A problem about of arcg yaml about lidar-bonnetal HOT 8 CLOSED

Hi, I calculate the statistics of x, y, z, range and depth. Actually, the means and STDs of y, z, range, depth are similar with yours. However, your mean of x is 10.88 while my result is around 0. I do not understand why the mean of x is 10.88 as the LIDAR is horizontally equipped.

from lidar-bonnetal.

Hi, I calculate the statistics of x, y, z, range and depth. Actually, the means and STDs of y, z, range, depth are similar with yours. However, your mean of x is 10.88 while my result is around 0. I do not understand why the mean of x is 10.88 as the LIDAR is horizontally equipped.

@LZDSJTU Hi，I also have this problem. This is the value I calculated:

img_means: #range,x,y,z,signal 
 -  11.71279 
 -  -0.1023471 
 -  0.4952 
 -  -1.0545 
 -  0.2877 
img_stds: #range,x,y,z,signal 
 -  10.24 
 -  12.295865  
 -  9.4287 
 -  0.8643 
 -  0.1450

I calculate the mean of all x, y, and z in the data set. I cannot get the data provided by the original arcg.yaml, how did you solve it?
@tano297

from lidar-bonnetal.

Hello,

Mean of each channel is the mean over the entire dataset for range, x, y,z and signal respectively. Std is the standard deviation in the same order.
You need to make a script to open your scans and accumulate the statistics over the whole data.
I am closing this because it is not an issue. Feel free to keep commenting.

from lidar-bonnetal.

@LZDSJTU How did those values work for you? I have values very similar to yours:
Means: x: 0.02926180063494279, y: -0.1316562397953751, z: -1.3030050504227546, r: 0.0, range: 12.73541935548366

STDs: x: 14.051696953243612, y: 13.017775592552525, z: 0.8700595079455102, r: 0.0, range: 14.394261975899619

(my r, or signal is 0 for each point)

Logically it kind of make sense. If you imagine a perfect 360º Lidar scan of a perfectly flat surface, the means for x, y would be 0 since for each x = 10, there would be x = -10 directly opposite.

Using this logic, if x and y means are close to 0, the Lidar scans would have been made in a fairly open environment. In Semantic-Kitti however, I imagine the presence of walls and other obstacles leads to the mean values being different.

from lidar-bonnetal.

hi, @LZDSJTU, @ivannson , how did you compute the the statistics of x, y, z, range and depth, do you compute the points in whole dataset? that's a large numbers, about one hundred million...

from lidar-bonnetal.

@xdtzzz was a long time ago, but I think I did it over the entire dataset. Despite the number of points it shouldn't be a difficult calculation and you're only doing it once anyway

from lidar-bonnetal.

from lidar-bonnetal.

from lidar-bonnetal.