-
Notifications
You must be signed in to change notification settings - Fork 4
/
prepare_data.py
190 lines (157 loc) · 9.28 KB
/
prepare_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
import os
import cv2
import glob
import numpy as np
from shutil import copy, copytree, rmtree
from config import DATA_PATH, RAW_DATA_PATH
def prepare_data():
for dataset in RAW_DATA_PATH.iterdir():
print(f'Working on {dataset.name}')
if dataset.name == 'DAVIS':
ds_path = DATA_PATH / 'davis'
elif dataset.name == 'SegTrackv2':
ds_path = DATA_PATH / 'segtrackv2'
elif dataset.name == 'FBMS59':
ds_path = DATA_PATH / 'fbms59'
else:
continue
ds_path.mkdir(exist_ok=True)
rgb_path = ds_path / 'JPEGImages'
anno_path = ds_path / 'annotations'
meta_path = ds_path / 'metadata'
rgb_path.mkdir(exist_ok=True)
anno_path.mkdir(exist_ok=True)
meta_path.mkdir(exist_ok=True)
if dataset.name == 'DAVIS':
rgb_source = dataset / 'JPEGImages' / '1080p'
anno_source = dataset / 'Annotations' / '1080p'
copy(dataset / 'ImageSets' / '1080p' / 'train.txt', meta_path)
copy(dataset / 'ImageSets' / '1080p' / 'trainval.txt', meta_path)
copy(dataset / 'ImageSets' / '1080p' / 'val.txt', meta_path)
copy(dataset / 'Annotations' / 'db_info.yml', meta_path)
elif dataset.name == 'MoCA_filtered':
rgb_source = dataset / 'JPEGImages'
anno_source = dataset / 'Annotations'
elif dataset.name == 'SegTrackv2':
rgb_source = dataset / 'JPEGImages'
anno_source = dataset / 'GroundTruth'
meta_source = dataset / 'ImageSets'
for meta_file in meta_source.iterdir():
copy(meta_file, meta_path)\
for cls in rgb_source.iterdir():
for idx, im_name in enumerate(sorted(cls.iterdir())):
if 'jpg' in im_name.name:
continue
im = cv2.imread(im_name.as_posix())
cv2.imwrite((cls / f'{idx:05}.jpg').as_posix(), im)
im_name.unlink()
for cls in anno_source.iterdir():
if cls.name in ['hummingbird', 'drift', 'bmx', 'monkeydog', 'cheetah']:
if cls.name == 'cheetah':
dir1 = sorted(glob.glob((cls / '1' / '*.bmp').as_posix()))
dir2 = sorted(glob.glob((cls / '2' / '*.png').as_posix()))
else:
dir1 = sorted(glob.glob((cls / '1' / '*.png').as_posix()))
dir2 = sorted(glob.glob((cls / '2' / '*.png').as_posix()))
for i in range(len(dir1)):
im1 = cv2.imread(dir1[i], cv2.IMREAD_GRAYSCALE)
im2 = cv2.imread(dir2[i], cv2.IMREAD_GRAYSCALE)
ims = np.clip(im1 + im2, 0, 255)
cv2.imwrite((cls / f'{i:05}.png').as_posix(), ims)
rmtree((cls / '1').as_posix(), ignore_errors=True)
rmtree((cls / '2').as_posix(), ignore_errors=True)
elif cls.name == 'penguin':
dir1 = sorted(glob.glob((cls / '1' / '*.png').as_posix()))
dir2 = sorted(glob.glob((cls / '2' / '*.png').as_posix()))
dir3 = sorted(glob.glob((cls / '3' / '*.png').as_posix()))
dir4 = sorted(glob.glob((cls / '4' / '*.png').as_posix()))
dir5 = sorted(glob.glob((cls / '5' / '*.png').as_posix()))
dir6 = sorted(glob.glob((cls / '6' / '*.png').as_posix()))
for i in range(len(dir1)):
im1 = cv2.imread(dir1[i], cv2.IMREAD_GRAYSCALE)
im2 = cv2.imread(dir2[i], cv2.IMREAD_GRAYSCALE)
im3 = cv2.imread(dir3[i], cv2.IMREAD_GRAYSCALE)
im4 = cv2.imread(dir4[i], cv2.IMREAD_GRAYSCALE)
im5 = cv2.imread(dir5[i], cv2.IMREAD_GRAYSCALE)
im6 = cv2.imread(dir6[i], cv2.IMREAD_GRAYSCALE)
ims = np.clip(im1 + im2 + im3 + im4 + im5 + im6, 0, 255)
