Motion detection is SOOOOOOOOOOOOO bad

[Hula_Rock]

Sometimes you just have to laugh.........

I pull up to my driveway and notice a box tucked behind a bush at the front door.  I have become so frustrated with these webcams that I was not surprised that I DID NOT receive an Alert.  Curious, I logged in and checked the time line for both cameras.  Mysteriously, according to Arlo, the white box just appeared out of thin air !!!!!!!!

Camera 1:

Camera 2:

You be the Judge,  would you expect at least one of the two cameras covering this zone to pick up motion?

Conveniently, and for some reason I am unable to attach any videos to this post.  hmmmmmmmmm.  But if you are curious about how the box got there, here is the YouTube link video captured by my door bell Motion Camera:

Door Bell Motion Camera Video

[JamesC]

Hula_Rock,

From the attached images it appears each camera captured a 20 second and a 31 second video. Did you manually capture these videos or did Arlo capture these while motion detection was enabled? 

Please take a look at the recently updated article for tips on optimizing motion detection: Optimizing motion detection performance on Arlo Wire-Free cameras

JamesC

[Hula_Rock]

---

JamesC wrote:

Hula_Rock,

From the attached images it appears each camera captured a 20 second and a 31 second video. Did you manually capture these videos or did Arlo capture these while motion detection was enabled? 

Please take a look at the recently updated article for tips on optimizing motion detection: Optimizing motion detection performance on Arlo Wire-Free cameras

JamesC

---

Good Morning James,

The videos where taken taken at a later time after the box was delivered.  As a matter of fact, a car driving on the street triggered the alert.  Arlo DID NOT catch the delivery.

[StuBee]

- What is your camera senstivity settings set to?   (wish everyone would post this with their motion detection complaints).

- I would think you can test the same scenario yourself, rather than waiting for the next drop off. Do you have the same issue on repeated tests?

- Have you considered repositioning your cameras so that the expected movement takes up a greater % of the viewable area?

Here's what mine looks like: 

https://arlo.netgear.com/#/viewShared/83F7ADEFDD4F6097

senstivity is set to 75%, so that the wind on that plant doesn't set it off.

[unco]

Out of curiosity, what doorbell camera are you using?

My Arlo that is in the front yard doesn't start recording when someone is walking to my front door until they are already out of sight under the porch.  A doorbell camera would be handy!

I'm too afraid to put my Arlo at the front door under the porch because it's low enough for someone to steal it even though I have a screw mount.

[TomMac]

My response is like Stubee's ...

take advantage of the sides and edges ot the PIR/Field of view where it is more sensitive... start up will be faster.

Also I then shrink the FOV by about 5% and re adjust the cam so the person is captured just as they enter my view.

[Hula_Rock]

---

StuBee wrote:

- What is your camera senstivity settings set to?   (wish everyone would post this with their motion detection complaints).

- I would think you can test the same scenario yourself, rather than waiting for the next drop off. Do you have the same issue on repeated tests?

- Have you considered repositioning your cameras so that the expected movement takes up a greater % of the viewable area?

Here's what mine looks like: 

https://arlo.netgear.com/#/viewShared/83F7ADEFDD4F6097

senstivity is set to 75%, so that the wind on that plant doesn't set it off.

---

@StuBee

@TomMac

What is your camera senstivity settings set to?    100 on both cameras

I would think you can test the same scenario yourself, rather than waiting for the next drop off. Do you have the same issue on repeated tests?

I have extensively tested dozens of locations to install the cameras that would afford me a FOV of the driveway and front door, even mounted the camera above my front door similar to your location.  Same Horrible results.

Have you considered repositioning your cameras so that the expected movement takes up a greater % of the viewable area?

In addition to my response to your second question above, I have installed the camera in incriments of 5 feet (left to right) above the garage door, 4 seperate locations giving a 100% view of my driveway and front yard.  Problem with those locations, a subject will be walking towards the camera (walking up the driveway) , center of the field of view where Arlo Motion detection is even Worse.

[TomMac]

Hey HulaRk...

I'm familiar your results/test for motion... Problematic to say the least.

I don't shoot the driveway much outside, but mostly the sidewalk.  Don't think any of my cams are set over 90% sense ( none less than 70% )

But I have ( for what ever reason  ) pretty good response from the cams loosing < 2 seconds of the front end ( or in steps about 3-4 by the delv man as he approaches ) I get a similar pic/video like above.

