Seeking Assistance with Eliminating False Positives in Motion Detection

General discussion about Blue Iris
User avatar
TimG
Posts: 2084
Joined: Tue Jun 18, 2019 10:45 am
Location: Nottinghamshire, UK.

Re: Seeking Assistance with Eliminating False Positives in Motion Detection

Post by TimG »

Another crucial point is can I still use zones with CPAI? This would be essential to avoid detecting people on the lower floor.
Yes. Your zones etc allow BI5 to decide what to send to CPAI.

Another thought. More cameras ! My Dahua cameras have "tripwire" where you can draw a line on the image for objects to cross. That can be picked up by BI5 as an Onvif trigger (No CPAI required). Alternatively, you could use a PIR with home automation to tell BI5 when to record.
Forum Moderator.
Problem ? Ask and we will try to assist, but please check the Help file.
User avatar
Pogo
Posts: 389
Joined: Tue Jul 18, 2023 7:21 pm
Location: Reportedly in the Area

Re: Seeking Assistance with Eliminating False Positives in Motion Detection

Post by Pogo »

Sardon wrote: Thu Nov 23, 2023 10:31 am
My objective is straightforward, all I want is to detect anyone approaching my front door while minimizing false positives caused by the sun's intermittent appearance.
The snapshot is very helpful in understanding the various considerations and possible options. I personally see it as one simple zone with possible background complications. Having triggers set up for both Motion and Object detection with CPAI thrown into the mix for what appears to be such a simple scene is rather extreme IMHO.

I gather by the reflections that the camera is actually indoors viewing the scene through a window? (I didn't quite grasp the hallway camera for testing purposes unless that was purely for CPAI evaluation and not for actually tuning the triggering.) If the camera is indeed indoors viewing the balcony through a window, I would obviously suggest mounting it outdoors with a direct view of approaching activity to instantly solve 90% of your problems..., unless the 'front door' is actually to a multi-unit complex and not your personal entrance. Either way, you are fighting two battles with the additional reflections, particularly with respect to the changes in lighting conditions -- day or night.

The balcony is the only area of interest with traffic entering the scene only from the left to reach the out of view entrance door all the way to the right? And you want triggers for traffic both entering and exiting the scene? While relatively easy to accomplish, my question (more out of curiosity than anything else) would be why both directions and not just for approaches?

And how much traffic are we concerned with observing/detecting?

With the scene now being known, a better understanding of the activity dynamics to be captured will provide additional help in offering a suggestion or two for simplifying the setup and hopefuly minimizing nuisance alerts.

This could also be a great use case for a clone camera for testing purposes if nothing else.
Last edited by Pogo on Wed Jan 17, 2024 11:08 am, edited 2 times in total.
User avatar
Pogo
Posts: 389
Joined: Tue Jul 18, 2023 7:21 pm
Location: Reportedly in the Area

Re: Seeking Assistance with Eliminating False Positives in Motion Detection

Post by Pogo »

As an additional observation, I notice in the zone example screenshot that the camera resolution appears to be rather high (based on the pixel size).

I mention this only because main stream triggers and sub stream triggers behave quite differently with each having their place. We can occasionally lose sight of this when exploring solutions to our nuisance alert issues.

One must keep in mind that a single pixel can be a very precise trigger. Add a few thousand more pixels and the ability to control trigger precision gets much more complicated!
Last edited by Pogo on Sat Nov 25, 2023 9:53 pm, edited 1 time in total.
User avatar
Pogo
Posts: 389
Joined: Tue Jul 18, 2023 7:21 pm
Location: Reportedly in the Area

Re: Seeking Assistance with Eliminating False Positives in Motion Detection

Post by Pogo »

So as to not cram all this stuff into one post, here is an example of the 'big pixels' as a fairly klunky bi-directional trip wire zone approach.

A simple example of a simple solution to achieving minimal nuisance triggers in what can sometimes be a very active scene..., using only 12 low resolution sub stream pixels of Zone A to trigger large object (human) activity approaching or exiting my back door.

Porch Trigger.jpg
Porch Trigger.jpg (146.14 KiB) Viewed 86083 times

The basic setup I generally use for starters.

Porch Sensor.jpg
Porch Sensor.jpg (91.86 KiB) Viewed 86083 times

Pre-trigger record times are a matter of choice, but with times less than 5 seconds no longer available from what I understand. I usually use 2 seconds since my BI version still supports the shorter times. I also use a clip oriented record method which essentially consolidates multiple short clips into 6 minute segments when continuous triggers are present, then compiles them into a maximum 15 minute file size if there's enough activity to do so. Makes viewing a days worth of clips very fast and easy for my particular setup of 16 cameras.

Porch Record.jpg
Porch Record.jpg (119.71 KiB) Viewed 86083 times

The result below: Notice the second counter reflecting the trigger reset for the opposite direction trigger. This can obviously be manipulated for various conditions/results.

Last edited by Pogo on Sat Nov 25, 2023 4:32 pm, edited 2 times in total.
User avatar
Pogo
Posts: 389
Joined: Tue Jul 18, 2023 7:21 pm
Location: Reportedly in the Area

Re: Seeking Assistance with Eliminating False Positives in Motion Detection

Post by Pogo »

My higher resolution driveway entrance trigger for MMS alerts. No nuisance triggers. None.

