Send Motion-user mailing list submissions to
        motion-user@lists.sourceforge.net

To subscribe or unsubscribe via the World Wide Web, visit
        https://lists.sourceforge.net/lists/listinfo/motion-user
or, via email, send a message with subject or body 'help' to
        motion-user-requ...@lists.sourceforge.net

You can reach the person managing the list at
        motion-user-ow...@lists.sourceforge.net

When replying, please edit your Subject line so it is more specific
than "Re: Contents of Motion-user digest..."


Today's Topics:

   1. Re: some comments on motion detection (rmbusy+mot...@gmail.com)
   2. Re: some comments on motion detection (QQQ www)


----------------------------------------------------------------------

Message: 1
Date: Thu, 6 Jun 2019 11:24:10 -0700
From: rmbusy+mot...@gmail.com
To: motion-user@lists.sourceforge.net
Subject: Re: [Motion-user] some comments on motion detection
Message-ID: <d1a08fa1-0dae-8fd0-19ff-deb927ddb...@gmail.com>
Content-Type: text/plain; charset="utf-8"; Format="flowed"

Hi Graham,

 ? Like you, I'm also seeing far too many false positives for my 
liking.? I've been using Motion for a couple years as well, and I've 
tried playing around with various options to reduce them.

 ? One suggestion I wanted to add was the possibility of using blocks of 
pixels that change, instead of a raw pixel count.? The idea is that 
something moving across the field of view would be detected as it moved 
from block to block, while something that spans the whole field of view 
would be ignored.

 ? One false positive example is a sudden change in light, like a cloud 
passing by (causing shadows to change).? I see a lot of false positives 
where the edge of a shadow from a tree is cast across the whole picture, 
and a cloud makes it appear or disappear.? Or if the wind is constantly 
moving the shadow of the tree.

 ? Another major cause of false positives I see is in I.R. night vision 
mode, when a single strand of a spider web is draped in front of the 
camera.? The IR reflects off the web, making it very bright, and the 
slightest breeze causes it to move all night.? I can easily get 15,000 
pictures of a moving spider web strand, from one camera, in just one 
night (configured for 1 pic every 2 seconds)!

 ? The above suggestion should help to eliminate events that cover the 
full field of view of the camera (edge to edge), while still detecting a 
moving object through the field of view.


--
Rob.


On 6/5/19 10:51 PM, Graham Toal wrote:
> Hi - I've been using motion on raspberry pies for a few years now.
>
> I have some thoughts I'ld like to pass on.? Motion's detection has too 
> many false positives for my liking, and the offered solutions haven't 
> ever really fixed it for me.? So...
>
> 1) perhaps a tweak to the algorithm, so that changed pixels which also 
> change hue are weighted more strongly than changes in intensity, so 
> that increases or decreases in light level don't register as strongly 
> as an object moving over the display (which with colour cameras 
> usually causes a change in pixel colour as well as intensity - not so 
> true at night or in IR mode, but then the current algorithm would work 
> as a fallback)
>
> (I know the downside of this is that someone could walk over a green 
> lawn in green camo, but I'm not suggesting using hue changes *only* - 
> just giving them a heavier weighting)
>
> 2) after detecting motion, the video could then be run through an 
> external program supplied by the user which implements a user-written 
> algorithm that returns motion/no-motion and allows the system to 
> cancel that detection and do nothing further with the video.? (This 
> can be done to some extent with the existing script hooks but maybe 
> not as cleanly)
>
> 3) Could there be a mechanism to add some external trigger to 
> supplement the motion detection, eg a PIR sensor or radar sensor - and 
> have motion handle the video only if it detects motion in the video 
> *and* the external trigger also has detected motion?? (and some users 
> might want the option of "OR" rather than "AND" depending on their 
> tolerance of false positives)
>
> just passing thoughts to consider some day...
>
> regards
>
> Graham
>
>
> _______________________________________________
> Motion-user mailing list
> Motion-user@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/motion-user
> https://motion-project.github.io/
>
> Unsubscribe: https://lists.sourceforge.net/lists/options/motion-user

-------------- next part --------------
An HTML attachment was scrubbed...

------------------------------

Message: 2
Date: Thu, 06 Jun 2019 21:41:36 +0300
From: QQQ www <s...@cord.ws>
To: Motion discussion list <motion-user@lists.sourceforge.net>
Subject: Re: [Motion-user] some comments on motion detection
Message-ID: <564271559846...@sas2-7b909973f402.qloud-c.yandex.net>
Content-Type: text/plain; charset="us-ascii"

An HTML attachment was scrubbed...
-------------- next part --------------
A non-text attachment was scrubbed...
Name: index.jpeg
Type: image/jpeg
Size: 280157 bytes
Desc: not available

------------------------------



------------------------------

Subject: Digest Footer

_______________________________________________
Motion-user mailing list
Motion-user@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/motion-user


------------------------------

End of Motion-user Digest, Vol 156, Issue 13
********************************************

Reply via email to