-
Notifications
You must be signed in to change notification settings - Fork 0
/
dataloader.py
366 lines (317 loc) · 18.5 KB
/
dataloader.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
import torch
from plyfile import PlyData
import numpy as np
from torch.utils.data import DataLoader,Dataset,random_split
import os
import random
import pandas as pd
from scipy.spatial import distance_matrix
os.environ["CUDA_VISIBLE_DEVICES"] = '1'
# labels = ((255, 255, 255), (255, 0, 0), (255, 125, 0), (255, 255, 0), (0, 255, 0), (0, 255, 255),
# (0, 0, 255), (255, 0, 255), (0, 125, 255))
def face(xyz):
face_normal = []
for f_idx in range(xyz.shape[0]):
x = xyz[f_idx, :3]
y = xyz[f_idx, 3:6]
z = xyz[f_idx, 6:9]
xy = x - y
xz = x - z
face = [xy[1] * xz[2]-xz[1] * xy[2], xy[2] * xz[0]-xz[2] * xy[0], xy[0] * xz[1]-xz[0] * xy[1]]
face_normal.append(face)
return np.asarray(face_normal)
def Adj_matrix_gen(face):
N = face.shape[1]
face0 = face.repeat(1, N).view(N*N, 3)
face1 = face.repeat(N, 1)
b = (face0 == face1)
b = b[:, 0] + b[:, 1] + b[:, 2]
a = b.view(N, N)
adj = torch.where(a == True, 1., 0.)
return adj
def get_data(path=""):
labels = (
# [255, 0, 0],[255, 255, 0], [0, 255, 0], [0, 255, 255],
# [0, 0, 255], [255, 0, 255], [30, 144, 255], [0, 255, 127], [127, 255, 0],
# [255, 246, 143],[60, 179, 113], [255, 106, 106], [131, 111, 255], [255, 211, 155], [255, 99, 71],[155, 48, 255],
[255, 48, 48],[0, 191, 255], [255, 165, 0], [202, 255, 112],
[200, 255, 255], [255, 228, 255], [255, 155, 255], [255, 69, 0], [139, 0, 0],
[144, 238, 144],[0, 139, 139], [0, 0, 139], [139, 0, 139], [255, 105, 180], [230, 230, 250],[255, 228, 181],
[255, 255, 255]
)
row_data = PlyData.read(path) # read ply file
points = np.array(pd.DataFrame(row_data.elements[0].data))
faces = np.array(pd.DataFrame(row_data.elements[1].data))
# print(faces)
n_face = faces.shape[0] # number of faces
xyz = points[:, :3] # coordinate of vertex shape=[N, 3]
normal = points[:, 3:] # normal of vertex shape=[N, 3]
label_face = np.zeros([n_face, 1]).astype('int32')
label_face_onehot = np.zeros([n_face, 33]).astype(('int32'))
""" index of faces shape=[N, 3] """
index_face = np.concatenate((faces[:, 0]), axis=0).reshape(n_face, 3)
# print(index_face)
""" RGB of faces shape=[N, 3] """
RGB_face = faces[:, 1:4]
""" coordinate of 3 vertexes shape=[N, 9] """
xyz_face = np.concatenate((xyz[index_face[:, 0], :], xyz[index_face[:, 1], :],xyz[index_face[:, 2], :]), axis=1)
""" normal of 3 vertexes shape=[N, 9] """
normal_vertex = np.concatenate((normal[index_face[:, 0], :], normal[index_face[:, 1], :],normal[index_face[:, 2], :]), axis=1)
normal_face = face(xyz_face)
x1, y1, z1 = xyz_face[:, 0], xyz_face[:, 1], xyz_face[:, 2]
x2, y2, z2 = xyz_face[:, 3], xyz_face[:, 4], xyz_face[:, 5]
x3, y3, z3 = xyz_face[:, 6], xyz_face[:, 7], xyz_face[:, 8]
x_centre = (x1 + x2 + x3) / 3
y_centre = (y1 + y2 + y3) / 3
z_centre = (z1 + z2 + z3) / 3
centre_face = np.concatenate((x_centre.reshape(n_face,1),y_centre.reshape(n_face,1),z_centre.reshape(n_face,1)), axis=1)
""" get points of each face, concat all of above, shape=[N, 24]"""
