Capacity use of Multiple Sensors with NI mate? Can NI do multiple Skeleton tracking?


#1

Firstly curious to see if someone can explain the capacity of NI Mate with multiple sensors? What does it achieve? A wider range of motion capture? More accuracy? I understand with NI Mate you can use multiple sensors, but what can you do with multiple sensors?

Secondly and semi-related can NI mate track multiple skeletons and apply this two or more different rigs? I have not been able to determine either a yes or a no from documentation I have come across thus far.


#2

Additional sensors enable, for example, a wider field of view and thus more tracked subjects on single computer. However, it’s important to understand that the sensors are ultimately unaware of each other and thus the data that they provide (for example the skeleton tracking) cannot be fused.

In regards to your other question, the amount of tracked individuals depends on the sensor and library being used. For example, on a Windows PC using an XBOX One Kinect (gen2) device you can track up to six individual users.


#3

Hey Julius, thanks for the informative reply.
This makes it more clear.

Yes so winder field of view. My understanding from what you said is that Skeletons can not move between sensors fields seamlessly, makes sense.

I am planning to use Kinect V2, but on a mac.

I understand the hardware, V1 could only do 2 Skeletons and yes as you said V2 can do up to six per device. Is it therefore possible to do 12 with two Kinects V2 on a PC?

Also do you know what the limitations are with Mac and multiple Skeletons with your software, I know that V2 Kinect misses some bones cause it’s not using the windows SDK.


#4

Actually I think gen1 Kinect can track 6 people as well.

You could have two sensors to track 12 people total, assuming each of them faces a different direction. Note that with the V2 Kinect you’d probably need a separate USB3 controller for both Kinects as the USB bandwidth requirements are rather huge. There are USB3 extension cards for PCI-e that you can use.

The limitations for skeleton tracking are largely related to the sensor’s single facing nature: They can only detect a skeleton if they see the user from the front or back. Sideways shots tend to be very inaccurate.

The bones that don’t exist on Mac are the hands (palms, finger bones, wrist) and collars.