0.15.0 Beta 5

[blakeblackshear]

Beta Documentation: https://deploy-preview-13787--frigate-docs.netlify.app/

Images

• ghcr.io/blakeblackshear/frigate:0.15.0-beta5
• ghcr.io/blakeblackshear/frigate:0.15.0-beta5-standard-arm64
• ghcr.io/blakeblackshear/frigate:0.15.0-beta5-tensorrt
• ghcr.io/blakeblackshear/frigate:0.15.0-beta5-tensorrt-jp4
• ghcr.io/blakeblackshear/frigate:0.15.0-beta5-tensorrt-jp5
• ghcr.io/blakeblackshear/frigate:0.15.0-beta5-rk
• ghcr.io/blakeblackshear/frigate:0.15.0-beta5-rocm
• ghcr.io/blakeblackshear/frigate:0.15.0-beta5-h8l

Changes since Beta 4

• Ability to save debug images for GenAI
• QSV ffmpeg changes for reolink compatibility
• Nested model config simplifications
• Fix for GPU stats when user has multiple Nvidia GPUs of same model
• Fix event cleanup error
• UI fixes and wording updates to simplify UI

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:

1. Simply copy your current config file to a new location
2. Stop Frigate and make a copy of the frigate.db file

• Rewritten SHM system
  • The SHM system has been rewritten to be more stable. This means that a large % of shared memory (SHM) will be used by default but the total usage will not fluctuate, and Frigate will not fail from SHM overflowing. However, this also means that having the correct SHM size for your cameras, detector, and hardware is more important to having a smooth experience with Frigate. A new calculation is done on startup, so if your shm_size is too low then a warning will be printed in the log stating that it needs to be increased.
• Recording configuration changed
  • The record config has been refactored to allow for direct control of how long alerts and detections are 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.
• ffmpeg updates
  • ffmpeg has been updated to version 7. Some Intel users may need to change what hwaccel preset they are using (preset-vaapi may need to now be preset-intel-qsv-h264 or preset-intel-qsv-h265) if camera feeds are not functioning correctly after upgrading. This may need to be adjusted on a per-camera basis. If a qsv preset is not working properly, you may still need to use a preset-vaapi or revert to the previous ffmpeg version as described below.
  • The new version provides many improvements, but the old version can still be run if needed by specifying path: "5.0" in your ffmpeg: config entry. For example:

    ffmpeg:
      path: "5.0"
    

  • ffmpeg presets have been updated. It is always recommended to use ffmpeg presets rather than manual arguments.
  • ffmpeg is no longer part of $PATH. In most cases this is handled automatically.
    • go2rtc exec streams will need to add the full path for ffmpeg which in most cases will be /usr/lib/ffmpeg/7.0/bin/ffmpeg.
    • If the go2rtc config has bin: ffmpeg defined, it needs to be removed.
• The nested model config under a detector has been simplified to just model_path as a string value. This change will be handled automatically through config migration.
• Submitting to Frigate+
  • The Frigate+ view in the UI has been replaced by the new Explore view. See the notes below.

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, and rocm Frigate 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

• Rework the Storage metrics pane to be more clear how much storage each camera is using
• Camera stream info button has been added to the camera metrics page allowing easy access to see camera codec information
• Add button for downloading the full set of logs
• The timeline handlebar will now automatically scroll while viewing recordings
• Review items and exports can now be shared via your browser's sharing pane or via direct link url. A secure context is required for the browser/OS sharing pane to function
• Add the ability to rename camera groups
• Add middle mouse button support in the Review pane to open full recordings view in a new tab
• Add Nvidia GPU encode and decode stats to system metrics
• New desktop keyboard shortcuts. In addition to the numerous keyboard shortcuts available with 0.14, 0.15 adds several more. In Live view, the numeric keys (1-9) will move PTZ cameras to presets. In Review, Page Up and Page Down will scroll through review items
• Accessibility updates for users with screen readers

Other Backend Changes

• FFMPEG updated to 7.0
• OpenVINO updated to 2024.3
• Use FastAPI instead of Flask for the Frigate API
• Add service manager infrastructure
• Install onevpl tools and update drivers for Intel
• If record is disabled for a camera then that camera will not save frames to the preview during that time
• Fix stalled connections to certain onvif cameras
• Fix autotracking absolute zoom interpolation logic
• TensorRT models are no longer generated by default, the ENV var must be set

---

This discussion was created from the release 0.15.0 Beta 5.

