-
Notifications
You must be signed in to change notification settings - Fork 2
/
frames_dataset.py
299 lines (254 loc) · 11.2 KB
/
frames_dataset.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
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
import os
from skimage import io, img_as_float32
from skimage.color import gray2rgb
from sklearn.model_selection import train_test_split
import imageio
from imageio import mimread
import numpy as np
from torch.utils.data import Dataset
import pandas as pd
from augmentation import AllAugmentationTransform
import glob
import time
def read_video(name, frame_shape, read_first_frame=False):
"""
Read video which can be:
- an image of concatenated frames
- '.mp4' and'.gif'
- folder with videos
"""
if os.path.isdir(name):
frames = sorted(os.listdir(name))
num_frames = len(frames)
if not read_first_frame:
video_array = np.array(
[img_as_float32(io.imread(os.path.join(name, frames[idx]))) for idx in range(num_frames)])
else:
video_array = np.array(
[img_as_float32(io.imread(os.path.join(name, frames[idx]))) for idx in range(1)])
elif name.lower().endswith('.png') or name.lower().endswith('.jpg'):
image = io.imread(name)
if len(image.shape) == 2 or image.shape[2] == 1:
image = gray2rgb(image)
if image.shape[2] == 4:
image = image[..., :3]
image = img_as_float32(image)
video_array = np.moveaxis(image, 1, 0)
video_array = video_array.reshape((-1,) + frame_shape)
video_array = np.moveaxis(video_array, 1, 2)
elif name.lower().endswith('.gif') or name.lower().endswith('.mp4') or name.lower().endswith('.mov'):
#video = np.array(mimread(name,memtest=False))
reader = imageio.get_reader(name)
driving_video = []
try:
for im in reader:
driving_video.append(im)
if read_first_frame:
break
except RuntimeError:
pass
reader.close()
video = np.array(driving_video)
if len(video.shape) == 3:
video = np.array([gray2rgb(frame) for frame in video])
if video.shape[-1] == 4:
video = video[..., :3]
video_array = img_as_float32(video)
else:
raise Exception("Unknown file extensions %s" % name)
return video_array
class FramesDataset(Dataset):
"""
Dataset of videos, each video can be represented as:
- an image of concatenated frames
- '.mp4' or '.gif'
- folder with all frames
"""
def __init__(self, root_dir, data_list=None, data_list_test=None, frame_shape=(256, 256, 3), id_sampling=False, is_train=True,
random_seed=0, pairs_list=None, augmentation_params=None,read_first_frame=False):
self.root_dir = root_dir
# self.videos = os.listdir(root_dir)
self.frame_shape = tuple(frame_shape)
self.pairs_list = pairs_list
self.id_sampling = id_sampling
self.read_first_frame = read_first_frame
f = open(data_list)
file_list = f.readlines()
if id_sampling:
train_video_ids = []
train_videos = {}
for file_name in file_list:
img_name = file_name.strip().split('/')[1]
instance_id = file_name.strip().split('/')[0]
video_id = instance_id.split('#')[0]
if video_id not in train_video_ids:
train_video_ids.append(video_id)
if video_id not in train_videos.keys():
train_videos[video_id] = {}
if instance_id not in train_videos[video_id].keys():
train_videos[video_id][instance_id] = []
train_videos[video_id][instance_id].append(img_name)
f.close()
else:
train_video_ids = []
train_videos = {}
for file_name in file_list:
img_name = file_name.strip().split('/')[1]
instance_id = file_name.strip().split('/')[0]
if instance_id not in train_video_ids:
train_video_ids.append(instance_id)
if instance_id not in train_videos.keys():
train_videos[instance_id] = []
train_videos[instance_id].append(img_name)
f.close()
f_test = open(data_list_test)
file_list = f_test.readlines()
test_video_ids = []
test_videos = {}
for file_name in file_list:
img_name = file_name.strip().split('/')[1]
instance_id = file_name.strip().split('/')[0]
if instance_id not in test_video_ids:
test_video_ids.append(instance_id)
if instance_id not in test_videos.keys():
test_videos[instance_id] = []
test_videos[instance_id].append(img_name)
f_test.close()
if is_train:
local_dir_name = os.path.join(self.root_dir, 'train')
self.root_dir = local_dir_name
else:
local_dir_name = os.path.join(self.root_dir, 'test')
# print(local_dir_name)
# print(os.path.join(self.root_dir+'test/', test_video_ids[0], test_videos[test_video_ids[0]][0]))
self.root_dir = local_dir_name
self.local_dir = local_dir_name
print(self.root_dir)
