why it can't use pool when calling pcd.estimate_normals(search_param=o3d.geometry.KDTreeSearchParamHybrid(radius=1, max_nn=30))

[liang00fan]

Checklist

• I have searched for similar issues.
• For Python issues, I have tested with the latest development wheel.
• I have checked the release documentation and the latest documentation (for main branch).

Describe the issue

I try to use pcd.estimate_normals and i want to use pool to accelerate. but it will stuck.

Steps to reproduce the bug

import os
import threading
from functools import partial
from multiprocessing import Pool

import open3d as o3d
import numpy as np
import glob
import torch
# from scipy.spatial import distance
from sklearn.neighbors import NearestNeighbors
from tqdm import tqdm

normals_lock = threading.Lock()
def process(pth_path, output_dir):
    print(pth_path)

    cityname = os.path.basename(os.path.dirname(os.path.dirname(pth_path)))
    # print("cityname:", cityname)

    type_name = os.path.basename(os.path.dirname(pth_path))
    # print("Subfolder name:", type_name)

    file_name = os.path.basename(pth_path)

    output_pth_dir = os.path.join(output_dir, cityname, type_name)
    os.makedirs(output_pth_dir, exist_ok=True)
    data = torch.load(pth_path)
    try:
        xyz_origin, rgb, label = data
        pcd = o3d.geometry.PointCloud()
        pcd.points = o3d.utility.Vector3dVector(xyz_origin)
        pcd.estimate_normals(search_param=o3d.geometry.KDTreeSearchParamHybrid(radius=1, max_nn=30))
        ...
        torch.save(...)
    except:
        print(f'skip {pth_path}')


def process_entry(data, args):
    output_dir = args
    process(pth_path=data, output_dir=output_dir)


if __name__ == '__main__':
    
    pths=['*.pcd']  # pcd files
    # for p in pths:           # work success
    #     process(p, output_dir)

    with Pool(processes=1) as pool:
        pool.map(partial(process_entry, args=(output_dir)), pths)

Error message

no error message , it will stuck whiling using Pool.

Expected behavior

no

Open3D, Python and System information

- Operating system: Ubuntu 20.04
- Python version: Python 3.8 
- Open3D version: 0.18.0
- System architecture: x86
- Is this a remote workstation?: yes
- How did you install Open3D?: pip 
- Compiler version (if built from source): gcc 7.5

Additional information

no

[ssheorey]

Open3D uses multithreading. If you want to mix multiprocessing and multithreading in Python, you need to use forkserver or spawn methods for creating new processes, not the default 'fork'. See Python multiprocessing docs.

[liang00fan]

very thanks
it works now after add multiprocessing.set_start_method("spawn")