[pyrodex]

Thanks for all the hard work! I just upgraded and encountering a bug on the UI Tracked objects detail page. If you look at the bottom right the button is gone or something since it shows the event ID down there.

[hawkeye217]

It's actually a timestamp. I'll have a fix up for this shortly.

[hawkeye217]

Fixed in #15946

[pyrodex]

Tested and fixed. Thanks!

[hawkeye217]

Great! Thanks for checking it.

[bagobones]

Just updated.. Excited for this to go stable but also slightly more excited to start testing 0.16 features.

[hawkeye217]

Glad you're excited, but we're still a good while off from 0.16 :)

[thelovemyth]

Hope we have 0.15 stable soon <3

[hawkeye217]

Hope we have 0.15 stable soon <3

We do too! This is dependent on the community testing the betas and reporting any issues.

[ElectronicBattle]

A painless "docker compose" method for me. I didn't bother backing up my config nor my db - the config got translated to the new format without problem as far as I can tell and the db did lose some arbitrary entries (but that doesn't trouble me).

CPU burden might be a bit lower ... I will monitor.

I stopped and exited my 0.15 container and created a new docker compose with 0.15-1 which, as described, worked fine. But how/where did the new 0.15-1 get my full config from? I presume it's because 0.15 and 0.15-1 docker compose files state the same path for host:container.

[hawkeye217]

I presume it's because 0.15 and 0.15-1 docker compose files state the same path for host:container.

That's correct.

[ElectronicBattle]

I am getting a lot (many hundreds) of these errors, which I might have been getting in previous versions, but I don't recall seeing quite this message:

Timestamp 2025-01-12 13:01:14
Tag
ffmpeg.ponds.record
Message
[vost#0:0/copy @ 0x6400ce21ce40] Non-monotonic DTS; previous: 3081, current: 3081; changing to 3082. This may result in incorrect timestamps in the output file.

One of the cameras ("ponds") with which these messages are associated seems to be working OK, but another is not working OK and has always been a problem.

[ElectronicBattle]

also - trying to go back into the logs causes a webUI hang:

I tried twice; attempting to reload the page has also failed, resulting in the "aww, snap!" message on Chromium (Version 131.0.6778.204 (Official Build) for Linux Mint (64-bit)). Navigating back up from http://192.168.1.247:5000/logs to http://192.168.1.247:5000 seemed to do the trick - all OK now.

[NickM-27]

This is just an ffmpeg line, you'll want to look at the end for errors.

[ElectronicBattle]

This is just an ffmpeg line, you'll want to look at the end for errors.

OK - I'll look into it as a separate matter. Thanks.

[leccelecce]

All my streams have broken with:

Option bsf:v (A comma-separated list of bitstream filters) cannot be applied to input url rtsp://127.0.0.1:8554/hikcam8_resized -- you are trying to apply an input option to an output file or vice versa. Move this option before the file it belongs to.

Assume relates to @NickM-27 #15865

go2rtc config:

    hikcam8:
      - rtsp://hikcam8.local.lan:554/Streaming/Channels/101
    hikcam8_resized:
      - ffmpeg:hikcam8#video=h264#width=1920#height=1080#raw=-fpsmax 5#hardware=vaapi

Camera config:

    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/hikcam8
          input_args: preset-rtsp-restream
          roles:
            - record
        - path: rtsp://127.0.0.1:8554/hikcam8_resized
          input_args: preset-rtsp-restream
          roles:
            - detect
      output_args:
        record: preset-record-generic-audio-copy
    live:
      stream_name: hikcam8

[NickM-27]

Ah right, it is only applicable to ffmpeg 7.0 not 5.0

[NickM-27]

fixed in #15956

[bmrowe]

Thanks for a new beta! I use Reolink cameras and was able to switch back to ffmpeg 7.0 (I was getting tons of ffmpeg errors in beta 4 without specifying v. 5.0). I do have to specify qsv264, else I get errors with vaapi (intel 13500), but it has been super stable so far.

[NickM-27]

I'd suggest using the config the docs recommend

[bmrowe]

I had tons of issues with flv on 0.14, which led to using rtsp. I'll switch half the cameras back and see how things look.

[bagobones]

I had tons of issues with flv on 0.14, which led to using rtsp. I'll switch half the cameras back and see how things look.

