-
Notifications
You must be signed in to change notification settings - Fork 46
/
Copy pathlogs_to_protobuf.py
244 lines (233 loc) · 12.1 KB
/
logs_to_protobuf.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
from collections import namedtuple
import numpy as np
import pathlib
import re
import scenenet_pb2 as sn
import sys
PoseData = namedtuple('PoseData', ['time', 'camera_position', 'camera_lookat'])
def get_info_log_lines(info_log):
with open(str(info_log),'r') as f:
readlines = f.readlines()
return readlines
def get_text_layout_lines(text_layout_file):
with open(text_layout_file,'r') as f:
readlines = f.readlines()
return readlines
def get_layout_type(line):
layouts = {'office':sn.SceneLayout.OFFICE,
'kitchen':sn.SceneLayout.KITCHEN,
'livingroom':sn.SceneLayout.LIVING_ROOM,
'bedroom':sn.SceneLayout.BEDROOM,
'bathroom':sn.SceneLayout.BATHROOM,
}
for key,layout_enum in layouts.items():
if key in line:
return layout_enum
return None
def process_objects_into_instances(layout_lines):
objects = []
current_object = None
next_line_type = None
for line in layout_lines:
if line.startswith('#first'):
break
if line.strip() == 'object':
next_line_type = 'object'
continue
if line.strip() == 'wnid':
next_line_type = 'wnid'
continue
if line.strip() == 'scale':
next_line_type = 'scale'
continue
if line.strip() == 'transformation':
next_line_type = 'trans'
continue
if next_line_type == 'object':
if current_object is not None:
objects.append(current_object)
current_object = {}
num_transformations = 0
current_object['hash'] = line.rstrip()
continue
if next_line_type == 'wnid':
current_object['wnid'] = line.rstrip()
continue
if next_line_type == 'scale':
current_object['scale'] = float(line.rstrip())
continue
if next_line_type == 'trans':
if num_transformations == 0:
current_object['transformation'] = []
if num_transformations < 3:
current_object['transformation'].append([float(x) for x in line.rstrip().split()])
num_transformations += 1
continue
if current_object is not None:
objects.append(current_object)
return objects
def parse_log_to_frame_pose_pairs(log_lines,skip_frames = 25):
def chunks(l, n):
for i in range(0, len(l), n):
yield l[i:i + n]
regex = re.compile("time:([e\.\-\d]+) pose:([e\.\-\d]+),([e\.\-\d]+),([e\.\-\d]+) lookat:([e\.\-\d]+),([e\.\-\d]+),([e\.\-\d]+)")
time_center_lookats = []
# Get the list of poses
for line in log_lines:
m = regex.search(line)
if m is not None:
time_center_lookat = PoseData(time=float(m.group(1)),
camera_position=np.array([float(m.group(i)) for i in range(2, 5)]),
camera_lookat=np.array([float(m.group(i)) for i in range(5, 8)]))
time_center_lookats.append(time_center_lookat)
# Take them in pairs (i.e. time and pose of shutter open, time and pose of shutter close),
# and filter out poses without a frame - using the pose_frame_skip info
frame_pose_pair_list = []
for idx,(shutter_open, shutter_close) in enumerate(chunks(time_center_lookats,2)):
if idx % skip_frames == 0:
frame_pose_pair_list.append((shutter_open,shutter_close))
return frame_pose_pair_list
def get_instances(info_lines):
instances = []
for line in info_lines:
if line.startswith('time:'):
break
if line.startswith('instance:'):
instance_dict = {}
try:
instance_num,wnid,english,shapenethash = line.strip().split(';')
except:
try:
instance_num,wnid,english,position,radius,power = line.strip().split(';')
position = position.replace('position[','').replace(']','').split(',')
instance_dict['light_position'] = [float(x) for x in position]
radius = radius.replace('radius[','').replace(']','')
instance_dict['light_radius'] = float(radius)
power = power.replace('power[','').replace(']','').split(',')
instance_dict['light_power'] = [float(x) for x in power]
except:
instance_num,wnid,english,position,v1,v2,power = line.strip().split(';')
position = position.replace('position[','').replace(']','').split(',')
instance_dict['light_position'] = [float(x) for x in position]
v1 = v1.replace('v1[','').replace(']','').split(',')
instance_dict['light_v1'] = [float(x) for x in v1]
v2 = v2.replace('v2[','').replace(']','').split(',')
instance_dict['light_v2'] = [float(x) for x in v2]
power = power.replace('power[','').replace(']','').split(',')
instance_dict['light_power'] = [float(x) for x in power]
shapenethash = None
instance_num = instance_num.split(':')[1]
wnid = wnid.split(',')[0]
english = english.split(',')[0]
if shapenethash == '':
shapenethash = None
instance_dict['instance_num'] = int(instance_num)
instance_dict['wnid'] = wnid
instance_dict['english'] = english
instance_dict['hash'] = shapenethash
instances.append(instance_dict)
return instances
def get_all_instances_dict(layout_lines,info_lines):
objects_and_transforms = process_objects_into_instances(layout_lines)
instances_in_log = get_instances(info_lines)
for idx,instance in enumerate(instances_in_log):
if instance['hash'] is not None:
shapenet_idx = idx
break
for objects_and_transform in objects_and_transforms:
shapenethash = objects_and_transform['hash']
assert instances_in_log[shapenet_idx]['hash'] == shapenethash
instances_in_log[shapenet_idx]['scale'] = objects_and_transform['scale']
instances_in_log[shapenet_idx]['transformation'] = objects_and_transform['transformation']
