-
Notifications
You must be signed in to change notification settings - Fork 42
/
pc_transforms.py
146 lines (104 loc) · 4.52 KB
/
pc_transforms.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
import numpy as np
import torch
import torch.nn as nn
class Compose(object):
def __init__(self,co_transforms):
self.co_transforms = co_transforms
def __call__(self, inputs, targets):
for transforms in self.co_transforms:
inputs,targets = transforms(inputs,targets) #
return inputs,targets
class ArrayToTensor(object):
def __call__(self,array):
assert(isinstance(array,np.ndarray))
#array = np.transpose(array, (2,0,1))
# handle numpy array
tensor = torch.from_numpy(array.copy())
tensor = torch.unsqueeze(tensor,dim=0)
return tensor.float()
class Jitter_PC(object):
def __init__(self,sigma, clip):
self.sigma = sigma
self.clip = clip
assert (clip > 0)
def __call__(self,input,target):
N,C = input.shape
jittered_data_input = np.clip(self.sigma * np.random.randn(N, C), -1*self.clip, self.clip)
jittered_data_input += input
N,C = target.shape
jittered_data_output = np.clip(self.sigma * np.random.randn(N, C), -1*self.clip, self.clip)
jittered_data_output += target
return jittered_data_input,jittered_data_output
class Scale(object):
def __init__(self,low, high):
self.low = low
self.high = high
def __call__(self,input,target):
scale = np.random.uniform(low=self.low, high=self.high)
input = input * scale
target = target * scale
return input, target
class Shift(object):
def __init__(self,low, high):
self.low = low
self.high = high
def __call__(self,input,target):
shift = np.random.uniform(self.low, self.high,(1,3)) #
input += shift
target += shift
return input, target
class Random_Rotate(object):
def __call__(self,input,target):
rotation_angle = np.random.uniform() * 2 * np.pi
cosval = np.cos(rotation_angle)
sinval = np.sin(rotation_angle)
rotation_matrix = np.array([[cosval, 0, sinval],
[0, 1, 0],
[-sinval, 0, cosval]])
rotated_input = np.dot(input.reshape((-1, 3)), rotation_matrix)
rotated_target = np.dot(target.reshape((-1, 3)), rotation_matrix)
return rotated_input, rotated_target
class Random_Rotate_90(object):
def __call__(self,input,target):
rotation_angle = np.random.randint(low=0, high=4) * (np.pi / 2.0)
cosval = np.cos(rotation_angle)
sinval = np.sin(rotation_angle)
rotation_matrix = np.array([[cosval, 0, sinval],
[0, 1, 0],
[-sinval, 0, cosval]])
rotated_input = np.dot(input.reshape((-1, 3)), rotation_matrix)
rotated_target = np.dot(target.reshape((-1, 3)), rotation_matrix)
return rotated_input, rotated_target
class Rotate_90(object):
def __init__(self,args,axis,angle=1.0):
self.angle = angle;
self.args = args;
self.axis = axis
def __call__(self,input,target):
if self.args.net_name == 'shape_completion':
rotation_angle = self.angle * (np.pi / 2.0)
cosval = np.cos(rotation_angle)
sinval = np.sin(rotation_angle)
if self.axis =='x':
rotation_matrix = np.array([[1, 0, 0],
[0, cosval, -sinval],
[0, sinval, cosval]])
if self.axis == 'y':
rotation_matrix = np.array([[cosval, 0, sinval],
[0, 1, 0],
[-sinval, 0, cosval]])
if self.axis == 'z':
rotation_matrix = np.array([[cosval, -sinval, 0],
[sinval, cosval, 0],
[0, 0, -1]])
if self.axis == 'shape_complete':
rotation_matrix = np.array([[1.0, 0.0, 0.0],
[0.0, 0.0, 1.0],
[0.0, 1.0, 0.0]])
# np.array([0.173178189568194, 0.378401247653964, - 0.909297426825682],
# [0.172881825917964, - 0.920591658450853, - 0.350175488374015],
# [0.969598467885110, 0.096558242344360, 0.224845095366153]])
rotated_input = np.dot(input.reshape((-1, 3)), rotation_matrix)
return rotated_input, target
else:
return input,target