What firmware is the camera running?

The most current firmware appears to be v3.1.0.3429_2404181313.

https://reolink.com/ca/download-center/

[bmrowe]

@bagobones , yeah 6 of the eight cameras are on latest. Two are intentionally on older versions because Reolink got rid of their motion activated spotlight settings and I need those in a couple locations. I do see the errors across all cameras (albeit, I'm not sure I've ever the issue on the doorbell).

[bmrowe]

Closing this out. I saw good results from shifting back to flv from rtsp w/ this newest beta. All cameras have been running on flv with zero of the errors above for over 24 hours. Thanks, Nick and team!

For anyone following along, or struggling with Reolink cameras, I'll summarize my setup. CX410 cameras plus a POE doorbell.

ffmpeg:
  hwaccel_args: preset-intel-qsv-h264
  output_args: record: preset-record-generic-audio-copy

All cameras configured as:

main:

ffmpeg:http://192.168.1.106/flv?port=1935&app=bcs&stream=channel0_main.bcs&user=user&password=password#video=copy#audio=copy

sub:

ffmpeg:http://192.168.1.106/flv?port=1935&app=bcs&stream=channel0_ext.bcs&user=user&password=password#video=copy#audio=copy

and then the camera specific portion:

driveway:
    ffmpeg:
      inputs:
        - path: rtsp://127.0.0.1:8554/driveway
          input_args: preset-rtsp-restream
          roles:
            - record
        - path: rtsp://127.0.0.1:8554/driveway_sub
          input_args: preset-rtsp-restream
          roles:
            - detect

[Polv89]

Can you tell me the right way to install it on proxmox?
I tried with debian 12 vm + docker but i have some issues
thanks

[NickM-27]

that is more or less the recommended way, I'd suggest creating a support discussion

[bagobones]

I don't know about RIGHT WAY, but here is what worked for me. With explanations.

VM Hardware

• Memory 8Gb or more, needs to have balloon DISABLED, when you do GPU pass through
• Machine type: q35 as it supports NEWER PCI pass through
• BIOS UEFI (to go with q35 hardware)
• Processor type: host (this will expose ALL CPU acceleration flags from the CPU, frigate uses many modern ones not in the templates)
• Controller VirtIO SCSI single
• Hard disks SCSI to go with the VirtIO SCSI controller (not to be confused with VirtIO disks which are no longer recommended) for lowest latency under current QEMU guidelines
• Network VirtIO (this should be the default anyway)

Pass through:

• iGPU pass through, PCI device, Raw device I located HD Graphics 630 on my host from the list (ROM-Bar should be checked by default)
• I have a USB Coral, I found USB device pass through made it VERY slow, so I used lsusb on the host to figure out which USB controller it was connected to and passed the WHOLE controller through as a RAW PCI device.

OS

I went with ubuntu-24.04.1 as it generally contains very modern drivers etc.. I chose the minimal install as docker and frigate will be the only things running other than adding the SSH server during install.. This results in only 4-5GB of disk use

Guest Agent

Don't forget to install the Qemu-guest-agent for shutdown and backup snapshot support https://pve.proxmox.com/wiki/Qemu-guest-agent also enable it on the VM settings in proxmox

Docker

I am not doing anything fancy with docker and ubuntu has a very current version in their repo so I just with with the ubuntu docker package docker.io
https://www.cherryservers.com/blog/install-docker-ubuntu

I then used docker compose to setup frigate, I made sure to first confirm the existence of the /dev/dri/renderD128 folder in my case in the ubuntu os and that the USB coral appeared under /dev/bus/usb

NOTES:

A PCI device can be only passed through to ONE VM at a time, you can't SHARE them unless you specifically have a very recent GPU that supports virtualization, there are more advanced proxmox guides for that.

Edit: Also a personal preference I made TWO virtual hard disks, one for the main OS and frigate install with DB and a second LARGER one for mapping /media/frigate to so that I could backup the main disk with snapshots easily without having a huge backup.. This is personal preference and requires extra steps of mounting and formatting the drive under linux.

Edit2: For further reference the host is using an Intel i7-7700T CPU @ 2.90GHz which I am using the integrated graphics for acceleration, exact device names for other iGPU and GPUs will differ.

