-
Notifications
You must be signed in to change notification settings - Fork 299
/
point_cloud_ops.py
202 lines (188 loc) · 7.02 KB
/
point_cloud_ops.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
import time
import numba
import numpy as np
@numba.jit(nopython=True)
def _points_to_voxel_reverse_kernel(
points,
voxel_size,
coors_range,
num_points_per_voxel,
coor_to_voxelidx,
voxels,
coors,
max_points=35,
max_voxels=20000,
):
# put all computations to one loop.
# we shouldn't create large array in main jit code, otherwise
# reduce performance
N = points.shape[0]
# ndim = points.shape[1] - 1
ndim = 3
ndim_minus_1 = ndim - 1
grid_size = (coors_range[3:] - coors_range[:3]) / voxel_size
# np.round(grid_size)
# grid_size = np.round(grid_size).astype(np.int64)(np.int32)
grid_size = np.round(grid_size, 0, grid_size).astype(np.int32)
coor = np.zeros(shape=(3,), dtype=np.int32)
voxel_num = 0
failed = False
for i in range(N):
failed = False
for j in range(ndim):
c = np.floor((points[i, j] - coors_range[j]) / voxel_size[j])
if c < 0 or c >= grid_size[j]:
failed = True
break
coor[ndim_minus_1 - j] = c
if failed:
continue
voxelidx = coor_to_voxelidx[coor[0], coor[1], coor[2]]
if voxelidx == -1:
voxelidx = voxel_num
if voxel_num >= max_voxels:
break
voxel_num += 1
coor_to_voxelidx[coor[0], coor[1], coor[2]] = voxelidx
coors[voxelidx] = coor
num = num_points_per_voxel[voxelidx]
if num < max_points:
voxels[voxelidx, num] = points[i]
num_points_per_voxel[voxelidx] += 1
return voxel_num
@numba.jit(nopython=True)
def _points_to_voxel_kernel(
points,
voxel_size,
coors_range,
num_points_per_voxel,
coor_to_voxelidx,
voxels,
coors,
max_points=35,
max_voxels=20000,
):
# need mutex if write in cuda, but numba.cuda don't support mutex.
# in addition, pytorch don't support cuda in dataloader(tensorflow support this).
# put all computations to one loop.
# we shouldn't create large array in main jit code, otherwise
# decrease performance
N = points.shape[0]
# ndim = points.shape[1] - 1
ndim = 3
grid_size = (coors_range[3:] - coors_range[:3]) / voxel_size
# grid_size = np.round(grid_size).astype(np.int64)(np.int32)
grid_size = np.round(grid_size, 0, grid_size).astype(np.int32)
lower_bound = coors_range[:3]
upper_bound = coors_range[3:]
coor = np.zeros(shape=(3,), dtype=np.int32)
voxel_num = 0
failed = False
for i in range(N):
failed = False
for j in range(ndim):
c = np.floor((points[i, j] - coors_range[j]) / voxel_size[j])
if c < 0 or c >= grid_size[j]:
failed = True
break
coor[j] = c
if failed:
continue
voxelidx = coor_to_voxelidx[coor[0], coor[1], coor[2]]
if voxelidx == -1:
voxelidx = voxel_num
if voxel_num >= max_voxels:
break
voxel_num += 1
coor_to_voxelidx[coor[0], coor[1], coor[2]] = voxelidx
coors[voxelidx] = coor
num = num_points_per_voxel[voxelidx]
if num < max_points:
voxels[voxelidx, num] = points[i]
num_points_per_voxel[voxelidx] += 1
return voxel_num
def points_to_voxel(
points, voxel_size, coors_range, max_points=35, reverse_index=True, max_voxels=20000
):
"""convert kitti points(N, >=3) to voxels. This version calculate
everything in one loop. now it takes only 4.2ms(complete point cloud)
with jit and 3.2ghz cpu.(don't calculate other features)
Note: this function in ubuntu seems faster than windows 10.
Args:
points: [N, ndim] float tensor. points[:, :3] contain xyz points and
points[:, 3:] contain other information such as reflectivity.
voxel_size: [3] list/tuple or array, float. xyz, indicate voxel size
coors_range: [6] list/tuple or array, float. indicate voxel range.
format: xyzxyz, minmax
max_points: int. indicate maximum points contained in a voxel.
reverse_index: boolean. indicate whether return reversed coordinates.
if points has xyz format and reverse_index is True, output
coordinates will be zyx format, but points in features always
xyz format.
max_voxels: int. indicate maximum voxels this function create.
for second, 20000 is a good choice. you should shuffle points
before call this function because max_voxels may drop some points.
Returns:
voxels: [M, max_points, ndim] float tensor. only contain points.
coordinates: [M, 3] int32 tensor.
num_points_per_voxel: [M] int32 tensor.
"""
if not isinstance(voxel_size, np.ndarray):
voxel_size = np.array(voxel_size, dtype=points.dtype)
if not isinstance(coors_range, np.ndarray):
coors_range = np.array(coors_range, dtype=points.dtype)
voxelmap_shape = (coors_range[3:] - coors_range[:3]) / voxel_size
voxelmap_shape = tuple(np.round(voxelmap_shape).astype(np.int32).tolist())
if reverse_index:
voxelmap_shape = voxelmap_shape[::-1]
# don't create large array in jit(nopython=True) code.
num_points_per_voxel = np.zeros(shape=(max_voxels,), dtype=np.int32)
coor_to_voxelidx = -np.ones(shape=voxelmap_shape, dtype=np.int32)
voxels = np.zeros(
shape=(max_voxels, max_points, points.shape[-1]), dtype=points.dtype
)
coors = np.zeros(shape=(max_voxels, 3), dtype=np.int32)
if reverse_index:
voxel_num = _points_to_voxel_reverse_kernel(
points,
voxel_size,
coors_range,
num_points_per_voxel,
coor_to_voxelidx,
voxels,
coors,
max_points,
max_voxels,
)
else:
voxel_num = _points_to_voxel_kernel(
points,
voxel_size,
coors_range,
num_points_per_voxel,
coor_to_voxelidx,
voxels,
coors,
max_points,
max_voxels,
)
coors = coors[:voxel_num]
voxels = voxels[:voxel_num]
num_points_per_voxel = num_points_per_voxel[:voxel_num]
return voxels, coors, num_points_per_voxel
@numba.jit(nopython=True)
def bound_points_jit(points, upper_bound, lower_bound):
# to use nopython=True, np.bool is not supported. so you need
# convert result to np.bool after this function.
N = points.shape[0]
ndim = points.shape[1]
keep_indices = np.zeros((N,), dtype=np.int32)
success = 0
for i in range(N):
success = 1
for j in range(ndim):
if points[i, j] < lower_bound[j] or points[i, j] >= upper_bound[j]:
success = 0
break
keep_indices[i] = success
return keep_indices