0.13.0 Release

[blakeblackshear]

Docs: https://deploy-preview-6262--frigate-docs.netlify.app/

• Export recordings to export directory next to (or in) recordings folder via API and WebUI
• Prevent recordings from being turned on if disabled in config
• Plus integration
  • False positive submissions
  • Custom model downloads
• Community supported boards framework
• Timeline data
  • Events framework revamp
  • Generic events model
  • Add custom event support
• Add audio event support
• Move recordings management to separate process
• False positive reduction
  • Tracking improvements
    • Implement a smarter algorithm
    • Occlusion handling
  • Smarter region sizes based on historical detections
  • Better handling of objects moving into motion masked areas
  • Better handling of labels that are frequently confused (cat/dog)
  • Use historical object sizes to determine if an object is too large/small to be a true positive
  • Minimum number of frames for zone presence
• Improve motion detection
  • Make dynamic to handle night mode automatically
  • Automatically handle lightning
  • Detect and handle rain
  • PTZ moves
• Object attributes (face on person, license plate on car, for example)
• remove old legacy recordings cleanup method

Cleanup:

• Make attribute scores visible
• Retain video when there is detected audio
• Remove proper case in UI for hover text

Bugs:

• Save button not working in mask editor
• Fix "Unable to create jpg because frame 1688851244.054868 is not in the cache" errors
• Recordings management still dropping segments for some users. Unsure why.

Nice to haves:

• View exports in UI

[netlify]

✅ Deploy Preview for frigate-docs ready!

Name	Link
🔨 Latest commit	27567b4
🔍 Latest deploy log	https://app.netlify.com/sites/frigate-docs/deploys/65b8e8421f24c80008cd6330
😎 Deploy Preview	https://deploy-preview-6262--frigate-docs.netlify.app
📱 Preview on mobile	Toggle QR Code... Use your smartphone camera to open QR code link.

---

To edit notification comments on pull requests, go to your Netlify site configuration.

[kdill00]

This seems to be a significant improvement on retaining lock on objects when they are close together and also looks to have the added side affect of decreasing my detector fps usage due to not having to relook for the target as often with large crowds of people. Will be seeing how it does with massive crowds tomorrow. Looks markedly better though. Thanks for all your hard work

[Melkor82]

Is there a way to differentiate between tags or create custom ones? I have 4 cats (each one of a different color) I would like to have them recognized differently.

[NickM-27]

Is there a way to differentiate between tags or create custom ones? I have 4 cats (each one of a different color) I would like to have them recognized differently.

That would be an external feature, something like doubletake or birdcage

[bagobones]

I wonder if rain / snow / small bug swarms are all essentially the same problem at least in IR mode where they all look like bright shiny streaks instead of solid moving masses.

[DrSpaldo]

I am not sure if I have missed a setting in updating to 7c629c1, however, I don't appear to be getting stationary objects for cars anymore. I keep getting the same car being detected in the same spot over & over. The event lasts about 25 seconds

[NickM-27]

I am not sure if I have missed a setting in updating to 7c629c1, however, I don't appear to be getting stationary objects for cars anymore. I keep getting the same car being detected in the same spot over & over. The event lasts about 25 seconds

The only thing that changed is that stationary objects are now verified by default every 10 seconds and there is no way to disable that (though the value is still configurable)

[DrSpaldo]

I am not sure if I have missed a setting in updating to 7c629c1, however, I don't appear to be getting stationary objects for cars anymore. I keep getting the same car being detected in the same spot over & over. The event lasts about 25 seconds

The only thing that changed is that stationary objects are now verified by default every 10 seconds and there is no way to disable that (though the value is still configurable)

Interesting. After updating I just don't seem to be getting 'in progress' events. So I am not sure if I am missing something...

[NickM-27]

nothing would have changed with that, and it's working as expected for me

[DrSpaldo]

Are the audio events supposed to come up in Home Assistant as an event / sensor?

[NickM-27]

Are the audio events supposed to come up in Home Assistant as an event / sensor?

Yes, but you need the updated integration which isn't available in HACS yet

[DrSpaldo]

Are the audio events supposed to come up in Home Assistant as an event / sensor?

Yes, but you need the updated integration which isn't available in HACS yet

You're going to make me work for it! Alright.,...

I am guessing this is the PR for this issue in the integration; blakeblackshear/frigate-hass-integration#516

With 8f26352 being the build? blakeblackshear/frigate-hass-integration@8f26352

So, how do I make HA load that build instead?

[DrSpaldo]

With 8f26352 being the build? blakeblackshear/frigate-hass-integration@8f26352

So, how do I make HA load that build instead?

I downloaded the 8f26352 build and copied across the integration files into my folder and it is working now

[DrSpaldo]

Can any consideration be given to being able to configure the audio sensitivity based on the time? Ie. I want it less sensitive to speech during the day when people are around, but more sensitive at nighttime when criminals may be speaking with each other more quietly and speech is not expected at that time..

[NickM-27]

Can any consideration be given to being able to configure the audio sensitivity based on the time? Ie. I want it less sensitive to speech during the day when people are around, but more sensitive at nighttime when criminals may be speaking with each other more quietly and speech is not expected at that time..

what do you mean by audio sensitivity?

[DrSpaldo]

Can any consideration be given to being able to configure the audio sensitivity based on the time? Ie. I want it less sensitive to speech during the day when people are around, but more sensitive at nighttime when criminals may be speaking with each other more quietly and speech is not expected at that time..

what do you mean by audio sensitivity?

I guess the min volume setting and the thread hold setting?

[NickM-27]

I guess the min volume setting and the thread hold setting?

I am trying to think through this based on my usage of this over the last month.

I don't think changing the threshold makes sense, the score of "speech" for example doesn't change whether the speech is happening at day or night, and regardless you'll just be opening up to more false positives by lowering the threshold at night.

min volume is similar, in my experience with some tuning I've been able to get my config where it really only creates an event for speech that happens on our property, unless someone is yelling on the sidewalk which is something you'd want to record / detect anyway. A person speaking the same volume will be "louder" to the camera the closer that person is to the camera.

It very well might make sense in the future, but I think for the initial implementation of this feature it makes sense to hold off until we can make sure that the audio features without this variability is working as expected for more users.

[DrSpaldo]

I guess the min volume setting and the thread hold setting?

I am trying to think through this based on my usage of this over the last month.

I don't think changing the threshold makes sense, the score of "speech" for example doesn't change whether the speech is happening at day or night, and regardless you'll just be opening up to more false positives by lowering the threshold at night.

min volume is similar, in my experience with some tuning I've been able to get my config where it really only creates an event for speech that happens on our property, unless someone is yelling on the sidewalk which is something you'd want to record / detect anyway. A person speaking the same volume will be "louder" to the camera the closer that person is to the camera.

It very well might make sense in the future, but I think for the initial implementation of this feature it makes sense to hold off until we can make sure that the audio features without this variability is working as expected for more users.

Alright, just a thought… what did you end up using for your settings?

[NickM-27]

For my camera on the front I ended up with min_volume of 750, obviously it is going to depend on the camera. Also most high quality cameras include options in the camera settings for mic volume which would naturally affect that setting as well

[Majestic7979]

Out of curiosity how far are we from the "stable" 0.13 release?

[towerhand]

Great work. I have been running it for a couple of weeks, and it has been solid. From the new features, I really like the option to create and export custom events and time-lapses from the UI. Way to go!!!

[ljmerza]

which commit are you running? last time i tried this i had too many issues and had to go back. just tried the latest and all my previous events broke

[towerhand]

I'm currently running 3c25c57.