-
Notifications
You must be signed in to change notification settings - Fork 2
/
hair_data.py
488 lines (425 loc) · 15.9 KB
/
hair_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
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
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
import math
import os
import pickle
import sys
import numpy as np
import torch
from matplotlib import pyplot as plt
from torch.utils.data import DataLoader, Dataset
from torchvision import transforms, utils
import tool_func
from component.data_transforms import (Exposure, Normalize, RandomCrop,
Rescale, ToTensor, Resize_Padding,
ToTensor2)
ROOT_DIR = os.path.dirname(os.path.abspath(__file__))
# where the real image library locate , path is linux style
sys.path.append(os.path.join('/mnt', 'd1p8', 'ming', 'jplin', 'FaceParsing'))
import Parsing as ps
class GeneralDataset(Dataset):
def __init__(self,
options,
mode='train',
from_to_ratio=None,
transform=None):
super(GeneralDataset, self).__init__()
self.options = options
if options:
self.im_size = options['im_size']
self.aug_setting_name = options['aug_setting_name']
self.query_label_names = options['query_label_names']
else:
# test data is pre_define
self.im_size = 512
self.aug_setting_name = 'aug_512_0.6_multi_person'
self.query_label_names = ['hair']
print(self.query_label_names)
self.transform = transform
if mode == 'train':
self.raw_dataset = self.gen_training_data(
self.query_label_names, self.aug_setting_name,
options.get('aug_ids', None), options.get('dataset_names', []))
else:
self.raw_dataset = self.gen_testing_data(
self.query_label_names, self.aug_setting_name,
options.get('dataset_names', []))
image_list = list(range(len(self.raw_dataset)))
if from_to_ratio is not None:
fr = int(from_to_ratio[0] * len(self.raw_dataset))
to = int(from_to_ratio[1] * len(self.raw_dataset))
self.image_ids = image_list[fr:to]
else:
self.image_ids = image_list[:]
def __len__(self):
return len(self.image_ids)
def __getitem__(self, idx):
image_id = self.image_ids[idx]
im = self.raw_dataset.load_image(image_id)
label = self.raw_dataset.load_labels(image_id)
im_info = self.raw_dataset[image_id]['image_path']
x_pos_map = None
y_pos_map = None
gaussian_map = None
pos_map_path = im_info.replace('.jpg', '.pk').replace(
'images', 'positions')
if self.options.get('position_map',
False) and os.path.exists(pos_map_path):
pos_map = pickle.load(open(pos_map_path, 'rb'))
x_pos_map, y_pos_map = pos_map['x_map'], pos_map['y_map']
elif self.options.get('center_map', False):
x_pos_map, y_pos_map = self.get_xy_map(self.im_size)
elif self.options.get('with_gaussian', False):
gaussian_map = self.get_gaussian_map(self.im_size)
res = {
'image': im,
'label': label,
'x_pos': x_pos_map,
'y_pos': y_pos_map,
'g_map': gaussian_map
}
if self.transform:
res = self.transform(res)
return res
def get_info(self, idx):
image_id = self.image_ids[idx]
im_info = self.raw_dataset[image_id]
return im_info
@staticmethod
def get_xy_map(im_size):
'''
im_size: int or list
return: x_mesh , y_mesh : HxW , [-0.5 , 0.5] center at center.
'''
if type(im_size) == int:
h, w = im_size, im_size
elif type(im_size) == list:
h, w = im_size[0], im_size[1]
x_center, y_center = w / 2, h / 2
y_range = np.arange(h)
x_range = np.arange(w)
x_mesh, y_mesh = np.meshgrid(x_range, y_range)
x_mesh = (x_mesh - x_center) / w
y_mesh = (y_mesh - y_center) / h
return x_mesh, y_mesh
@staticmethod
def get_gaussian_map(im_size):
'''
im_size: int or list
return: H x W gaussian map
'''
if type(im_size) == int:
h, w = im_size, im_size
elif type(im_size) == list:
h, w = im_size[0], im_size[1]
x_center, y_center = w / 2, h / 2
y_range = np.arange(h)
x_range = np.arange(w)
x_mesh, y_mesh = np.meshgrid(x_range, y_range)
def _gaussian(x, y, x_center, y_center, sigma):
x = np.abs(x - x_center)
y = np.abs(y - y_center)
ret = np.exp(-(x**2 + y**2) /
(2 * sigma**2)) / (sigma * math.sqrt(2 * math.pi))
ret = ret / (np.max(ret) - np.min(ret))
return ret
gaussian_map = _gaussian(x_mesh, y_mesh, x_center, y_center, 50)
return gaussian_map
def gen_training_data(self,
query_label_names,
aug_setting_name='aug_512_0.8',
aug_ids=[0, 1, 2, 3],