[igiannakas]

Have been testing 0.15 for a while now and unfortunately FFmpeg v7 appears to be less resilient to camera stream / wifi latency hiccups/spikes compared to v5. For example see below. This is also repeated on pretty much all my cameras even though they all have good wifi coverage.

System is quite low utilisation wise (despite running 17 cameras). The detector spike is when rebooting due to the motion detection process generating a bunch of frames till it settles.

With v5 I seldomly get stream errors like the above.

Stats with V5 look very similar:

Using an intel pentium gold 8505 with 2 coral TPU's connected via USB3. Wifi delivered from a set of unifi access points with excellent reported coverage (cameras are connecting at their max speed). Cameras are Tapo C325WB's and C100's connected over Wifi and RTSP like so:

DownstairsLivingRoomSide:
  - rtsp://<<creds>>@192.168.1.61:554/stream1
  - ffmpeg:DownstairsLivingRoomSide#video=copy#audio=aac
DownstairsLivingRoomSide_sub:
  - rtsp://<<creds>>@192.168.1.61:554/stream2

[igiannakas]

Seems to have done the trick! thank you!

[igiannakas]

@NickM-27 another error now:

Any thoughts?

[NickM-27]

unrelated, potentially some issue just from restarting recently

[igiannakas]

Last one hopefully :)

how about these ones?

[NickM-27]

First one looks like the same thing, other one is just an opencv print I think, that should be cleared out via env variable unless it was overridden

[pyrodex]

I don't know if this is a bug or just testament to the hard work by the devs but I've noticed a pretty big diminishing usage in my Intel GPU sort of... When I say sort of if I reboot the server it seems to return to pre beta5 levels but it's hard to reproduce but right now I am sitting at sub-1% utilized (0.1X% at the moment) when I normally would run anywhere from 5-10% usage.

My configuration is 13 cameras 10 of them doing detection and recording as well.

I've included a graph for the last 30 days and you can see the change. Now when I upgraded I restarted the container but had not rebooted my box until around the 12th due to the fact I upgrade the kernel and didn't reboot but after restarting the container you can see the drop off again. Kind of a head scratcher and wanted to report the anomaly. Happy to provide logs or do any other testing.

[bagobones]

I normally don't worry about drops, most of the time I have found persistent HIGH CPU use I have found object detection going NUTS resizing the detection box on one object and not resetting till a reboot.

• go2RTC stream handling is very CPU bound (in general more active streams more CPU, not talking about transcoding)
• detect is very CPU bound (Resolution and frame rate are big factors)
• audio detection is CPU bound

[NickM-27]

you can check intel_gpu_top on the host to see what it says, but in general yeah decoding is a low load especially if you are now using qsv and you weren't before

[pyrodex]

you can check intel_gpu_top on the host to see what it says, but in general yeah decoding is a low load especially if you are now using qsv and you weren't before

Shows the same and I was using it before b5 just thought it was weird...

[brianfeucht]

Running on a Beelink N100 and just updated to Beta5 from 0.14. I have one camera that ffmpeg looks to be unhappy with. It had been running rock solid on 0.14.1

My config is fairly basic

detectors:
  coral:
    type: edgetpu
    device: usb

model:
  path: plus://**

go2rtc:
  streams:
    east_gate:
      - rtsp://**/cam/realmonitor?channel=1&subtype=0
    east_gate_sub:
      - rtsp://**/cam/realmonitor?channel=1&subtype=2

cameras:	  
  EastGate:
    ffmpeg:
      inputs:
        - path: rtsp://**/cam/realmonitor?channel=1&subtype=0
          roles:
            - record
        - path: rtsp://127.0.0.1:8554/east_gate_sub
          input_args: preset-rtsp-restream
          roles:
            - detect
            - audio
    audio:
      enabled: true
    motion:
      mask: 0.96,0.038,0.96,0.088,0.885,0.088,0.885,0.038
      threshold: 40
      contour_area: 20
    detect:
      width: 1280
      height: 720
      fps: 5

record:
  enabled: true
  retain:
    days: 7
    mode: all
  alerts:
    retain:
      days: 14
  detections:
    retain:
      days: 14
snapshots:
  enabled: true
  retain:
    default: 30

birdseye:
  enabled: false
 
version: 0.15-1

[hawkeye217]

Have a look at the breaking changes in the release notes, you likely need to change your ffmpeg hwaccel preset.