# if os.path.exists(os.path.join(root_dir, 'train')):
# assert os.path.exists(os.path.join(root_dir, 'test'))
# print("Use predefined train-test split.")
# if id_sampling:
# train_videos = {os.path.basename(video).split('#')[0] for video in
# os.listdir(os.path.join(root_dir, 'train'))}
# train_videos = list(train_videos)
# else:
# train_videos = os.listdir(os.path.join(root_dir, 'train'))
# test_videos = os.listdir(os.path.join(root_dir, 'test'))
# self.root_dir = os.path.join(self.root_dir, 'train' if is_train else 'test')
# # test_videos = os.listdir(os.path.join(root_dir, 'train'))
# # self.root_dir = os.path.join(self.root_dir, 'train')
# else:
# print("Use random train-test split.")
# train_videos, test_videos = train_test_split(self.videos, random_state=random_seed, test_size=0.2)
if is_train:
# self.videos = train_videos
self.videos = train_video_ids
self.video_dicts = train_videos
else:
# self.videos = test_videos
self.videos = test_video_ids
self.video_dicts = test_videos
# print(len(test_video_ids), len(list(test_videos.keys())))
self.is_train = is_train
if self.is_train:
self.transform = AllAugmentationTransform(**augmentation_params)
else:
self.transform = None
def __len__(self):
return len(self.videos)
def __getitem__(self, idx):
if self.is_train and self.id_sampling:
# name = self.videos[idx]
# path = np.random.choice(glob.glob(os.path.join(self.root_dir, name + '*.mp4')))
name = self.videos[idx]
name_list = sorted(list(self.video_dicts[name].keys()))
video_name = np.random.choice(name_list)
path = os.path.join(self.root_dir, video_name)
else:
name = self.videos[idx]
path = os.path.join(self.root_dir, name)
video_name = name
# video_name = os.path.basename(path)
if self.is_train:
if self.id_sampling:
frames = self.video_dicts[name][video_name]
else:
frames = self.video_dicts[name]
num_frames = len(frames)
frame_idx = np.sort(np.random.choice(num_frames, replace=True, size=2))
video_array = []
for idx in frame_idx:
try:
img = img_as_float32(io.imread(os.path.join(path, frames[idx])))
except TypeError:
img = img_as_float32(io.imread(os.path.join(path, frames[idx].decode())))
if len(img.shape) == 2:
img = gray2rgb(img)
if img.shape[-1] == 4:
img = img[..., :3]
video_array.append(img)
else:
video_array = read_video(path, frame_shape=self.frame_shape, read_first_frame=self.read_first_frame)
num_frames = len(video_array)
frame_idx = np.sort(np.random.choice(num_frames, replace=True, size=2)) if self.is_train else range(num_frames)
video_array = video_array[frame_idx]
if self.transform is not None:
video_array = self.transform(video_array)
out = {}
if self.is_train:
source = np.array(video_array[0], dtype='float32')
driving = np.array(video_array[1], dtype='float32')
# out['driving'] = np.concatenate(driving.transpose((0, 3, 1, 2)))
out['driving'] = driving.transpose((2, 0, 1))
out['source'] = source.transpose((2, 0, 1))
else:
video = np.array(video_array, dtype='float32')
out['video'] = video.transpose((3, 0, 1, 2))
out['name'] = video_name
return out
class DatasetRepeater(Dataset):
"""
Pass several times over the same dataset for better i/o performance
"""
def __init__(self, dataset, num_repeats=100):
self.dataset = dataset
self.num_repeats = num_repeats
def __len__(self):
return self.num_repeats * self.dataset.__len__()
def __getitem__(self, idx):
return self.dataset[idx % self.dataset.__len__()]
class PairedDataset(Dataset):
"""
Dataset of pairs for animation.
"""
def __init__(self, initial_dataset, number_of_pairs, seed=0):
self.initial_dataset = initial_dataset
pairs_list = self.initial_dataset.pairs_list
np.random.seed(seed)
if pairs_list is None:
max_idx = min(number_of_pairs, len(initial_dataset))
nx, ny = max_idx, max_idx
xy = np.mgrid[:nx, :ny].reshape(2, -1).T
number_of_pairs = min(xy.shape[0], number_of_pairs)
self.pairs = xy.take(np.random.choice(xy.shape[0], number_of_pairs, replace=False), axis=0)
else:
videos = self.initial_dataset.videos
name_to_index = {name: index for index, name in enumerate(videos)}
pairs = pd.read_csv(pairs_list)
pairs = pairs[np.logical_and(pairs['source'].isin(videos), pairs['driving'].isin(videos))]
number_of_pairs = min(pairs.shape[0], number_of_pairs)
self.pairs = []
self.start_frames = []
for ind in range(number_of_pairs):
self.pairs.append((name_to_index[pairs['driving'].iloc[ind]], name_to_index[pairs['source'].iloc[ind]]))
def __len__(self):
return len(self.pairs)
def __getitem__(self, idx):
pair = self.pairs[idx]
self.initial_dataset.read_first_frame=False
first = self.initial_dataset[pair[0]]
self.initial_dataset.read_first_frame=True
second = self.initial_dataset[pair[1]]
first = {'driving_' + key: value for key, value in first.items()}
second = {'source_' + key: value for key, value in second.items()}
return {**first, **second}