpegis no longer part of$PATH. In most cases this is handled automatically.execstreams will need to add the full path for ffmpeg which in most cases will be/usr/lib/ffmpeg/7.0/bin/ffmpeg.bin: ffmpegdefined, it needs to be removed.Explore
The new Explore pane in Frigate 0.15 makes it easy to explore every object tracked by Frigate. It offers a variety of filters and supports keyword and phrase-based text search, searching for similar images, and searching through descriptive text generated by AI models.
The default Explore pane shows a summary of your most recent tracked objects organized by label. Clicking the small arrow icon at the end of the list will bring you to an infinitely scrolling grid view. The grid view can also be set as the default by changing the view type from the Settings button in the top right corner of the pane.
The Explore pane also serves as the new way to submit images to Frigate+. Filters can be applied to only display tracked objects with snapshots that have not been submitted to Frigate+. The left/right arrow keys on the keyboard allow quick navigation between tracked object snapshots when looking at the Tracked Object Details pane from the grid view.
AI/ML Search
Frigate 0.15 introduces two powerful search features: Semantic Search and GenAI Search. Semantic Search can be enabled on its own, while GenAI Search works in addition to Semantic Search.
Semantic Search
Semantic Search uses a CLIP model to generate embeddings (numerical representations of images) for the thumbnails of your tracked objects, enabling searches based on text descriptions or visual similarity. This is all done locally.
For instance, if Frigate detects and tracks a car, you can use similarity search to see other instances where Frigate detected and tracked that same car. You can also quickly search your tracked objects using an "image caption" approach. Searching for "red car driving on a residential street" or "person in a blue shirt walking on the sidewalk at dawn" or even "a person wearing a black t-shirt with the word 'SPORT' on it" will produce some stunning results.
Semantic Search works by running an AI model locally on your system. Small or underpowered systems like a Raspberry Pi will not run Semantic Search reliably or at all. A dedicated GPU and 16GB of RAM is recommended for best performance.
See the Semantic Search docs for system requirements, setup instructions, and usage tips.
Generative AI
GenAI Search employs generative AI models to create descriptive text for the thumbnails of your tracked objects, which are stored in the Frigate database to enhance future searches. Supported providers include Google Gemini, Ollama, and OpenAI, so you can choose whether you want to send data to the cloud or use a locally hosted provider.
See the GenAI docs for setup instructions and use case suggestions.
Improved Tools for Debugging
Review Item Details Pane
A new Review Item Details pane can be viewed by clicking / tapping on the gray chip on a review item in the Review pane. This shows more information about the review item as well as thumbnails or snapshots for individual objects (if enabled). The pane also provides links to share the review item, download it, submit images to Frigate+, view object lifecycles, and more.
Object Lifecycle Pane
The Recordings Timeline from Frigate 0.13 has been improved upon and returns to 0.15 as the
Object Lifecycle, viewable in the Review Details pane as well as the new Explore page. The new pane shows the significant moments during the object's lifecycle: when it was first seen, when it entered a zone, became stationary, etc. It also provides information about the object's area and size ratio to assist in configuring Frigate to tune out false positives.Native Notifications
Frigate now supports notifications using the WebPush protocol. This allows Frigate to deliver notifications to devices that have registered to receive notifications in the Frigate settings, delivering them in a timely and secure manner. Currently, notifications will be delivered for all review items marked as alerts. More options for native notifications will be supported in the future.
See the notifications docs.
New Object Detectors
ONNX
ONNX is an open model standard which allows for a single model format that can run on different types of GPUs. The
default,tensorrt, androcmFrigate build variants include GPU support for efficient object detection via ONNX models, simplifying configuration and support more models. There are no default included ONNX models for object detection.AMD MiGraphX
Support has been added for AMD GPUs via ROCm and MiGraphX. Currently there is no default included model for this detector.
Hailo-8
Support has been added for the Hailo8 and Hailo-8L hardware for object detection on both arm64 and amd64 platforms.
Other UI Changes
Other Backend Changes
This discussion was created from the release 0.15.0 Beta 2.
Beta Was this translation helpful? Give feedback.
All reactions