Need help with my setup for monitoring the outside

[John Gammel]

I’m running motioneyeos version 0.39.2 motion version 4.1.1 on a Raspberry Pi 3 with a 32GB microSD card.  The USB camera is an old, old Logitech 500 1.3 megapixel which the system sees as a V4L2 camera.  It’s installed in SW Florida so the environmental concerns are related to rain and heat (the Pi is in the garage with the USB camera underneath the eaves).

 

First question: I’ve recycled this one from another project so I have the MPEG-2 and VC-1 license keys from the Raspberry Pi store.  Will they help anything if I install them and if so how?

 

Much more importantly is getting the setup for outdoor monitoring so that it has very few false positives while capturing all of the real movement of people and cars.  If I can do this in my test setup, which has been a flop after two days, then I’ll purchase a good outside high resolution camera.

 

Here are the main setup items that I think count:

Video resolution: 1280x1024 (per the vendor but I think that’s after heavy interpolation)

Video rotation: 0 degrees (I had it mounted upside down but that seemed to cause real problems with the mask)

Frame rate: 10 (I’ve tried anywhere from 1 to 20 as well)

Still images off, movies on

Frame change threshold: 2% (I’ve tried from 1 to 10)

Auto Noise Detection: On

Light Switch Detection: 0%

Despeckle Filter: Off

Motion Gap: 10 seconds

Captured Before: 5 frames

Captured After: 5 frames

Minimum motion frames: 10

Mask: ON (more on this later)

Mask Type: Editable

Show Frame Changes: ON

 

I’ve tried lots of combinations of the above parameters over two days without finding a workable combination.

 

Mask information:

There is part of my driveway, the street, palm trees and their shadows

I have adjusted the mask to block out all of the tree shadow areas during the daylight hours

Most of the time the change value in the upper right hand corner of the display will show a value less than 100, and sometimes less than 10 for awhile

When there are recordings the change value spikes well above 10,000 even though the concrete area doesn’t seem to have any visible change.  As you can see from the driveway picture the change value is 5.

 

[David Chew]

I feel the pain, same issues here

--
You received this message because you are subscribed to the Google Groups "motioneye" group.
To unsubscribe from this group and stop receiving emails from it, send an email to motioneye+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/motioneye/491d411e-c3c7-41d9-a073-2fa894e9dab6%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

David T. Chew

610-996-7165 (C)

[Adrien]

Hi

Did you try playing with light switch detection and enable despeckle filter ?
Or maybe you can move down the point of view to not have tree shades.
Did you try dynamic mask ?
Maybe enable debug media file to understand your change.

Sorry no answer :/

[John Gammel]

I have tried a number of options but without obvious change except that I either get nothing or too much.  The front of our house and garage is all trees except for the driveway so that's not an option unless I want an extremely limited view which I don't.  Can't do much about the shade as the front orientation of the house is southeast so there is sun from sunup to late afternoon.

[Adrien]

Sorry,
Only thing I can recommend is to read motion documentation : https://motion-project.github.io/motion_guide.html

Maybe you will find more information.

Good luck 👍

[John Gammel]

I've spent a lot of time reading the documentation.  Tuning isn't really covered to the extent that I need it.

I made some headway.  Despeckle didn't help but changing to smart mask did.  With the old USB camera set to 320x240 (which I think is it's real optical resolution) I found the following for pixel change counts: me walking down the street (300), cars driving down the street (1500 peak), me walking closer to the camera 1000 (I can't move very fast).  Right now I'm running with 10 fps, change 1%, noise ON, gap 5 seconds (note that this isn't frames), before after and minimum of 5, 5, and 10 frames, and sluggishness of 8.  This seems to work decently.  The static frame right now is changing from about 50 to 250 but we have a rain storm coming.  Last night it was 5 - 75.

If I can get this stable with my new high res camera outdoor camera think I'll investigate using a Pi Zero W.

[John Gammel]

I've continued working away at this as time permits.  I did find a combination that actually works somewhat OK.

These are the settings with my old DCS-932L network security camera.

latest firmware, set 640 x 480 and 5 FPS

PI:

network camera at 5 frame rate

frame change: 1%

auto noise: on

despeckle: on (the biggest change I made)

10 seconds, 1 F, 1 F, 30 secs

smart mask

sluggishness: 1

On Friday, August 3, 2018 at 2:14:23 PM UTC-4, John Gammel wrote: