-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathbatch_load_scannet_data.py
203 lines (172 loc) · 7.63 KB
/
batch_load_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
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
import os
import datetime
import numpy as np
import shutil
import pickle
# import cv2
from PIL import Image
BASE_DIR = os.path.dirname(os.path.abspath(__file__))
ROOT_DIR = BASE_DIR
SCANNET_RGB_DIR = os.path.join(ROOT_DIR, 'data/frames_square/')
SCANNET_MASK_DIR = os.path.join(ROOT_DIR, 'data/scannet_frame_labels/')
SCANNET_BOX_DIR = os.path.join(ROOT_DIR, 'data/scannet_frame_bbox/')
OUTPUT_FOLDER = os.path.join(ROOT_DIR, 'data/maskrcnn_training/')
ALL_SCAN_NAMES = [line.rstrip() for line in open('meta_data/scannet_all_scans.txt')]
TRAIN_SCAN_NAMES = [line.rstrip() for line in open('meta_data/scannetv2_train.txt')]
VAL_SCAN_NAMES = [line.rstrip() for line in open('meta_data/scannetv2_val.txt')]
TEST_SCAN_NAMES = [line.rstrip() for line in open('meta_data/scannetv2_test.txt')]
IMG_W = 320
IMG_H = 240
def export_one_scan(scan_name, output_dir):
rgb_folder = os.path.join(SCANNET_RGB_DIR, scan_name, 'color')
mask_folder = os.path.join(SCANNET_MASK_DIR, scan_name, 'instance-filt')
bbox_folder = os.path.join(SCANNET_BOX_DIR, scan_name)
rgb_fnames = os.listdir(rgb_folder)
mask_fnames_all = os.listdir(mask_folder)
bbox_fnames = os.listdir(bbox_folder)
# check correspondence of filenames of [rgb, mask, bbox], and filter out redundant mask_files
assert len(rgb_fnames) == len(bbox_fnames)
mask_fnames = list()
rgb_remove_list = list()
mask_remove_list = list()
bbox_remove_list = list()
for i in range(len(rgb_fnames)):
fname = rgb_fnames[i].split('.')[0]
# remove files that have no bounding box annotation
bbox_dict_list = pickle.load( open(os.path.join(SCANNET_BOX_DIR, scan_name, fname+'.p'), "rb"))
for bbox_dict in bbox_dict_list:
# Check: valid bounding-boxes should not have `xmin==xmax or ymin==ymax`
bbox = bbox_dict['bbox']
if bbox[0] == bbox[2] or bbox[1] == bbox[3]:
bbox_dict_list.remove(bbox_dict)
if bbox_dict_list == []:
rgb_remove_list.append(fname + '.jpg')
mask_remove_list.append(fname + '.png')
bbox_remove_list.append(fname + '.p')
if fname+'.p' not in bbox_fnames:
print("{} not exist!".format(SCANNET_BOX_DIR + fname + '.p'))
rgb_remove_list.append(fname + '.jpg')
continue
if fname+'.png' in mask_fnames_all:
mask_fnames.append(fname+'.png')
else:
rgb_remove_list.append(fname + '.jpg')
bbox_remove_list.append(fname + '.p')
# remove inconsistent files
for fname in rgb_remove_list:
rgb_fnames.remove(fname)
for fname in mask_remove_list:
mask_fnames.remove(fname)
for fname in bbox_remove_list:
bbox_fnames.remove(fname)
# ======================================================
# Store images and mask-labels in npy
# ======================================================
# img_npy_fpath = os.path.join(OUTPUT_FOLDER, scan_name + "_rgb.npy")
# im_array = []
# mask_npy_fpath = os.path.join(OUTPUT_FOLDER, scan_name + "_mask.npy")
# mask_array = []
#
# for fname in rgb_fnames:
# fname = fname.split('.')[0]
# # RGB
# image = cv2.imread(os.path.join(SCANNET_RGB_DIR, scan_name, 'color', fname+'.jpg'))
# im_array.append(image)
# # label-mask
# mask = cv2.imread(os.path.join(SCANNET_MASK_DIR, scan_name, 'instance-filt', fname+'.png'))
# mask = cv2.resize(mask, (IMG_W, IMG_H))
# mask_array.append(mask)
#
# np.save(img_npy_fpath, np.array(im_array))
# np.save(mask_npy_fpath, np.array(mask_array))
# ======================================================
# ======================================================
# ======================================================
# Store images and mask-labels in organized folders
# ======================================================
# copy and rename files
for fname in rgb_fnames:
dst = os.path.join(output_dir, 'raw_rgb/', '_'.join([scan_name, fname]))
if not os.path.exists(os.path.join(output_dir, 'raw_rgb/')):
os.mkdir(os.path.join(output_dir, 'raw_rgb/'))
# dst = os.path.join(OUTPUT_FOLDER, 'raw_rgb/', scan_name)
# if not os.path.exists(dst):
# os.makedirs(dst)
shutil.copy(rgb_folder + "/" + fname, dst)
for fname in mask_fnames:
dst = os.path.join(output_dir, 'label_mask/', '_'.join([scan_name, fname]))
if not os.path.exists(os.path.join(output_dir, 'label_mask/')):
os.mkdir(os.path.join(output_dir, 'label_mask/'))
# dst = os.path.join(OUTPUT_FOLDER, 'label_mask/', scan_name)
# if not os.path.exists(dst):
# os.makedirs(dst)
im = Image.open(os.path.join(SCANNET_MASK_DIR, scan_name, 'instance-filt', fname))
im_resized = im.resize((IMG_W, IMG_H))
# im_resized.save(dst + '/' + fname)
im_resized.save(dst)
for fname in bbox_fnames:
dst = os.path.join(output_dir, 'bbox/', '_'.join([scan_name, fname]))
if not os.path.exists(os.path.join(output_dir, 'bbox/')):
os.mkdir(os.path.join(output_dir, 'bbox/'))
# bbox = pickle.load( open(os.path.join(SCANNET_BOX_DIR, scan_name, fname), "rb"))
shutil.copy(bbox_folder + "/" + fname, dst)
# ======================================================
# ======================================================
def batch_export(data_split):
output_dir = ''
scan_names = ''
if data_split == 'train':
output_dir = os.path.join(OUTPUT_FOLDER, 'train')
scan_names = TRAIN_SCAN_NAMES
elif data_split == 'valid':
output_dir = os.path.join(OUTPUT_FOLDER, 'valid')
scan_names = VAL_SCAN_NAMES
elif data_split == 'test':
output_dir = os.path.join(OUTPUT_FOLDER, 'test')
scan_names = TEST_SCAN_NAMES
elif data_split == 'all':
output_dir = os.path.join(OUTPUT_FOLDER, 'all')
scan_names = ALL_SCAN_NAMES
else:
Exception("Invalid `data_split`, please choose from: `all`, `train`,`valid`, `test`")
if not os.path.exists(output_dir):
print('Creating new data folder: {}'.format(output_dir))
os.mkdir(output_dir)
for scan_name in scan_names:
print('-' * 20 + 'begin')
print(datetime.datetime.now())
print(scan_name)
# if os.path.isfile(output_filename_prefix + '_vert.npy'):
# print('File already exists. skipping.')
# print('-' * 20 + 'done')
# continue
# export_one_scan(scan_name, output_filename_prefix)
try:
export_one_scan(scan_name, output_dir)
except:
print('Failed export scan: %s' % scan_name)
print('-' * 20 + 'done')
if __name__ == '__main__':
import argparse
parser = argparse.ArgumentParser(
description=__doc__)
parser.add_argument('--data-split', default='train', help='valid options: `all`, `train`,`valid`, `test` ')
args = parser.parse_args()
batch_export(args.data_split)
# # TEST
# fpath = OUTPUT_FOLDER + '/scene0000_00_rgb.npy'
# img_array = np.load(fpath)
# mask_path = OUTPUT_FOLDER + '/scene0000_00_mask.npy'
# mask_array = np.load(mask_path)
# assert len(img_array) == len(mask_array)
# count = 0
# # TODO: fix PIL.Image read image as BGR
# for i in range(5):
# im = Image.fromarray(img_array[i])
# # im.show()
# im.save(OUTPUT_FOLDER + '/raw_{}.jpg'.format(count))
#
# mask = Image.fromarray(mask_array[i])
# # mask.show()
# mask.save(OUTPUT_FOLDER + '/mask_{}.png'.format(count))
# count += 1