-
Notifications
You must be signed in to change notification settings - Fork 2
/
prepare_scannet_data.py
164 lines (135 loc) · 6.38 KB
/
prepare_scannet_data.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
import argparse
import pprint
import time
import os.path as osp
import multiprocessing as mp
from tqdm import tqdm
import numpy as np
from referit3d.in_out.scannet_scan import ScannetScan, ScannetDataset
from referit3d.utils import immediate_subdirectories, create_dir, pickle_data, str2bool
"""
-top-scan-dir /home/e/scannet_dataset/scannet/scans -top-save-dir /home/e/scannet_dataset/scannet/scan_4_nr3d --load-dense False --save-jpg True --imgsize 128 --apply-global-alignment False --geo True --twoStreams True
"""
def parse_args():
parser = argparse.ArgumentParser(description='ReferIt3D')
parser.add_argument('-top-scan-dir', required=True, type=str,
help='the path to the downloaded ScanNet scans')
parser.add_argument('-top-save-dir', required=True, type=str,
help='the path of the directory to be saved preprocessed scans as a .pkl')
# Optional arguments.
parser.add_argument('--n-processes', default=-1, type=int,
help='the number of processes, -1 means use the available max')
parser.add_argument('--process-only-zero-view', default=True, type=str2bool,
help='00_view of scans are used')
parser.add_argument('--verbose', default=True, type=str2bool, help='')
parser.add_argument('--apply-global-alignment', default=True, type=str2bool,
help='rotate/translate entire scan globally to aligned it with other scans')
# Eslam
parser.add_argument('--load-dense', default=False, type=str2bool, help='Load dense version of point-clouds')
parser.add_argument('--save-jpg', default=False, type=str2bool, help='Save projected images directly')
parser.add_argument('--imgsize', default=32, type=int, help='Size of projected image')
parser.add_argument('--cocoon', default=False, type=str2bool, help='Save cocoon images for each object')
parser.add_argument('--twoStreams', default=False, type=str2bool, help='Save 2d images for each object and raw pc')
parser.add_argument('--geo', default=False, type=str2bool, help='Save 2d images for each object and raw pc')
parser.add_argument('--camaug', default=0, type=int, help='number of camera augmented frames.'
' set it to 0 to deactivate camaug')
parser.add_argument('--savepickle', default=True, type=str2bool, help='This flag indicates whether u wanna save'
' a pickle or not')
ret = parser.parse_args()
# Print the args
args_string = pprint.pformat(vars(ret))
print(args_string)
return ret
if __name__ == '__main__':
args = parse_args()
if args.process_only_zero_view:
tag = 'keep_all_points_00_view'
else:
tag = 'keep_all_points'
if args.apply_global_alignment:
tag += '_with_global_scan_alignment'
else:
tag += '_no_global_scan_alignment'
if args.load_dense:
tag += '_densePCLoaded'
if args.save_jpg:
tag += '_saveJPG'
if args.cocoon:
tag += '_cocoon'
if args.twoStreams:
tag += '_twoStreams'
if args.geo:
tag += '_GEO'
if args.camaug:
tag += '_camaug'
# Read all scan files.
all_scan_ids = [osp.basename(i) for i in immediate_subdirectories(args.top_scan_dir)]
print('{} scans found.'.format(len(all_scan_ids)))
kept_scan_ids = []
if args.process_only_zero_view:
for si in all_scan_ids:
if si.endswith('00'):
kept_scan_ids.append(si)
all_scan_ids = kept_scan_ids
print('Working with {} scans.'.format(len(all_scan_ids)))
if args.load_dense and False:
list_all_scans = np.array_split(np.array(all_scan_ids), 7)
else:
list_all_scans = [all_scan_ids]
# Prepare ScannetDataset
idx_to_semantic_class_file = 'referit3d/data/mappings/scannet_idx_to_semantic_class.json'
instance_class_to_semantic_class_file = 'referit3d/data/mappings/scannet_instance_class_to_semantic_class.json'
axis_alignment_info_file = 'referit3d/data/scannet/scans_axis_alignment_matrices.json'
scannet = ScannetDataset(args.top_scan_dir,
idx_to_semantic_class_file,
instance_class_to_semantic_class_file,
axis_alignment_info_file)
def scannet_loader(scan_id):
"""Helper function to load the scans in memory.
:param scan_id:
:return: the loaded scan.
"""
global scannet, args
print("scan_id = ", scan_id)
scan_i = ScannetScan(scan_id, scannet, args.apply_global_alignment, load_dense=args.load_dense,
save_jpg=args.save_jpg, img_size=args.imgsize, top_scan_dir=args.top_scan_dir,
cocoon=args.cocoon, camaug=args.camaug)
if args.load_dense:
scan_i.load_point_clouds_of_all_objects_dense()
else:
scan_i.load_point_clouds_of_all_objects()
if args.save_jpg:
scan_i.pc = None
scan_i.color = None
return scan_i
if args.verbose:
print('Loading scans in memory...')
start_time = time.time()
for id, all_scan_ids in enumerate(list_all_scans):
print("Start processing of id = ", id)
n_items = len(all_scan_ids)
if args.n_processes == -1:
n_processes = min(mp.cpu_count(), n_items)
pool = mp.Pool(n_processes)
chunks = int(n_items / n_processes)
all_scans = dict()
for i, data in enumerate(pool.imap(scannet_loader, all_scan_ids, chunksize=chunks)):
all_scans[all_scan_ids[i]] = data
pool.close()
pool.join()
if args.verbose:
print("Loading raw data took {:.4} minutes.".format((time.time() - start_time) / 60.0))
# Save data
if args.savepickle:
if args.verbose:
print('Saving the results.')
all_scans = list(all_scans.values())
save_dir = create_dir(osp.join(args.top_save_dir, tag))
if args.load_dense:
save_file = osp.join(save_dir, tag + "_" + str(id) + '.pkl')
else:
save_file = osp.join(save_dir, tag + '.pkl')
pickle_data(save_file, scannet, all_scans)
print("Finish processing of id = ", id)
if args.verbose:
print('All done.')