dataset_names=[]):
datasets = []
if len(dataset_names) == 0:
dataset_names = [
'HELENRelabeled', 'MultiPIE', 'HangYang', 'Portrait724'
]
for dataset_name in dataset_names:
datasets.append(
ps.Dataset(
dataset_name,
category='train',
aug_ids=aug_ids,
aug_setting_name=aug_setting_name,
query_label_names=query_label_names))
return ps.CombinedDataset(datasets)
def gen_testing_data(self,
query_label_names,
aug_setting_name='aug_512_0.8',
dataset_names=[]):
datasets = []
if len(dataset_names) == 0:
dataset_names = [
'HELENRelabeled', 'MultiPIE', 'HangYang', 'Portrait724'
]
for dataset_name in dataset_names:
datasets.append(
ps.Dataset(
dataset_name,
category='test',
aug_ids=[0],
aug_setting_name=aug_setting_name,
query_label_names=query_label_names))
return ps.CombinedDataset(datasets)
class AttnSegDataset(Dataset):
def __init__(self,
options,
mode='train',
from_to_ratio=None,
transform=None):
super(AttnSegDataset, self).__init__()
self.options = options
if options:
self.im_size = options['im_size']
self.query_label_names = options['query_label_names']
else:
# test data is pre_define
self.im_size = 512
self.query_label_names = ['hair']
print('query_label_names', self.query_label_names)
self.transform = transform
# create basic dataset instance
if mode == 'train':
self.raw_dataset = self.gen_training_data(
self.query_label_names, options.get('training_dataset', []))
else:
self.raw_dataset = self.gen_testing_data(
self.query_label_names, options.get('test_dataset', []))
# generate index mapping
image_list = list(range(len(self.raw_dataset)))
if from_to_ratio is not None:
fr = int(from_to_ratio[0] * len(self.raw_dataset))
to = int(from_to_ratio[1] * len(self.raw_dataset))
self.image_ids = image_list[fr:to]
else:
self.image_ids = image_list[:]
def __len__(self):
return len(self.image_ids)
def __getitem__(self, idx):
image_id = self.image_ids[idx]
im = self.raw_dataset.load_image(image_id)
label_list = self.raw_dataset.load_labels(image_id)
fa_point_list = self.raw_dataset.load_fa_points(image_id)
res = {'image': im, 'labels': label_list, 'fa_points': fa_point_list}
if self.transform:
res = self.transform(res)
return res
def gen_training_data(self, query_label_names, dataset_names):
ret_dataset = ps.MergeDataset(
dataset_names,
category='train',
query_label_names=query_label_names)
return ret_dataset
def gen_testing_data(self, query_label_names, dataset_names):
ret_dataset = ps.MergeDataset(
dataset_names,
category='test',
query_label_names=query_label_names)
return ret_dataset
# for evaluate
def get_helen_test_data(query_label_names, aug_setting_name):
return ps.Dataset(
'HELENRelabeled_wo_pred',
category='test',
aug_ids=[0],
aug_setting_name=aug_setting_name,
query_label_names=query_label_names)
# calling general dataset
def gen_transform_data_loader(options,
mode='train',
batch_size=1,
shuffle=True,
dataloader=True,
use_original=False):
# define composition of transforms
transform_list = []
if mode == 'train':
transform_list = [
Exposure(options['grey_ratio']),
Rescale(options['crop_size'], options.get('random_scale', 0)),
RandomCrop(options['im_size']),
Normalize(),
ToTensor()
]
elif mode == 'test':
if not use_original:
transform_list = [
Rescale(options['crop_size'], options.get('random_scale',
400)),
RandomCrop(options['im_size']),
Normalize(),
ToTensor()
]
else:
transform_list = [Normalize(), ToTensor()]
_transforms = transforms.Compose(transform_list)
# define pytorch dataset
ds = GeneralDataset(options, mode=mode, transform=_transforms)
print("=> generate data loader: mode({0}) , length({1})".format(
mode, len(ds)))
# define pytorch dataloader
ds_loader = DataLoader(
ds, batch_size=batch_size, shuffle=shuffle, num_workers=12)
if dataloader:
return ds_loader
else:
return ds
# for unit test , test pytorch dataset
def test_dataset(options):
transform = transforms.Compose([
Exposure(options['grey_ratio']),
Rescale(options['crop_size'], options.get('random_scale', 400)),
RandomCrop(options['im_size']),
ToTensor()
])
ds = GeneralDataset(options, mode='train', transform=transform)
for i in range(len(ds)):
sample = ds[i]
print(i, sample['image'].size(), sample['label'].size())
