Training on my own dataset

[abdur4373]

Hello everyone, I am using the same toolbox, It is doing well on raw dataset I downloaded from NYU page however I want to process my own kinect data and make my own depth dataset for training of neural network. I wanted to know before giving it to project_depth_map.m function they are using swapbytes.m function of matlab.

First question is why would they use that function Secondly there is this condition of imgDepth(imgDepth < 600) = 2047; imgDepth(noiseMask) = 2047;

Why are they taking these values because my data is in mm and max value of like 4000 which makes it 4 meters.

I am really stuck in this issue could you help me please, I would appreciate that. Regards