I am interested in using multiple cameras to increase the accuracy of body tracking. There are some services out there that do this, and the results are fantastic! How can this be done with Nuitrack? How do you sync the cameras so they know they are staring at the same space?
If you want to run two Nuitrack instances in the same binary, you can use the fork function in Linux or the CreateProcess function in Windows for example.
I am interested in using multiple cameras to increase the accuracy of body tracking.
What sensors do you use? Why don’t you use one sensor, but for example with the best depth channel quality?
There are some services out there that do this, and the results are fantastic!
Could you share the names of these services with me?
How can this be done with Nuitrack? How do you sync the cameras so they know they are staring at the same space?
Unfortunately, Nuitrack does not have this feature at the moment. But now we are discussing the implementation of this.
Yes, gladly. IpiSoft is the name of the service. They mostly specialize in motion capture, but it’s machine learned skeletal tracking like your own software. And they have this crazy technology that allows them to sync multiple cameras and create tracking that is almost 100% accurate (no problems with limb occlusion when two cameras cover both sides of the person).
Presently we are using Kinect V2, which probably has the second best depth channel quality after Kinect Azure. Unfortunately, Nuitrack AI won’t run on our machine.
