Dataset preparation about anchorudf

[zhaofang0627]

Actually, the keypoints/anchorpoints can be easily obtained by the following code:

import numpy as np
import trimesh
from sklearn.cluster import KMeans

sample_num = 100000 
point_num = 600

mesh = trimesh.load(input_mesh_path)
surface_points, _ = trimesh.sample.sample_surface(mesh, sample_num)
kmeans = KMeans(n_clusters=point_num, random_state=0).fit(surface_points)
key_points = kmeans.cluster_centers_

trimesh.Trimesh(vertices=key_points, faces=[]).export(out_file_off)
np.savez(out_file_path, kp=key_points)

[kampelmuehler]

Thanks you very much for sharing!