Drive Trigger.jpg
Drive Trigger.jpg (237.45 KiB) Viewed 86079 times

So clone the balcony camera. No 'clone master' stuff. Just a clean clone. (This leaves all your hard work and frustration in place to re-contemplate after creating a new triggering approach to your ongoing problem.)

Ensure the clone is basically a newly created camera with none of the triggering or recording parameters of the original. Follow the general examples previously provided to set up the highlighted area in the example below as a bi-directional trigger..., just the highlighted area at that particular ballister as it provides the best backround shielding for any unwanted activity behind it. Use a high res setting to keep the pixels small. What is currently in use would appear to be suitable (if not perfect) for the application. No additional zone area required beyond the ballister itself and on the walkway below.

Simpler Balcony Trigger.jpg
Simpler Balcony Trigger.jpg (74.78 KiB) Viewed 86079 times

De-activate the main camera to eliminate all the extra trigger processing baggage currently taxing your system and see how the clone performs. You may be pleasantly surprised.

Good luck.
User avatar
Pogo
Posts: 389
Joined: Tue Jul 18, 2023 7:21 pm
Location: Reportedly in the Area

Re: Seeking Assistance with Eliminating False Positives in Motion Detection

Post by Pogo »

Apologies for dragging this out further, but I omitted a screenshot of the basic motion settings that come into play when determining simple trigger reset and clip duration for my particular 6 minute clip purposes. Obviously feel free to modify these settings to suit your own reqirements.

Motion A.jpg
Motion A.jpg (100.3 KiB) Viewed 86072 times
Last edited by Pogo on Sat Nov 25, 2023 10:05 pm, edited 2 times in total.
User avatar
TimG
Posts: 2084
Joined: Tue Jun 18, 2019 10:45 am
Location: Nottinghamshire, UK.

Re: Seeking Assistance with Eliminating False Positives in Motion Detection

Post by TimG »

Oh the simplicity of it.

OP: Let us know if it works :D
Forum Moderator.
Problem ? Ask and we will try to assist, but please check the Help file.
User avatar
Sardon
Posts: 111
Joined: Sun Nov 19, 2023 10:50 am

Re: Seeking Assistance with Eliminating False Positives in Motion Detection

Post by Sardon »

I've just noticed the amount of follow ups on the thread.. Looks like I'll have to have a good read just to refresh. I'm going to re-install Blue Iris on a fresh install of Windows then I'm going to have another stab at the AI. Thanks for all the input.. if I run into any problems I'll be in touch.. Thanks again for all the valuable input
User avatar
Sardon
Posts: 111
Joined: Sun Nov 19, 2023 10:50 am

Re: Seeking Assistance with Eliminating False Positives in Motion Detection

Post by Sardon »

Pogo wrote: Fri Nov 24, 2023 7:42 pm My higher resolution driveway entrance trigger for MMS alerts. No nuisance triggers. None.

Drive Trigger.jpg

So clone the balcony camera. No 'clone master' stuff. Just a clean clone. (This leaves all your hard work and frustration in place to re-contemplate after creating a new triggering approach to your ongoing problem.)

Ensure the clone is basically a newly created camera with none of the triggering or recording parameters of the original. Follow the general examples previously provided to set up the highlighted area in the example below as a bi-directional trigger..., just the highlighted area at that particular ballister as it provides the best backround shielding for any unwanted activity behind it. Use a high res setting to keep the pixels small. What is currently in use would appear to be suitable (if not perfect) for the application. No additional zone area required beyond the ballister itself and on the walkway below.

Simpler Balcony Trigger.jpg
De-activate the main camera to eliminate all the extra trigger processing baggage currently taxing your system and see how the clone performs. You may be pleasantly surprised.

Good luck.
Hi Pogo.. So are you suggesting I delete the entire zone (after cloning), and simply create a new zone (same as your white line) instead of covering the complete railings ? I most certainly can try that out if this is the case. As for putting a camera outside is not really an option for me where I live, people are funny about this sort of thing as I reside in a block of flats. It's in my kitchen window the camera that overlooks the balcony and it's discrete and private and only serves a single purpose, to capture anyone walking to my front door of my flat from left to right.

The current setup captures everything 100% but it's the overcast weather which then just causes me so much stress as the phone keeps going of with motion detection alerts constantly. Maybe AI is overkill in this case. My objective is to simply capture anyone walking back and forth and it has to work like clockwork.

It's been a problem for me for very long time and if that simple white line (zone?) is the solution I'd be over totally over the moon!

Thanks for the insight and providing that short clip you created on YouTube..
User avatar
TimG
Posts: 2084
Joined: Tue Jun 18, 2019 10:45 am
Location: Nottinghamshire, UK.

Re: Seeking Assistance with Eliminating False Positives in Motion Detection

Post by TimG »

I believe Pogo is telling you to keep it really simple. Dump your existing rules for a clean fresh clone with none of the original rules, and try the single line in high resolution so it doesn't overlap the vertical bars.

You can run this in parallel to the existing camera, so you don't lose anything but can then see the different results.
Forum Moderator.
Problem ? Ask and we will try to assist, but please check the Help file.
Post Reply