3D object detection for custom objects; dataset creation

Question

How can I create a custom dataset for 3D object detection, I want to use the "Stanford3dDataset" or "Scannet" as baseline and add my object of interest in the dataset. I have the PCD files captured from the 3D camera [Realsense] and for 3D object detection, I am using the Pointnet model.

I see the dataset has the text file as input instead of PCD or PLY format, how do I convert PLY/PCD files to text files.

Answer 1

import open3d as o3d

import numpy as np

Load saved point cloud

pcd_load = o3d.io.read_point_cloud("try.ply")

convert PointCloud to numpy array

xyz_load = np.asarray(pcd_load.points)

Save points into a text file

np.savetxt('test.txt', xyz_load)`

More information is available in the link - NumPy <-> open3d.PointCloud