Hello. In our scenario, a user sitting in a chair in front of the sensor would be tracked, who, in some cases, would stand up and leave the scene (and return later). In such cases, obviously, the absence of the user could be detected as the tracking should stop.
However, when tested, we experienced that, quite frequently, the tracking would not stop as it would continue to track the chair as the user (see images attached).
Are there any known solutions for solving this problem?
Hello.
I think almost every body tracking SDK has this problem.
Actually, it also happens in kinect too if you try this action at the end of screen.
And it may not happen if you get enough far position.
Anyway, I think you must accept it sadly.
Sometimes, I use it to track more than two user.
FWIW - none of the current crop of skeletal tracking systems are optimized for SEATED tracking
And in the case of Nuitrack - they dont seem to use the color data as part of their user selection / discrimination processing
This to me is a fairly serious design flaw - as simple HSV based skin color detection would go a long way to rejection of this sort of issue - UNLESS of course the seat was skin colour - but well …
Westa
I think you could base your tracking on face detection, only track if a face is detected, and it would be easy as nuitrack only tracks the face if a skeleton is detected, hope this hack can work for you