Don't have an answer to the problem as you know.

The only thing I can think of is the cheat method of shrinking the FOV ( pan/zoom smaller ) a bit as mentioned, last messg, so it allows the subject to enter the visable area with the delay off screen ( so to speak ).

[StuBee]

TomMac,

Can you explain your "cheat method of shrinking the FOV ( pan/zoom smaller )"

That doesn't really change when motion activation detection occurs does it?  You would just think it does because you are zoomed in.  So no real point in that I would think.  Let me know if I am misunderstanding.

These devices do not use the Camera lens to detection motion but the separate PIR sensor that is the little black ball you see on the device under the lens.  People should make sure that part of the camera is not being obscured by anything.

[Hula_Rock]

---

TomMac wrote:

Hey HulaRk...

I'm familiar your results/test for motion... Problematic to say the least.

I don't shoot the driveway much outside, but mostly the sidewalk.  Don't think any of my cams are set over 90% sense ( none less than 70% )

But I have ( for what ever reason  ) pretty good response from the cams loosing < 2 seconds of the front end ( or in steps about 3-4 by the delv man as he approaches ) I get a similar pic/video like above.

Don't have an answer to the problem as you know.

The only thing I can think of is the cheat method of shrinking the FOV ( pan/zoom smaller ) a bit as mentioned, last messg, so it allows the subject to enter the visable area with the delay off screen ( so to speak ).

---

I just shruck the FOV by 5 -10 %,  going to test it this weekend.  

Thanks for the suggestion.

[unco]

I tried shrinking my FOV in previous tests.

I'm pretty certain it just crops the image as StuBee suggests and has no effect on the PIR sensor.

Even when I completely cut off the front street with the cropping function, I would get motion detection alerts when a car drove past.

This is the type of video I would get every time a car drove past, keeping in mind cars drive past at a rather slow speed due to speed bumps.

The frustrating thing was it never got activated when I would come out or drive in to my own driveway.

https://arlo.netgear.com/hmsweb/users/library/share/link/754672AF432C3C6D

I tried all sensitivity settings, ranging from 25% to try and avoid activations from cars going past (still got activated) up to 100% to try and get me driving in and out of my driveway (only captured about once per week).

I then adjusted the aim to try and get more of my driveway movement and still don't get motion in my driveway or front door, yet I get cars activating it after they pass my mailbox.

https://arlo.netgear.com/hmsweb/users/library/share/link/6C51CAD3D38DCDBC

[TomMac]

"Can you explain your "cheat method of shrinking the FOV ( pan/zoom smaller )"

We know the PIR seems to be more sensitive on the sides and edge corners.... The Pan/crop feature can change the size of the  "seen" view but of course does not change how the PIR senses ( it senses the entire fov as its not pixel activated ).  Also due to this it is easier to capture a trigger event if the subject moves across the FOV instead of directly towards the cam.

So, define the area you want to capture, then using the pan/crop feature, decrease the size of what you see by 5-10%. At optimized video, you picture isn't degraded much... now re-align the cam to see the defined area you want best as possible.  The trigger event starts from the PIR sensing movement but since you don't see it ( happens before your visible area now ), it appears as if the event happens the exact time the subject/trigger enters the FOV.

Creating , yes... as your trigger event is off the actually viewed area.... it appears as if the triggered event is much quicker and you have a full capture of what you wanted ( in exchange for a small amount of quality )

hope this made sense

PS... this of course does't allow for cars in streets but more of a defined area closer to home where there is normally no movement.... make sure that regular movement pattens are excluded from full FOV ( or it will of course get false triggers )

[TomMac]

@Hula_Rock

PS;

My cams outside only cover about 1/4+ your area shown in the pic of your driveway... I think the PIR  ( except for large objects ) is only good to a max of 20ft and I try to keep it to a max of 15-20 ft out from cam.

It is actually more like your ring cam FOV

[Hula_Rock]

@TomMac

That is what I found out too.  The system does a great job picking up vehicle traffic moving at 25 mph on the street 30 feet out in the FOV, fast enough to see the car in a cpature video.  If someone walks up my driveway, well, we all know thats another story .