points_face = np.concatenate((xyz_face, centre_face, normal_vertex, normal_face), axis=1).astype('float32')
""" get label of each face """
for i, label in enumerate(labels):
label_face[(RGB_face == label).all(axis=1)] = i
label_face_onehot[(RGB_face == label).all(axis=1), i] = 1
return index_face, points_face, label_face, label_face_onehot, points, index_face
# def get_data(path=""):
# labels = ([255,255,255], [255, 0, 0], [255, 125, 0], [255, 255, 0], [0, 255, 0], [0, 255, 255],
# [0, 0, 255], [255, 0, 255])
# row_data = PlyData.read(path) # read ply file
# points = np.array(pd.DataFrame(row_data.elements[0].data))
# faces = np.array(pd.DataFrame(row_data.elements[1].data))
# n_face = faces.shape[0] # number of faces
# xyz = points[:, :3] # coordinate of vertex shape=[N, 3]
# normal = points[:, 3:] # normal of vertex shape=[N, 3]
# label_face = np.zeros([n_face,1]).astype('int32')
# label_face_onehot = np.zeros([n_face,8]).astype(('int32'))
# """ index of faces shape=[N, 3] """
# index_face = np.concatenate((faces[:, 0]), axis=0).reshape(n_face, 3)
# """ RGB of faces shape=[N, 3] """
# RGB_face = faces[:, 1:4]
# """ coordinate of 3 vertexes shape=[N, 9] """
# xyz_face = np.concatenate((xyz[index_face[:, 0], :], xyz[index_face[:, 1], :],xyz[index_face[:, 2], :]), axis=1)
# """ normal of 3 vertexes shape=[N, 9] """
# normal_vertex = np.concatenate((normal[index_face[:, 0], :], normal[index_face[:, 1], :],normal[index_face[:, 2], :]), axis=1)
#
# normal_face = faces[:, 5:]
# x1, y1, z1 = xyz_face[:, 0], xyz_face[:, 1], xyz_face[:, 2]
# x2, y2, z2 = xyz_face[:, 3], xyz_face[:, 4], xyz_face[:, 5]
# x3, y3, z3 = xyz_face[:, 6], xyz_face[:, 7], xyz_face[:, 8]
# x_centre = (x1 + x2 + x3) / 3
# y_centre = (y1 + y2 + y3) / 3
# z_centre = (z1 + z2 + z3) / 3
# centre_face = np.concatenate((x_centre.reshape(n_face,1),y_centre.reshape(n_face,1),z_centre.reshape(n_face,1)), axis=1)
# """ get points of each face, concat all of above, shape=[N, 24]"""
# points_face = np.concatenate((xyz_face, centre_face, normal_vertex, normal_face), axis=1).astype('float32')
# """ get label of each face """
# for i, label in enumerate(labels):
# label_face[(RGB_face == label).all(axis=1)] = i
# label_face_onehot[(RGB_face == label).all(axis=1), i] = 1
# return index_face, points_face, label_face, label_face_onehot, points
def generate_plyfile(index_face, point_face, label_face, path= " "):
"""
Input:
index_face: index of points in a face [N, 3]
points_face: 3 points coordinate in a face + 1 center point coordinate [N, 12]
label_face: label of face [N, 1]
path: path to save new generated ply file
Return:
"""
unique_index = np.unique(index_face.flatten()) # get unique points index
flag = np.zeros([unique_index.max()+1, 2]).astype('uint64')
order = 0
with open(path, "a") as f:
f.write("ply\n")
f.write("format ascii 1.0\n")
f.write("comment VCGLIB generated\n")
f.write("element vertex " + str(unique_index.shape[0]) + "\n")
f.write("property float x\n")
f.write("property float y\n")
f.write("property float z\n")
f.write("property float nx\n")
f.write("property float ny\n")
f.write("property float nz\n")
f.write("element face " + str(index_face.shape[0]) + "\n")
