Hello, about the random kernel point selection

[longxiangxu321]

Hello Hugues,

I would really appreciate it if you are able to help me with the following two questions.

https://github.com/HuguesTHOMAS/KPConv-PyTorch/blob/73e444d486cd6cb56122c3dd410e51c734064cfe/kernels/kernel_points.py#L109

Is there a specific reason why you set a normal distribution of N(-radius0, 2*radius0) for the sampling of random kernel points?

https://github.com/HuguesTHOMAS/KPConv-PyTorch/blob/73e444d486cd6cb56122c3dd410e51c734064cfe/kernels/kernel_points.py#L133 Also, what is side_n? And why it is computed in this way?

Best regards, Leon

[HuguesTHOMAS]

Is there a specific reason why you set a normal distribution of N(-radius0, 2*radius0) for the sampling of random kernel points?

Well, it is not a normal distribution, it is uniform, between -radius0 and radius0. There is no specific reason, it is just initialization, not very important.

what is side_n? And why it is computed in this way?

For a volume, it would be the cubic root of the total number of points, so the number of points on a linear segment. This function. But we do not use this part of the code anyway. This is some debug code that I implemented some time, but do not use anymore