cv2.imwrite((cls / f'{i:05}.png').as_posix(), ims)
rmtree((cls / '1').as_posix(), ignore_errors=True)
rmtree((cls / '2').as_posix(), ignore_errors=True)
rmtree((cls / '3').as_posix(), ignore_errors=True)
rmtree((cls / '4').as_posix(), ignore_errors=True)
rmtree((cls / '5').as_posix(), ignore_errors=True)
rmtree((cls / '6').as_posix(), ignore_errors=True)
if cls.name == 'worm':
for anno_name in sorted(cls.iterdir()):
if anno_name.name == '0000000000.png':
anno_name.unlink()
for idx, anno_name in enumerate(sorted(cls.iterdir())):
if anno_name.is_dir():
continue
os.rename(anno_name, cls / f'{idx:05}.png')
elif dataset.name == 'FBMS59':
train_path = dataset / 'Trainingset'
test_path = dataset / 'Testset'
for path_iter in [train_path, test_path]:
for cls in path_iter.iterdir():
rgb_path_cls = rgb_path / cls.name
anno_path_cls = anno_path / cls.name
rgb_path_cls.mkdir(exist_ok=True)
anno_path_cls.mkdir(exist_ok=True)
for im in sorted(cls.iterdir()):
if im.is_dir():
is_ppm = len(list(im.glob('*.ppm'))) > 0
for mask in sorted(im.iterdir()):
if '.dat' in mask.suffix:
continue
if is_ppm and mask.suffix == '.pgm':
continue
if 'PROB' in mask.name:
continue
if is_ppm:
mask_im = cv2.imread(mask.as_posix())
pmask = cv2.cvtColor(mask_im, cv2.COLOR_RGB2GRAY)
final_mask = ((~(pmask/255 > 0.9))*255).astype(np.uint8)
cv2.imwrite((anno_path_cls / f'{int(mask.name.split(".")[0].split("_")[1]) - 1:05}.png').as_posix(), final_mask)
else:
mask_im = cv2.imread(mask.as_posix())
pmask = cv2.cvtColor(mask_im, cv2.COLOR_RGB2GRAY)
if cls.name == 'marple2':
final_mask = ((pmask/255 > 0.4) * 255).astype(np.uint8)
elif cls.name == 'marple7':
final_mask = ((pmask/255 > 0.05) * 255).astype(np.uint8)
else:
final_mask = (pmask.astype(bool)*255).astype(np.uint8)
if cls.name == 'tennis':
mdx = ''.join([s for s in mask.name if s.isdigit()])
mdx = int(mdx) - 454
cv2.imwrite((anno_path_cls / f'{mdx:05}.png').as_posix(), final_mask)
elif cls.name == 'marple4':
cv2.imwrite((anno_path_cls / f'{int(mask.name.split(".")[0].split("_")[1]) - 324:05}.png').as_posix(), final_mask)
else:
cv2.imwrite((anno_path_cls / f'{int(mask.name.split(".")[0].split("_")[1]) - 1:05}.png').as_posix(), final_mask)
else:
if im.suffix == '.bmf':
continue
if cls.name == 'tennis':
idx = ''.join([s for s in im.name if s.isdigit()])
idx = int(idx) - 454
os.rename(im, cls / f'{idx:05}.jpg')
copy(cls / f'{idx:05}.jpg', rgb_path_cls / f'{idx:05}.jpg')
elif cls.name == 'marple4':
os.rename(im, cls / f'{int(im.name.split(".")[0].split("_")[-1]) - 324:05}.jpg')
copy(cls / f'{int(im.name.split(".")[0].split("_")[-1]) - 324:05}.jpg', rgb_path_cls / f'{int(im.name.split(".")[0].split("_")[-1]) - 324:05}.jpg')
else:
os.rename(im, cls / f'{int(im.name.split(".")[0].split("_")[-1]) - 1:05}.jpg')
copy(cls / f'{int(im.name.split(".")[0].split("_")[-1]) - 1:05}.jpg', rgb_path_cls / f'{int(im.name.split(".")[0].split("_")[-1]) - 1:05}.jpg')
if dataset.name != 'FBMS59':
for obj_rgb in rgb_source.iterdir():
copytree(obj_rgb, rgb_path / obj_rgb.name)
for obj_anno in anno_source.iterdir():
if dataset.name == 'MoCA_filtered':
copy(obj_anno, anno_path / obj_anno.name)
else:
copytree(obj_anno, anno_path / obj_anno.name)
if __name__ == "__main__":
prepare_data()