forked from charlesq34/pointnet2
-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathfold_dataset.py
193 lines (165 loc) · 7.04 KB
/
fold_dataset.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
'''
ModelNet dataset. Support ModelNet40, ModelNet10, XYZ and normal channels. Up to 10000 points.
'''
import os
import os.path
import numpy as np
import sys
from glob import glob
from collections import Counter
import tensorflow as tf
BASE_DIR = os.path.dirname(os.path.abspath(__file__))
ROOT_DIR = BASE_DIR
sys.path.append(os.path.join(ROOT_DIR, 'utils'))
import provider
def pc_normalize(pc):
l = pc.shape[0]
centroid = np.mean(pc, axis=0)
pc = pc - centroid
m = np.max(np.sqrt(np.sum(pc ** 2, axis=1)))
pc = pc / m
return pc
class PFRDataset:
def __init__(
self,
root,
batch_size=32,
npoints=1024,
split='train',
normalize=True,
normal_channel=True,
cache_size=15000,
shuffle=None,
shuffle_points=False,
scale_low=0.7,
scale_high=1.3,
shift_range=0.3,
jitter_sigma=0.005,
add_n_c_info=True,
omit_parameters_ranges=[],
to_categorical_indexes=[],
to_categorical_sizes=[]
):
self.root = root
self.batch_size = batch_size
self.npoints = npoints
self.normalize = normalize
self.add_n_c_info = add_n_c_info
if add_n_c_info:
self.n_c = np.expand_dims(np.array(np.arange(npoints) / npoints), axis=1)
self.classes_names = ['.'.join(j) for j in sorted([i.split('.') for i in os.listdir(self.root)], key=lambda x: (x[0], int(x[1])))]
self.classes = dict(zip(self.classes_names, range(len(self.classes_names))))
self.normal_channel = normal_channel
self.shuffle_points = shuffle_points
self.scale_low = scale_low
self.scale_high = scale_high
self.shift_range = shift_range
self.jitter_sigma = jitter_sigma
self.omit_parameters_ranges = omit_parameters_ranges
self.to_categorical_indexes = to_categorical_indexes
self.to_categorical_sizes = to_categorical_sizes
assert split == 'train' or split == 'test'
# list of (shape_name, shape_txt_file_path) tuple
self.datapath = sorted(
[(i.split('/')[2], i) for i in sorted(glob(self.root + f'/*/{split}/*.npy'))],
key=lambda x: (x[0][0], int(x[0][2:]))
)
self.cache_size = cache_size # how many data points to cache in memory
self.cache = {} # from index to (point_set, cls) tuple
self.get_classes_weights()
if shuffle is None:
if split == 'train':
self.shuffle = True
else:
self.shuffle = False
else:
self.shuffle = shuffle
self.reset()
def _augment_batch_data(self, batch_data):
if self.normal_channel:
rotated_data = provider.rotate_point_cloud_with_normal(batch_data)
rotated_data = provider.rotate_perturbation_point_cloud_with_normal(rotated_data)
else:
rotated_data = provider.rotate_point_cloud(batch_data)
rotated_data = provider.rotate_perturbation_point_cloud(rotated_data)
jittered_data = provider.random_scale_point_cloud(rotated_data[:, :, 0:3], scale_low=self.scale_low, scale_high=self.scale_high)
jittered_data = provider.shift_point_cloud(jittered_data, shift_range=self.shift_range)
jittered_data = provider.jitter_point_cloud(jittered_data, sigma=self.jitter_sigma, clip=0.1)
rotated_data[:, :, 0:3] = jittered_data
if self.shuffle_points:
return provider.shuffle_points(rotated_data)
else:
return rotated_data
def _get_item(self, index):
if index in self.cache:
point_set, cls = self.cache[index]
else:
fn = self.datapath[index]
cls = self.classes[self.datapath[index][0]]
cls = np.array([cls]).astype(np.int32)
if self.normal_channel:
point_set = np.load(fn[1])[:, :]
for i in range(len(self.omit_parameters_ranges) - 1, -1, -2):
point_set = np.concatenate(
[point_set[:, :self.omit_parameters_ranges[i - 1]], point_set[:, self.omit_parameters_ranges[i]:]], axis=1
)
else:
point_set = np.load(fn[1])[:, :3]
if len(self.cache) < self.cache_size:
self.cache[index] = (point_set, cls)
# Take exactly n npoints
ind = np.arange(point_set.shape[0])
if len(ind) > self.npoints:
ind = np.sort(np.random.choice(ind, self.npoints, replace=False))
else:
ind = np.sort(np.random.choice(ind, self.npoints, replace=True))
point_set = point_set[ind, :]
for cat_ind, cat_size in zip(self.to_categorical_indexes, self.to_categorical_sizes):
cat = tf.keras.utils.to_categorical(point_set[:, cat_ind], num_classes=cat_size)
point_set = np.concatenate([point_set[:, :cat_ind], cat, point_set[:, cat_ind+1:]], axis=1)
if self.add_n_c_info:
point_set = np.concatenate([point_set, self.n_c], axis=1)
if self.normalize:
point_set[:, 0:3] = pc_normalize(point_set[:, 0:3])
if not self.normal_channel:
point_set = point_set[:, 0:3]
return point_set, cls
def __getitem__(self, index):
return self._get_item(index)
def __len__(self):
return len(self.datapath)
def num_channel(self):
return self._get_item(0)[0].shape[1]
def reset(self):
self.idxs = np.arange(0, len(self.datapath))
if self.shuffle:
np.random.shuffle(self.idxs)
self.num_batches = (len(self.datapath) + self.batch_size - 1) // self.batch_size
self.batch_idx = 0
def has_next_batch(self):
return self.batch_idx < self.num_batches
def next_batch(self, augment=False):
''' returned dimension may be smaller than self.batch_size '''
start_idx = self.batch_idx * self.batch_size
end_idx = min((self.batch_idx + 1) * self.batch_size, len(self.datapath))
bsize = end_idx - start_idx
batch_data = np.zeros((bsize, self.npoints, self.num_channel()))
batch_label = np.zeros((bsize), dtype=np.int32)
batch_cls_weights = np.zeros((bsize), dtype=np.float32)
for i in range(bsize):
ps, cls = self._get_item(self.idxs[i + start_idx])
batch_data[i] = ps
batch_label[i] = cls
batch_cls_weights[i] = self.weights[cls[0]]
self.batch_idx += 1
if augment:
batch_data = self._augment_batch_data(batch_data)
return batch_data, batch_label, batch_cls_weights
def get_classes_weights(self):
classes = [j[0] for j in self.datapath]
weights = {k: 1/v for k,v in Counter(classes).items()}
mean = np.mean(list(weights.values()))
weights = {k: v / mean for k, v in weights.items()}
sorted_weights = sorted(weights.items(), key=lambda x: (x[0][0], x[0].split('.')[2:]))
# return [i[1] for i in sorted_weights]
self.weights = [i[1] for i in sorted_weights]