search

SysCV / LiDAR_snow_sim

LiDAR snowfall simulation
https://www.trace.ethz.ch/lidar_snow_sim
Other
181 stars 29 forks source link

The label of snow point cloud #23

Closed Grandzxw closed 1 year ago

Grandzxw commented 1 year ago

I have a question, in the function “process_single_channel”，it is understandable that pc[j, 4] = 1 or pc[j, 4] = 2 can be defined as a snowflake point, while pc[j, 4] = 0 is a normal point? right?

And then I find in wads datasets and cadcd datasets, The intensity of most snowflake points is 0(zero), So from your simulation, most snowflake point is not 0(zero). Can you explain my question?

Grandzxw commented 1 year ago

And the function is in the simulation.py. The link is https://github.com/SysCV/LiDAR_snow_sim/blob/760c42e46e10c6aa6b57048c5c1892289f0c2d08/tools/snowfall/simulation.py#L50

MartinHahner commented 1 year ago

I have a question, in the function “process_single_channel”，it is understandable that pc[j, 4] = 1 or pc[j, 4] = 2 can be defined as a snowflake point, while pc[j, 4] = 0 is a normal point? right?

pc[j, 4] = 0 # same as original point
pc[j, 4] = 1 # only alter intensity (meaning the location [x,y,z] stays the same
pc[j, 4] = 2 # replace point of hard target with snowflake, meaning not also the intensity, but also the location changes

We use these labels for this visualization in the paper: image

And then I find in wads datasets and cadcd datasets, The intensity of most snowflake points is 0(zero), So from your simulation, most snowflake point is not 0(zero). Can you explain my question?

0 is also a valid intensity, it is the lowest measurable intensity possible by the sensor, it is basically zero-indexed. Our code was designed to work with the SeeingThroughFog dataset, it is possible, that some parameters (e.g. here and there) need to be adjusted to work well with other datasets while the overall algorithm should still work.