About the quantization method of KITTI point cloud

[Thomas-pku]

Hello, I am very interested in your work. I used your open source weights and SparsePCGC to quantize to 1mm for testing. The experimental results are quite different from those in your article. Could you please tell me about your quantization method? If it is convenient, can you provide the processing script? Thank you!

[youkangGG]

Thank you for your interest in our paper! The KITTI dataset we utilized comprises densemap data, rather than the original KITTI point clouds in SparsePCGC. A simple conversion of the densemap provided in the aforementioned link is as follows:

import numpy as np
import pandas as pd
from pyntcloud import PyntCloud

input_bin_path = './40m_ILEN/08/000028.bin'
output_ply_path = './000028.ply'

points = np.fromfile(input_bin_path, dtype='float32').reshape(11, -1).T
points = points[:, :3]

df = pd.DataFrame(points, columns=['x', 'y', 'z'])
cloud = PyntCloud(df)
cloud.to_file(output_ply_path)