I’m a student studying deep learning-based human gesture recognition.
It is being developed by Unity using the nuitrack sdk and realsense camera.
There’s something I’m curious about in my research, and I’m asking you here.
If you know how to save 3D coordinate values for all 16 joints(In progress as one of the nuitrak sdk tutorials) in a .skeleton file, please help me.
Hi jina,
Nuitrack doesn’t have embedded export of data. Nuitrack provides skeleton tracking data and you can export it to any suitable format yourself.
Oh, I see! Thank you for the good information.
Let me ask you one more question.
Can I extract data such as depth 2d coordinates, ir 2d coordinates, leanX, and Y etc… like kinect sdk? If possible, what should I do? It’s too difficult.;(
You can get the information about joints such as a joint type, real coordinates and projective coordinates, please see the relevant section in our docs (C++ and C#).
Thank you for your sincere reply every time.
It’s been so helpful!! 
Let me ask you one last question.
We are using the First Project( nuitrack SDK / tutorial) to extract the necessary joint data (in unity) , we would also like to extract the 2D joint coordinates from the Depth frame and RGB frame.
But in the case of the First Project, we don’t get the RGB or Depth frame.
Is there a way or function to obtain simultaneously the 3D coordinates of the joint and the 2D joint coordinates of each frame (RGB,Depth) in “First Project”?
Our First Project shows how to get 3D coordinates of joints. Displaying Skeletons on an RGB Image project shows how to get 2D coordinates of joints. Unfortunately, we don’t have a Unity project that shows how to get 2D and 3D coordinates at the same time but you can use the projects mentioned above for reference.
Your answer was too helpful! Thank you every time.
I want to use “Vector3 nuitrack. DepthSensor.ConvertRealToProjCoords” to get dept 2D , but I’m worried that “ConvertRealToProjCoords” didn’t appear.
I don’t know how to use the Depth Sensor module.
We are now proceeding on “Displaying Skeletons on an RGB Image,” as you advise.
If our previous method of obtaining Depth 2D is wrong, I think it would be helpful if you could let me know if you knew any other way.
Thank you every time:)