[brianfeucht]

Okay setting the hwaccel_args to preset-intel-qsv-h264 appears to have resolved it. All my cams are on h264 so maybe the auto detection is weird there.

[NickM-27]

the auto detection will only do preset-vaapi as there is not an easy way to know if qsv should be favored

[YasinMh]

Beta5 is playing nicer with my reolink cameras under qsv264, all 4 cameras displaying correctly. Beta4 had only 1 out of the 4 working, which meant having to revert to vaapi. Ffmpeg CPU usage for all of the cameras are now in single digits with qsv compared to ~20% for vaapi

[NathanMTL]

Thank you for the latest update Frigate Team ❤️

Can anyone give me more information on Model Nesting? Does that mean running multiple models?

[NickM-27]

What do you mean model nesting?

[NathanMTL]

Changes since Beta 4

Ability to save debug images for GenAI
QSV ffmpeg changes for reolink compatibility
Nested model config simplifications
Fix for GPU stats when user has multiple Nvidia GPUs of same model
Fix event cleanup error
UI fixes and wording updates to simplify UI

[NickM-27]

That is just referring to this config:

detectors:
  my_detector:
    type: ...
   model:   # <--- this field
     path: ...  # <--- this field

becoming simpler

[NathanMTL]

Thank you!

[thelovemyth]

How could I end Object Lifecycle, if Object became stationary.

[NickM-27]

You don't, if the object is still there then it will continue

[Thorsten1982]

I get error messages saying that recordings cannot be deleted. But in the system overview it shows me that 0 MB are used under recordings.

I saw this bug for the first time in this beta
frigate-logs.txt

[NickM-27]

sounds like an issue with the DB

[Thorsten1982]

okay, what can I do? does it make sense to add this entry first?

record:
sync_recordings: True

[NickM-27]

No, if you truly have 0MB of recordings used then the DB has already been reset and it won't delete a full block of recordings, you would want to delete those manually.

[Thorsten1982]

Okay, just so I understand correctly. So I just have to delete all the recordings. Otherwise I don't have to do anything with the database Nicolas Mowen ***@***.***> schrieb am Mi., 15. Jan. 2025, 16:15:

…

No, if you truly have 0MB of recordings used then the DB has already been reset and it won't delete a full block of recordings, you would want to delete those manually. — Reply to this email directly, view it on GitHub <#15945 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAIEABJGI5HCVKI7BWGANR32KZ3RNAVCNFSM6AAAAABVAJR7GOVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTCOBUGQZDSNQ> . You are receiving this because you commented.Message ID: ***@***.*** com>

[NickM-27]

right, based on your description of the issue, the DB is unaware of any recordings (as it shows a size of 0MB) meaning the DB was likely deleted or lost so it doesn't know about the existing recordings on the disc

[Thorsten1982]

Okay, so I have to delete the recordings using a Linux command, right? Or can I do it via the Frigate web interface? Am Mi., 15. Jan. 2025 um 16:27 Uhr schrieb Nicolas Mowen < ***@***.***>:

…

right, based on your description of the issue, the DB is unaware of any recordings (as it shows a size of 0MB) meaning the DB was likely deleted or lost so it doesn't know about the existing recordings on the disc — Reply to this email directly, view it on GitHub <#15945 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAIEABM6EURKRV4WYDS3SV32KZ443AVCNFSM6AAAAABVAJR7GOVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTCOBUGQ2DCMY> . You are receiving this because you commented.Message ID: ***@***.*** com>

[NickM-27]

It can't be done within Frigate. It could be done via CLI, file explorer, or anything else depending on your host environment

[Thorsten1982]

Thank you, it was somehow strange after all. I couldn't find any recordings in recordings but a lot of storage space was still used. I then set up the container completely. I hoped that the errors with my Reolink cameras would then go away. But unfortunately I still have the problem that it only works with

ffmpeg:
   Path: '5.0'

As soon as I remove this entry again I have massive problems in the log again

[Thorsten1982]

with this argument it works really well, no error message for a few hours

ffmpeg:
hwaccel_args: preset-intel-qsv-h264

[Polv89]

Is this configuration right ?
(i have enabled recording only for one for test it)

I want to:

