Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[FR/Bug] - "Best image timeout" for clips (0.8.0) #346

Closed
tubalainen opened this issue Dec 9, 2020 · 7 comments
Closed

[FR/Bug] - "Best image timeout" for clips (0.8.0) #346

tubalainen opened this issue Dec 9, 2020 · 7 comments
Labels

Comments

@tubalainen
Copy link
Contributor

tubalainen commented Dec 9, 2020

I've got quite a lot of short video clips when there is more than one person moving within a camera - like when my kids are playing in the front yard.

Would it be possible to have a "re-trigger" threshold to just continue recording on "the first best detection" rather than creating many short videos? The event/scene is quite hard to follow with so many clips.
Something similar to "best image timeout" for snaps?

I am not 100% sure that I understand the global max_seconds: 300 setting fully, is it suppose do prevent "re-triggers"?

image

@blakeblackshear
Copy link
Owner

There are several things going on here. I see something similar on my cameras, but my clips aren't as short. What is the frame rate of your cameras?

@blakeblackshear
Copy link
Owner

max_seconds is a cache setting. It is the maximum clip length to prevent the cache from filling forever during long events.

@tubalainen
Copy link
Contributor Author

tubalainen commented Dec 9, 2020

There are several things going on here. I see something similar on my cameras, but my clips aren't as short. What is the frame rate of your cameras?

All cams are 8 fps.
I use a ram based spool area (tmpfs) and the clips folder are pointed towards an USB3.0 external disk.
I see no CPU bottleneck even if I am stressing the machine.

What would be the symptom if there are missing frames or other artifacts in the stream while the clip is being created? Might this be the cause of my short clips? BUT in t he example below, there was only ONE (broken?) clip created, not several.

AND

I have also noticed that some clips claiming to be "5 s" are much longer, but the clips are not complete/broken (in my case). (See screenshot below).
The (broken?) clip can be downloaded here for your reference, no errors could be found in the frigate logs during this event.
image

Sorry for the COVID-19 style "work from home" outfit "viewer discretion is advised" ;)
Clip download: http://tuben.lugn.nu:3000/0ad1648910ed

@blakeblackshear
Copy link
Owner

It's not a resource issue. Frigate determines that an object is gone after a certain number of frames go by without it being detected. With higher frame rates, this can cause objects to be marked as gone in less than a second. This would cause some events to be very short. Also, the duration shown in the media browser is the length of the event. It does not include the time for "pre capture" time.

@rjgrandy
Copy link

I also see this issue, where many short clips of the same "event" show up in the media browser, would like a way to have them combined if possible

@lawrence-jeff
Copy link

lawrence-jeff commented Jan 2, 2021

Also fighting this -I would like it to start a clip and continue as one clip until a user defined period of no detection occurs - Say 3 minutes, if it clipped the end of the video to correspond to the beginning of that period it would be ideal (so no 3 minutes of nothing). I also have kids go in and out of frame and it piles up the videos

@stale
Copy link

stale bot commented Feb 22, 2021

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

@stale stale bot added the stale label Feb 22, 2021
@stale stale bot closed this as completed Feb 25, 2021
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

No branches or pull requests

4 participants