shapenet_idx += 1
return instances_in_log
def fill_trajectory(info_lines,layout_lines,trajectory,frame_skip=25):
layout = trajectory.layout
layout.layout_type = get_layout_type(layout_lines[0])
layout.model = layout_lines[0].split(': ./')[1].strip()
all_instance_data = get_all_instances_dict(layout_lines,info_lines)
instance = trajectory.instances.add()
instance.instance_id = 0
instance.instance_type = sn.Instance.BACKGROUND
for instance_dict in all_instance_data:
# This is the background object
instance = trajectory.instances.add()
instance.instance_id = instance_dict['instance_num']
instance.semantic_wordnet_id = instance_dict['wnid']
instance.semantic_english = instance_dict['english'].split('.')[0].lower()
if 'light_power' in instance_dict:
instance.instance_type = sn.Instance.LIGHT_OBJECT
instance.light_info.light_output.r = instance_dict['light_power'][0]
instance.light_info.light_output.g = instance_dict['light_power'][1]
instance.light_info.light_output.b = instance_dict['light_power'][2]
instance.light_info.position.x = instance_dict['light_position'][0]
instance.light_info.position.y = instance_dict['light_position'][1]
instance.light_info.position.z = instance_dict['light_position'][2]
if 'light_radius' in instance_dict:
instance.light_info.light_type = sn.LightInfo.SPHERE
instance.light_info.radius = instance_dict['light_radius']
elif 'light_v1' in instance_dict:
instance.light_info.light_type = sn.LightInfo.PARALLELOGRAM
instance.light_info.v1.x = instance_dict['light_v1'][0]
instance.light_info.v1.y = instance_dict['light_v1'][1]
instance.light_info.v1.z = instance_dict['light_v1'][2]
instance.light_info.v2.x = instance_dict['light_v2'][0]
instance.light_info.v2.y = instance_dict['light_v2'][1]
instance.light_info.v2.z = instance_dict['light_v2'][2]
else:
assert False
elif instance_dict['hash'] is None:
instance.instance_type = sn.Instance.LAYOUT_OBJECT
else:
instance.instance_type = sn.Instance.RANDOM_OBJECT
instance.object_info.height_meters = instance_dict['scale']
instance.object_info.shapenet_hash = instance_dict['hash']
instance.object_info.object_pose.translation_x = instance_dict['transformation'][0][3]
instance.object_info.object_pose.translation_y = instance_dict['transformation'][1][3]
instance.object_info.object_pose.translation_z = instance_dict['transformation'][2][3]
instance.object_info.object_pose.rotation_mat11 = instance_dict['transformation'][0][0]
instance.object_info.object_pose.rotation_mat12 = instance_dict['transformation'][0][1]
instance.object_info.object_pose.rotation_mat13 = instance_dict['transformation'][0][2]
instance.object_info.object_pose.rotation_mat21 = instance_dict['transformation'][1][0]
instance.object_info.object_pose.rotation_mat22 = instance_dict['transformation'][1][1]
instance.object_info.object_pose.rotation_mat23 = instance_dict['transformation'][1][2]
instance.object_info.object_pose.rotation_mat31 = instance_dict['transformation'][2][0]
instance.object_info.object_pose.rotation_mat32 = instance_dict['transformation'][2][1]
instance.object_info.object_pose.rotation_mat33 = instance_dict['transformation'][2][2]
pose_data = parse_log_to_frame_pose_pairs(info_lines,skip_frames=frame_skip)
for idx,(shutter_open,shutter_close) in enumerate(pose_data):
view = trajectory.views.add()
view.frame_num = idx * frame_skip
shutter_open_pose = view.shutter_open
shutter_open_pose.camera.x = shutter_open.camera_position[0]
shutter_open_pose.camera.y = shutter_open.camera_position[1]
shutter_open_pose.camera.z = shutter_open.camera_position[2]
shutter_open_pose.lookat.x = shutter_open.camera_lookat[0]
shutter_open_pose.lookat.y = shutter_open.camera_lookat[1]
shutter_open_pose.lookat.z = shutter_open.camera_lookat[2]
shutter_open_pose.timestamp = shutter_open.time
shutter_close_pose = view.shutter_close
shutter_close_pose.camera.x = shutter_close.camera_position[0]
shutter_close_pose.camera.y = shutter_close.camera_position[1]
shutter_close_pose.camera.z = shutter_close.camera_position[2]
shutter_close_pose.lookat.x = shutter_close.camera_lookat[0]
shutter_close_pose.lookat.y = shutter_close.camera_lookat[1]
shutter_close_pose.lookat.z = shutter_close.camera_lookat[2]
shutter_close_pose.timestamp = shutter_close.time
if __name__ == '__main__':
trajectories = sn.Trajectories()
if len(sys.argv) < 3:
print('Please run as python logs_to_protobuf.py /path/to/scenenetrgbd/renderer/build/render_info.log /path/to/scenenetrgbd/camera_trajectory_generator/build/scene_and_trajectory_description.txt')
sys.exit(1)
# Here we add only one trajectory, but it can be a full list
render_log_path = sys.argv[1]
scene_and_trajectory_description_path = sys.argv[2]
info_lines = get_info_log_lines(render_log_path)
layout_lines = get_text_layout_lines(scene_and_trajectory_description_path)
trajectory = trajectories.trajectories.add()
trajectory.render_path = str(pathlib.Path(render_log_path).parent)
fill_trajectory(info_lines,layout_lines,trajectory)
output_path = './scenenet_metadata.pb'
print('Finished processing - writing to:{0}'.format(output_path))
with open(output_path,'wb') as f:
f.write(trajectories.SerializeToString())