Deepstack rejection person type on white listed faces?

sorka

Pulling my weight
Joined
Aug 9, 2015
Messages
366
Reaction score
197
I haven't yet installed Deepstack, so some of these questions might seem naive.

Since Deepstack itself can filter out white listed faces, does this work with BI integration? Is there a way to identify unique faces and white list them with an exclusion file? If an object is both detected as a person and a face, can the person detection be thrown out for alert purposes if that object is also a face that is known?

Is there any way to filter objects unless multiple cameras in a group all see the same object within x seconds? I have an area where 3 cameras intersect a large area and I want person and face detection in just the intersection of those views which are at different angles. I'd like to reduce the confidence level to really low, like 40% but only alert a detection if all 3 of those views have a person detected within say 10 seconds.
 

wittaj

IPCT Contributor
Joined
Apr 28, 2019
Messages
24,436
Reaction score
47,561
Location
USA
Most of us have found that the BI/DS integration as it relates to faces is hit or miss at best. Still a major work in progress.

Keep in mind that most of us have found in home settings that facial recognition is more gimmicky and novelty than anything else. If you have to put in 5 or 10 or 15 or 35 pictures of yourself in the NVR or DeepStack for it to recognize it is you...then you shouldn't expect much.... My success rate was under 5% so I moved on to other things LOL.

It can work in certain situations like a business that requires everyone to stop in front of the camera and the camera is at head height. Outside of that, the percentage of being accurate is probably not going to be super high. Especially someone coming to a front door.

Unless you spend the big bucks that casinos and airports have LOL.

Heck even in ideal situations like a business with the camera at ideal height and optimal lighting it fails....

 

kklee

Pulling my weight
Joined
May 9, 2020
Messages
187
Reaction score
201
Location
Vancouver, BC
@wittaj summed it up perfectly.

I just had a quick look at the faces that were captured, some are wildly inaccurate. The back of my head, a swirl of steam, and the weirdest one, a shot of the top half of a rolling recycling bin show that it's far from reliable, at least for home use, where the cameras are mounted higher and aimed downwards.

I don't rely on it for alerting, I have it enabled just out of curiosity (and for a chuckle).
 
Top