A Different Approach to Eliminating Collection of Unwanted Videos Caused by False Triggers

[looker]

I installed a new Arlo Pro system the day after Christmas 2016.  On the night of December 26th, after successfully standing the Arlo system up, after midnight wind speed picked up quite a bit.  Leaves flying by and tree limbs moving around triggered quite a false alerts by the 7AM disarm which I’d programmed via selecting “Schedule” mode in the Arlo app.

The next morning my wife grumbled that our new video system keep her awake half the night so I investigated how to fix the false triggering immediately after breakfast.  The solution I came up with was to create a new mode I called “Armed_Windy”.  That new mode was just a tweak of "Armed" via moving video and sound sensitivity down by an integer of 1 on all the cameras.  From creation of “Armed_Windy” I’ve activated that mode any time it’s been windy outside when we retire for the evening.

The restored bliss in the household held until last night.  About 3AM the wind kicked up again and the frequent false triggers where like having sleep apnea until I got up, turned on my iPad, accessed the Arlo app and MANUALLY moved mode from “Armed” to “Armed_Windy”.  Since I was wide awake I decided to do some more false trigger research.  I discovered multiple requests from the Community for the ability to draw polygons that limit a camera’s motion sensitive area.  Common sense, however, tells me that leaves blowing around could come from any direction and go anywhere in a camera’s field of view.  In my opnion limiting polygons will not work well for false triggers caused by swirling winds.

So … since about 4AM I brainstormed a better solution for wind-related false triggers.  I came up with two ways to accomplish this. 

The first involves my Netatmo Weather Station and their new Wind Gauge that I’m installing.  Soon I’ll have wind speed data that could be shared via IFTTT.  I had that thought because I use IFTTT currently to “Arm”/”Disarm” my Arlo Pro system via my cell phone’s GPS position.  That’s worked better for me than the beta geofencing functionality now in the Arlo app.  So that was idea number 1.  The obvious problem with that idea … not every Arlo user has a Netatmo for wind speed.  My next approach … how to use what’s already built into Arlo to detect windy conditions.

Idea 2 is using sound instead of wind speed to trigger a change from “Armed” to “Armed_Windy”.  I think that might be quire feasible for Arlo’s software engineers to implement.  Physical evidence that it can be achieved exists in almost all false triggered videos my system's recorded since the first night it was up and running. 

If I play back any of my USB stored false triggered recordings I can HEAR that it’s windy outside where the cameras are installed.  I’m no sound engineer but it seems to me that windy conditions outside sound quite different if compared to guys talking about breaking in my home or a cardinal singing on my birdfeeder announcing dawn each morning.  When I Googled “detecting wind noise in a video recording” I the search gave me back about 651,000 results (0.46 seconds).  The third link was to an article titled algorithm - Detecting wind noise - Stack Overflow.  I’m not suggesting this article contains the the solution … it just points to the potential of a solution beginning available as easily as clicking on a link from results of a Google search.

How important is this to me as an Arlo Pro owner?  After reviewing this morning all the locally stored videos since my system came online I’ve had one movement triggered recording of my neighbor’s cat and two collected because deer intruded into sensing space.  All 176 other false triggered videos were recored because of wind related movement and/or noise.  AUTOMATICALLY changing mode from “Armed” to “Armed_Windy” could have eliminated all of those.

[ironKingkong]

The audio detection works on my arlo pro.
However, I just think it is too sensitive even though I already decrease the sensitivity to 2.
Today I still receive two audio alerts, in one of which, I could hear wind. In another, I could hear bird.

I just want to let you know that you need to add a filter to ignore those sounds like wind and give filter options to customers.

I know it may be hard to filter a specific kind of sound, but definitely you could train it.

And you could give users options to skip a kind of sound like wind sound.

[JamesC]

The Arlo development team routinely reviews posts in the Arlo Idea Exchange to assess which features the community would like to see implemented. We greatly appreciate the communities contribution and will keep the status of this idea updated as we get new information on it's potential implementation. Thank you for posting your idea!