index -2147483648 is out of bounds for axis 0 with size 64

[liujw-tobias]

I use my own point cloud data，and convert PCD into bin file for testing and output labels, but I can't visualize it. What's the reason?

Sematickitti's data visualization succeeded

error：index -2147483648 is out of bounds for axis 0 with size 64

[willy8898]

I met the same problem, too. In my situation, the error comes from the range image projection.
Maybe you should comment the range image visualization in auxiliary/laserscanvis.py.

Then, recheck your data and labels.
If the point xyz and label are correct, you can successfully get the result.

In my situation, I found that I should use ".astype('float32')" before saving the downsampling 16-beam point cloud.
reference from here: https://stackoverflow.com/questions/10535687/write-a-raw-binary-file-with-numpy-array-data

[l9761116]

Hey! I'm also on a project of my own dataset. So I'm very interest in how you get your predictions. Which net did you use and did you trained on it? May I have some suggestions?

[L-Reichardt]

To adapt to other sensors:

in laserscanvis.py change lines 78, 79 to your sensor configuration.
in laserscan.py change LaserScan __init__(...) defaults to your sensor configuration.
in laserscan.py change LaserScan open_scan(...) make sure your file is loaded as float32.
in laserscan.py change SemLaserScan __init__(...) defaults to your sensor configuration.

Division by 0:
Change laserscan.py line 127 to prevent NAN.
pitch = np.arcsin(scan_z / depth) -> pitch = np.arcsin(scan_z / (depth + 1e-8))