f.write("property list uchar int vertex_indices\n")
f.write("property uchar red\n")
f.write("property uchar green\n")
f.write("property uchar blue\n")
f.write("property uchar alpha\n")
f.write("end_header\n")
for i, index in enumerate(index_face):
for j, data in enumerate(index):
if flag[data, 0] == 0: # if this point has not been wrote
xyz = point_face[i, 3*j:3*(j+1)] # Get coordinate
xyz_nor = point_face[i, 3*(j+3):3*(j+4)]
f.write(str(xyz[0]) + " " + str(xyz[1]) + " " + str(xyz[2]) + " " + str(xyz_nor[0]) + " "
+ str(xyz_nor[1]) + " " + str(xyz_nor[2]) + "\n")
flag[data, 0] = 1 # this point has been wrote
flag[data, 1] = order # give point a new index
order = order + 1 # index add 1 for next point
# labels_change_color = [[255, 255, 255], [255, 0, 0], [255, 125, 0], [255, 255, 0], [0, 255, 0], [0, 255, 255],
# [0, 0, 255], [255, 0, 255]]
labels_change_color = [
# [255, 0, 0],[255, 255, 0], [0, 255, 0], [0, 255, 255],
# [0, 0, 255], [255, 0, 255], [30, 144, 255], [0, 255, 127], [127, 255, 0],
# [255, 246, 143],[60, 179, 113], [255, 106, 106], [131, 111, 255], [255, 211, 155], [255, 99, 71],[155, 48, 255],
[255, 48, 48],[0, 191, 255], [255, 165, 0], [202, 255, 112],
[200, 255, 255], [255, 228, 255], [255, 155, 255], [255, 69, 0], [139, 0, 0],
[144, 238, 144],[0, 139, 139], [0, 0, 139], [139, 0, 139], [255, 105, 180], [230, 230, 250],[255, 228, 181],
[255, 255, 255]
]
for i, data in enumerate(index_face): # write new point index for every face
RGB = labels_change_color[label_face[i, 0]] # Get RGB value according to face label
f.write(str(3) + " " + str(int(flag[data[0], 1])) + " " + str(int(flag[data[1], 1])) + " "
+ str(int(flag[data[2], 1])) + " " + str(RGB[0]) + " " + str(RGB[1]) + " "
+ str(RGB[2]) + " " + str(255) + "\n")
f.close()
class plydataset(Dataset):
def __init__(self, path="data/train", mode='train', model='normal'):
self.mode = mode
self.model = model
self.root_path = path
self.file_list = os.listdir(path)
def __len__(self):
return len(self.file_list)
def __getitem__(self, item):
read_path = os.path.join(self.root_path, self.file_list[item])
index_face, points_face, label_face, label_face_onehot, points, RGB_face = get_data(path=read_path)
RGB_face = torch.from_numpy(RGB_face.astype(float))
raw_points_face = points_face.copy()
x_bias = random.uniform(-6, 6)
y_bias = random.uniform(-8, 8)
z_bias = random.uniform(-5, 5)
theta = random.uniform(-np.pi*0.15, np.pi*0.15)
X = np.array([[1, 0, 0], [0, np.cos(theta), -np.sin(theta)], [0, np.sin(theta), np.cos(theta)]])
Y = np.array([[np.cos(theta), 0, np.sin(theta)], [0, 1, 0], [-np.sin(theta), 0, np.cos(theta)]])
Z = np.array([[np.cos(theta), -np.sin(theta), 0], [np.sin(theta), np.cos(theta), 0], [0, 0, 1]])
if self.mode == 'train':
points_face[:, 0] = points_face[:, 0] + x_bias
points_face[:, 3] = points_face[:, 3] + x_bias
points_face[:, 6] = points_face[:, 6] + x_bias
points_face[:, 9] = points_face[:, 9] + x_bias
points_face[:, 12] = points_face[:, 12] + x_bias
points_face[:, 15] = points_face[:, 15] + x_bias
points_face[:, 18] = points_face[:, 18] + x_bias
points_face[:, 21] = points_face[:, 21] + x_bias
points_face[:, 1] = points_face[:, 1] + y_bias