1. watch cameras with better resolution
2. record when detect motion at low resolution

mqtt:
  enabled: false

ffmpeg:
  hwaccel_args: preset-vaapi

cameras:
  Retro: # <------ Nome della telecamera
    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://xx:[email protected]:554//h264Preview_01_sub # <----- Stream sub per la rilevazione
          roles:
            - detect
        - path: rtsp://xx:[email protected]:554//h264Preview_01_Main # <----- Stream principale per la visualizzazione e la registrazione
          roles:
            - record
    detect:
      enabled: true # <---- disable detection until you have a working camera feed
      width: 640
      height: 480

  Dietro_casa: # <------ Nome della telecamera
    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://xx:[email protected]:554//h264Preview_01_sub  # <----- Stream sub per la rilevazione
          roles:
            - detect
        - path: rtsp://xx:[email protected]:554//h264Preview_01_Main # <----- Stream principale per la visualizzazione e la registrazione
          roles:
            - record
    detect:
      enabled: false # <---- disable detection until you have a working camera feed
      width: 640
      height: 480

  Ingresso: # <------ Nome della telecamera
    enabled: true
    ffmpeg:
      inputs:
        - path: rtsp://admin:[email protected]/cam/realmonitor?channel=1&subtype=01&authbasic=xx # <----- Stream sub per la rilevazione
          roles:
            - detect
        - path: rtsp://admin:[email protected]/cam/realmonitor?channel=1&subtype=00&authbasic=xx# <----- Stream principale per la visualizzazione e la registrazione
          roles:
            - record
    detect:
      enabled: false # <---- disable detection until you have a working camera feed
      width: 640
      height: 480

record:
  enabled: True
  retain:
    days: 3
    mode: motion
  alerts:
    retain:
      days: 3
      mode: motion
  detections:
    retain:
      days: 3
      mode: motion



version: 0.15-1

Another thing, i set up date and time in the ubuntu server correctly, the same in docker, but i see that recordings are saved in the folder with 1 hour ago..can you help me?

[NickM-27]

watch cameras with better resolution

you need to setup go2rtc for that

Another thing, i set up date and time in the ubuntu server correctly, the same in docker, but i see that recordings are saved in the folder with 1 hour ago..can you help me?

recordings are always saved in UTC time

[Polv89]

How to setup go2rtc? what is it?

I know that recordings are always saved in UTC time, isn't possible to change it?
Could be convenient to have files with correct timestamp, because i check these files somewhere without opening frigate..

[NickM-27]

How to setup go2rtc? what is it?

https://docs.frigate.video/guides/configuring_go2rtc

I know that recordings are always saved in UTC time, isn't possible to change it?

No, there are many reasons why it is required, like daylight savings time, converting between timezones while traveling, etc.

Could be convenient to have files with correct timestamp, because i check these files somewhere without opening frigate..

You should not be accessing the files directly, Frigate is not designed to work that way. There are many APIs available to access recordings as needed.

[Polv89]

Ok, thanks a lot for informations, i will think about the way to access these recordings.

I will read the docs about go2rtc
Thanks

[MayorBug]

[NickM-27]

you should make a support discussion

[vcdx71]

I'm not seeing sub_labels in mqtt topic frigate/events. Should I be using a different topic?

payload_json:
    before:
      id: 1737049635.094058-c5j2ya
      camera: walkway
      frame_time: 1737049635.094058
      snapshot: null
      label: car
      sub_label: null
      top_score: 0
      false_positive: true
      start_time: 1737049635.094058
      end_time: null
      score: 0.7109375
      box:
        - 235
        - 94
        - 505
        - 209
      area: 31050
      ratio: 2.347826086956522
      region:
        - 180
        - 0
        - 556
        - 376
      active: true
      stationary: false
      motionless_count: 0
      position_changes: 0
      current_zones: []
      entered_zones: []
      has_clip: false
      has_snapshot: false
      attributes: {}
      current_attributes: []
      pending_loitering: false
      max_severity: alert
    after:
      id: 1737049635.094058-c5j2ya
      camera: walkway
      frame_time: 1737049661.296121
      snapshot:
        frame_time: 1737049661.296121
        box:
          - 269
          - 94
          - 506
          - 204
        area: 26070
        region:
          - 243
          - 0
          - 563
          - 320
        score: 0.859375
        attributes: []
      label: car
      sub_label: null
      top_score: 0.859375
      false_positive: false
      start_time: 1737049635.094058
      end_time: null
      score: 0.859375

[NickM-27]

you're only looking at one message here, and in this case you are looking at the new label which is very much unlikely to have a sub label assigned

[vcdx71]

Can you point me to what I should be looking at? I'm just trying to setup an automation that fires a notification when a delivery car (usps, ups, amazon, fedex) is detected in a zone. I'm trying to use the blueprint with a custom filter for this but it never fires and when I look at the traces for the time I see the car in frigate the sub label is always null.