image = np.transpose(sample['image'].numpy(), [1, 2, 0])
fig, axes = plt.subplots(ncols=4)
axes[0].imshow(image)
axes[0].set(title='image')
axes[1].imshow(sample['label'].numpy())
axes[1].set(title='ground-truth')
axes[2].imshow(sample['x_pos'].numpy())
axes[2].set(title='pos_map')
axes[3].imshow(sample['y_pos'].numpy())
axes[3].set(title='pos_map')
plt.show()
if i == 3:
break
# for unit test , test pytorch dataloader
def test_dataloader(options):
transform = transforms.Compose([
Exposure(options['grey_ratio']),
Rescale(options['crop_size'], options.get('random_scale', 400)),
RandomCrop(options['im_size']),
ToTensor()
])
ds = GeneralDataset(options, mode='train', transform=transform)
ds_loader = DataLoader(ds, batch_size=4, shuffle=True, num_workers=1)
def _show_batch(sample_batch):
image_batch, label_batch = sample_batch['image'], sample_batch['label']
batch_size = len(image_batch)
grid = utils.make_grid(image_batch)
plt.figure()
plt.imshow(grid.numpy().transpose((1, 2, 0)))
grid = label_batch.numpy()
print(np.unique(grid))
grids = []
for i in range(batch_size):
grids.append(grid[i])
plt.figure()
plt.imshow(np.concatenate(grids, 1))
for i_batch, sample_batch in enumerate(ds_loader):
print(i_batch, sample_batch['image'].size(),
sample_batch['label'].size())
'''
if i_batch == 3:
_show_batch(sample_batch)
plt.axis('off')
plt.ioff()
plt.show()
break
'''
# calling AttnSegDataset
def gen_transform_data_loader_2(options,
mode='train',
batch_size=1,
shuffle=True,
dataLoader=True):
# define composition of transforms
transform_list = []
if mode == 'train':
transform_list = [
Exposure(options['grey_ratio']),
Resize_Padding(options['im_size']),
Normalize(),
ToTensor2()
]
elif mode == 'test':
transform_list = [
Resize_Padding(options['im_size']),
Normalize(),
ToTensor2()
]
_transforms = transforms.Compose(transform_list)
# define pytorch dataset
ds = AttnSegDataset(options, mode=mode, transform=_transforms)
print("=> generate data loader: mode({0}) , length({1})".format(
mode, len(ds)))
# define pytorch dataloader
ds_loader = DataLoader(
ds, batch_size=batch_size, shuffle=shuffle, num_workers=12)
if DataLoader:
return ds_loader
else:
return ds
def test_dataloader2(options):
transform = transforms.Compose([
Exposure(options['grey_ratio']),
Resize_Padding(options['im_size']),
Normalize(),
ToTensor2()
])
ds = AttnSegDataset(options, mode='train', transform=transform)
print("=> generate data loader: mode({0}) , length({1})".format(
'train', len(ds)))
ds_loader = DataLoader(ds, batch_size=4, shuffle=True, num_workers=1)
def _show_batch(sample_batch):
image_batch, label_batch, fa_point_batch = sample_batch[
'image'], sample_batch['labels'], sample_batch['fa_points']
batch_size = len(image_batch)
# visualize image
grid = utils.make_grid(image_batch)
plt.figure()
plt.imshow(grid.numpy().transpose((1, 2, 0)))
# visualize labels
grid = label_batch[0].numpy()
print(np.unique(grid))
grids = []
for i in range(batch_size):
grids.append(grid[i])
plt.figure()
plt.imshow(np.concatenate(grids, 1))
# grid = label_batch[1].numpy()
print(np.unique(grid))
grids = []
for i in range(batch_size):
grids.append(grid[i])
plt.figure()
plt.imshow(np.concatenate(grids, 1))
# visualize fa points
grid = fa_point_batch[0].numpy()
tool_func.vis_points(image_batch[0].numpy().transpose((1, 2, 0)),
grid[0])
# tool_func.vis_points(image_batch[0].numpy().transpose((1, 2, 0)),
# grid[0])
for i_batch, sample_batch in enumerate(ds_loader):
image = sample_batch['image']
labels = sample_batch['labels']
fa_points = sample_batch['fa_points']
print(image.shape, labels[0].shape, fa_points[0].shape)
if i_batch == 3:
_show_batch(sample_batch)
plt.axis('off')
plt.ioff()
plt.show()
break
if __name__ == '__main__':
import yaml
from torchvision import transforms, utils
# test dataloader
# plt.ion()
# options = yaml.load(
# open(
# os.path.join(ROOT_DIR, 'options',
# 'dfn_hairseg_attention_randomcrop.yaml')))
# print(options)
# #test_dataset(options)
# test_dataloader(options)
# test dataloader2
plt.ion()
options = yaml.load(
open(os.path.join(ROOT_DIR, 'options', 'attnseg.yaml')))
test_dataloader2(options)