points_face[:, 4] = points_face[:, 4] + y_bias
points_face[:, 7] = points_face[:, 7] + y_bias
points_face[:, 10] = points_face[:, 10] + y_bias
points_face[:, 13] = points_face[:, 13] + y_bias
points_face[:, 16] = points_face[:, 16] + y_bias
points_face[:, 19] = points_face[:, 19] + y_bias
points_face[:, 22] = points_face[:, 22] + y_bias
points_face[:, 2] = points_face[:, 2] + z_bias
points_face[:, 5] = points_face[:, 5] + z_bias
points_face[:, 8] = points_face[:, 8] + z_bias
points_face[:, 11] = points_face[:, 11] + z_bias
points_face[:, 14] = points_face[:, 14] + z_bias
points_face[:, 17] = points_face[:, 17] + z_bias
points_face[:, 20] = points_face[:, 20] + z_bias
points_face[:, 23] = points_face[:, 23] + z_bias
points_face[:, :3] = points_face[:, :3].dot(X.transpose())
points_face[:, 3:6] = points_face[:, 3:6].dot(X.transpose())
points_face[:, 6:9] = points_face[:, 6:9].dot(X.transpose())
points_face[:, 9:12] = points_face[:, 9:12].dot(X.transpose())
points_face[:, 12:15] = points_face[:, 12:15].dot(X.transpose())
points_face[:, 15:18] = points_face[:, 15:18].dot(X.transpose())
points_face[:, 18:21] = points_face[:, 18:21].dot(X.transpose())
points_face[:, 21:24] = points_face[:, 21:24].dot(X.transpose())
points_face[:, :3] = points_face[:, :3].dot(Y.transpose())
points_face[:, 3:6] = points_face[:, 3:6].dot(Y.transpose())
points_face[:, 6:9] = points_face[:, 6:9].dot(Y.transpose())
points_face[:, 9:12] = points_face[:, 9:12].dot(Y.transpose())
points_face[:, 12:15] = points_face[:, 12:15].dot(Y.transpose())
points_face[:, 15:18] = points_face[:, 15:18].dot(Y.transpose())
points_face[:, 18:21] = points_face[:, 18:21].dot(Y.transpose())
points_face[:, 21:24] = points_face[:, 21:24].dot(Y.transpose())
points_face[:, :3] = points_face[:, :3].dot(Z.transpose())
points_face[:, 3:6] = points_face[:, 3:6].dot(Z.transpose())
points_face[:, 6:9] = points_face[:, 6:9].dot(Z.transpose())
points_face[:, 9:12] = points_face[:, 9:12].dot(Z.transpose())
points_face[:, 12:15] = points_face[:, 12:15].dot(Z.transpose())
points_face[:, 15:18] = points_face[:, 15:18].dot(Z.transpose())
points_face[:, 18:21] = points_face[:, 18:21].dot(Z.transpose())
points_face[:, 21:24] = points_face[:, 21:24].dot(Z.transpose())
# p = random.uniform(0, 3)
# if p >= 0 and p < 1:
# points_face[:, :3] = points_face[:, :3].dot(X.transpose())
# points_face[:, 3:6] = points_face[:, 3:6].dot(X.transpose())
# points_face[:, 6:9] = points_face[:, 6:9].dot(X.transpose())
# points_face[:, 9:12] = points_face[:, 9:12].dot(X.transpose())
# points_face[:, 12:15] = points_face[:, 12:15].dot(X.transpose())
# points_face[:, 15:18] = points_face[:, 15:18].dot(X.transpose())
# points_face[:, 18:21] = points_face[:, 18:21].dot(X.transpose())
# points_face[:, 21:24] = points_face[:, 21:24].dot(X.transpose())
# elif p >= 1 and p < 2:
# points_face[:, :3] = points_face[:, :3].dot(Y.transpose())
# points_face[:, 3:6] = points_face[:, 3:6].dot(Y.transpose())
# points_face[:, 6:9] = points_face[:, 6:9].dot(Y.transpose())
# points_face[:, 9:12] = points_face[:, 9:12].dot(Y.transpose())
# points_face[:, 12:15] = points_face[:, 12:15].dot(Y.transpose())
