IVS person and vehicle false positives

105437

BIT Beta Team
Joined
Jun 8, 2015
Messages
2,059
Reaction score
954
At that tight of a field of view, drop the sensitivity way down. You could probably go to 3 or 4 and still pick up every human or vehicle.

And if that doesn't fix it, then global config probably will.
Okay, I'll drop the sensitivity. I haven't done the global config because I really haven't found a good explanation of how to do it. Thanks!
 

CanCuba

Known around here
Joined
Dec 9, 2020
Messages
1,132
Reaction score
3,418
Location
Havana, Cuba
Okay, I'll drop the sensitivity. I haven't done the global config because I really haven't found a good explanation of how to do it. Thanks!
I played with Global config for a few minutes before giving up. May be worthwhile to try again should we find, or be given, a decent tutorial.
 

wittaj

IPCT Contributor
Joined
Apr 28, 2019
Messages
25,175
Reaction score
49,062
Location
USA
He is a comparable field of view to yours - I get the late afternoon sun casting shadows for several hours across the entire field of view in a pattern similar to yours

I have never got a false trigger from the moving shadows.

My sensitivity is set to the default 5 position.

TREES.jpg
 

105437

BIT Beta Team
Joined
Jun 8, 2015
Messages
2,059
Reaction score
954
He is a comparable field of view to yours - I get the late afternoon sun casting shadows for several hours across the entire field of view in a pattern similar to yours

I have never got a false trigger from the moving shadows.

My sensitivity is set to the default 5 position.

View attachment 161905
I just set it to 4, we'll see how that works.
 

wittaj

IPCT Contributor
Joined
Apr 28, 2019
Messages
25,175
Reaction score
49,062
Location
USA
I only have two cameras that I had to do the global configuration for because they got the occasional false trigger.

Basically what you do is take an object of a known size (or use a person LOL) and you place the object in 3 different places in the field of view and you draw a vertical line the height of the object. You will want to zoom in to make sure you get as close to the actual bottom and top of the item as possible.

Then you also take that same object and lay it on the ground and draw a line for its length.

I usually do an object close, middle, and far within the area I am going to be using IVS.

Then you can test it by placing the object somewhere else in the field of view and draw a line the height of it and see if it spits out a height close to the actual height of the object. If it is close, call it good. If they are far off, then redo the calibration until it is close.
 

CanCuba

Known around here
Joined
Dec 9, 2020
Messages
1,132
Reaction score
3,418
Location
Havana, Cuba
I only have two cameras that I had to do the global configuration for because they got the occasional false trigger.

Basically what you do is take an object of a known size (or use a person LOL) and you place the object in 3 different places in the field of view and you draw a vertical line the height of the object. You will want to zoom in to make sure you get as close to the actual bottom and top of the item as possible.

Then you also take that same object and lay it on the ground and draw a line for its length.

I usually do an object close, middle, and far within the area I am going to be using IVS.

Then you can test it by placing the object somewhere else in the field of view and draw a line the height of it and see if it spits out a height close to the actual height of the object. If it is close, call it good. If they are far off, then redo the calibration until it is close.
I know what the wife and I are doing tomorrow! Lol
 

1fxman

Pulling my weight
Joined
Aug 29, 2018
Messages
273
Reaction score
143
Location
Oklahoma
I only have two cameras that I had to do the global configuration for because they got the occasional false trigger.

Basically what you do is take an object of a known size (or use a person LOL) and you place the object in 3 different places in the field of view and you draw a vertical line the height of the object. You will want to zoom in to make sure you get as close to the actual bottom and top of the item as possible.

Then you also take that same object and lay it on the ground and draw a line for its length.

I usually do an object close, middle, and far within the area I am going to be using IVS.

Then you can test it by placing the object somewhere else in the field of view and draw a line the height of it and see if it spits out a height close to the actual height of the object. If it is close, call it good. If they are far off, then redo the calibration until it is close.
What! No video. Just kidding. But I’m going to look into this.
 
Joined
May 1, 2019
Messages
2,215
Reaction score
3,504
Location
Reno, NV
I believe the intent of this Global Configuration is to help the Dahua AI determine depth from a 2D image. It's not 100% foolproof but it really did help knock down false positives (such as small animals walking by).
 

CanCuba

Known around here
Joined
Dec 9, 2020
Messages
1,132
Reaction score
3,418
Location
Havana, Cuba
I believe the intent of this Global Configuration is to help the Dahua AI determine depth from a 2D image. It's not 100% foolproof but it really did help knock down false positives (such as small animals walking by).
I haven't gotten a single false positive on any of my SMD 3 cameras despite heavy contrast scenes with lots of shadows.

The Color4K/X does miss the odd person on the tripwire at night so maybe this will help that.

Any false positives I get are during face detection but I also have the quality requirement turned down to the minimum and the sensitivity up to the max so that's so be expected.
 

105437

BIT Beta Team
Joined
Jun 8, 2015
Messages
2,059
Reaction score
954
I only have two cameras that I had to do the global configuration for because they got the occasional false trigger.

Basically what you do is take an object of a known size (or use a person LOL) and you place the object in 3 different places in the field of view and you draw a vertical line the height of the object. You will want to zoom in to make sure you get as close to the actual bottom and top of the item as possible.

Then you also take that same object and lay it on the ground and draw a line for its length.

I usually do an object close, middle, and far within the area I am going to be using IVS.

Then you can test it by placing the object somewhere else in the field of view and draw a line the height of it and see if it spits out a height close to the actual height of the object. If it is close, call it good. If they are far off, then redo the calibration until it is close.
@wittaj I'm taking a look at this today, I can't find a way to zoom in to be more precise with the measurements.
 
Top