[Red-ington]

if running the 0.15.0 beta5 & having AMD GPU PCI R9 390x should i be loading the ROCM?? ghcr.io/blakeblackshear/frigate:0.15.0-beta5-rocm

[NickM-27]

It's not 100% known as it's not that popular of a platform and none of the devs have recent AMD GPUs to test currently, but my older Vega GPU doesn't run it well. So 5700+ seems like a safer bet.

[Red-ington]

all good i was running 0.13.0 beta 2 before updating to this & have notice the GPU usage has changed heaps for my current config. I am looking to update to a 3050 or 3060 gpu & keep my CPU AMD 3800x going for a while longer. I'm just seeing a heap of CPU load compared to the older version. would this be due to the newer 7.0 FFMPEG update

[NickM-27]

I have a 3050 and it is way more than enough for just Frigate with 8 cameras. I don't suspect ffmpeg 7.0 will have changed the CPU usage but you can use the path config to change ffmpeg back to 5.0to test

[Red-ington]

so just to confirm AMD GPU taking some load off the CPU i should us hwaccewl_args: prerset-vaapi.

    - -hwaccel
    - vaapi
    - -hwaccel_device
    - /dev/dri/renderD128
    - -hwaccel_output_format
    - yuv420p```
    

[NickM-27]

Only preset-vaapi nothing else

[hkdce]

Thank you so much for adding ONNX support! I'm using it with TensorRT.
Since the official TensorRT Docker image includes TensorRT 8.x, could we downgrade onnxruntime_gpu to version 1.17.*?
onnxruntime_gpu 1.18.* is linked against TensorRT 10.x, which is causing Frigate to fail to start due to the missing shared object.

[NickM-27]

Frigate uses the cuda execution provider as the speed is barely affected but cuda execution provider uses significantly less host memory

[thelovemyth]

I think the thumbnail images have lower quality compared to the original videos. How can I improve the image quality so that Gemini AI can read more detailed information?

[thelovemyth]

Thank you for your feedback. I used Gemini AI to read the license plate number (if any) from the image and noted it in the description. I tried to find (in explore) the license plate number again but couldn't get the exact result for that video. What should I do to find the most accurate keyword?

[hawkeye217]

Are you using the "description" search type? You can specify it in Explore's Settings dropdown if not.

[hawkeye217]

On a related note, Frigate 0.16 will be able to do native LPR.

[thelovemyth]

Are you using the "description" search type? You can specify it in Explore's Settings dropdown if not.

of cource , yes
Thank you for your contribution

[thelovemyth]

Are you using the "description" search type? You can specify it in Explore's Settings dropdown if not.

Could I search this description data anywhere (link folder location) ?

[thelovemyth]

"Google Gemini has a free tier allowing 15 queries per minute to the API, which is more than sufficient for standard Frigate usage."
Each description is one request, or the number of requests depends on the number of images in the lifecycle ?
Thanks a lot for your help.

[thelovemyth]

It's probably just hallucinating. You can always enable debug logs for frigate.embeddings.maintainer to see how many images are being sent

Thank you for your reply. I'm new to this, so I'm still learning. Could you please show me where to enable this feature (debug logs for frigate.embeddings.maintainer)?

[NickM-27]

https://docs.frigate.video/configuration/advanced#frigate-logger

[thelovemyth]

I've noticed that when Frigate automatically request to Gemini, the response includes many images, but when I manually request new descriptions from the Regenerate button, it (Gemini) answer only 1 image.

[NickM-27]

Yes that's correct

[hawkeye217]

This is expected behavior. See the GenAI documentation. https://deploy-preview-13787--frigate-docs.netlify.app/configuration/genai/#custom-prompts

Optionally, you can generate the description using a snapshot (if enabled) by setting use_snapshot to True. By default, this is set to False, which sends the uncompressed images from the detect stream collected over the object's lifetime to the model. Once the object lifecycle ends, only a single compressed and cropped thumbnail is saved with the tracked object.