# points_face[:, 15:18] = points_face[:, 15:18].dot(Y.transpose())
# points_face[:, 18:21] = points_face[:, 18:21].dot(Y.transpose())
# points_face[:, 21:24] = points_face[:, 21:24].dot(Y.transpose())
# else:
# points_face[:, :3] = points_face[:, :3].dot(Z.transpose())
# points_face[:, 3:6] = points_face[:, 3:6].dot(Z.transpose())
# points_face[:, 6:9] = points_face[:, 6:9].dot(Z.transpose())
# points_face[:, 9:12] = points_face[:, 9:12].dot(Z.transpose())
# points_face[:, 12:15] = points_face[:, 12:15].dot(Z.transpose())
# points_face[:, 15:18] = points_face[:, 15:18].dot(Z.transpose())
# points_face[:, 18:21] = points_face[:, 18:21].dot(Z.transpose())
# points_face[:, 21:24] = points_face[:, 21:24].dot(Z.transpose())
# move all mesh to origin
centre = points_face[:, 9:12].mean(axis=0)
points_face[:, 0:3] -= centre
points_face[:, 3:6] -= centre
points_face[:, 6:9] -= centre
points_face[:, 9:12] = (points_face[:, 0:3] + points_face[:, 3:6] + points_face[:, 6:9]) / 3
points[:, :3] -= centre
max = points.max()
points_face[:, :12] = points_face[:, :12] / max
# normalized data
maxs = points[:, :3].max(axis=0)
mins = points[:, :3].min(axis=0)
means = points[:, :3].mean(axis=0)
stds = points[:, :3].std(axis=0)
nmeans = points[:, 3:].mean(axis=0)
nstds = points[:, 3:].std(axis=0)
nmeans_f = points_face[:, 21:].mean(axis=0)
nstds_f = points_face[:, 21:].std(axis=0)
for i in range(3):
#normalize coordinate
points_face[:, i] = (points_face[:, i] - means[i]) / stds[i] # point 1
points_face[:, i + 3] = (points_face[:, i + 3] - means[i]) / stds[i] # point 2
points_face[:, i + 6] = (points_face[:, i + 6] - means[i]) / stds[i] # point 3
points_face[:, i + 9] = (points_face[:, i + 9] - mins[i]) / (maxs[i] - mins[i]) # centre
#normalize normal vector
points_face[:, i + 12] = (points_face[:, i + 12] - nmeans[i]) / nstds[i] # normal1
points_face[:, i + 15] = (points_face[:, i + 15] - nmeans[i]) / nstds[i] # normal2
points_face[:, i + 18] = (points_face[:, i + 18] - nmeans[i]) / nstds[i] # normal3
points_face[:, i + 21] = (points_face[:, i + 21] - nmeans_f[i]) / nstds_f[i] # face normal
# if self.model=='meshsegnet':
# S1 = np.zeros([16000, 16000], dtype='float32')
# S2 = np.zeros([16000, 16000], dtype='float32')
#
# # if torch.cuda.is_available():
# # TX = torch.as_tensor(points_face[:, 9:12]).cuda()
# # TD = torch.cdist(TX, TX)
# # D = TD.cpu().numpy()
# # else:
# D = distance_matrix(points_face[:, 9:12], points_face[:, 9:12])
#
# S1[D<0.1] = 1.0
# S1 = S1 / np.dot(np.sum(S1, axis=1, keepdims=True), np.ones((1, 16000)))
#
# S2[D<0.2] = 1.0
# S2 = S2 / np.dot(np.sum(S2, axis=1, keepdims=True), np.ones((1, 16000)))
#
# return index_face, points_face, label_face, label_face_onehot, self.file_list[item], raw_points_face, RGB_face, S1, S2
# else:
return index_face, points_face, label_face, label_face_onehot, self.file_list[item], raw_points_face, RGB_face
if __name__ == "__main__":
# print(" ")
index_face, points_face, label_face, label_face_onehot, points, _ = get_data('data/test/001.ply')
print(index_face)
# print(index_face.shape, points_face.shape, label_face.shape, label_face_onehot.shape, points.shape)