/
data_loader.py
151 lines (121 loc) · 5.8 KB
/
data_loader.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
from torch.utils import data
from torchvision import transforms as T
from torchvision.datasets import ImageFolder
from PIL import Image
import torch
import os
import random
import tqdm
class EncoderFolder(data.Dataset):
"""Dataset class for the CelebA dataset."""
def __init__(self, image_dir, transform, mode):
"""Initialize and preprocess the CelebA dataset."""
self.image_dir = image_dir
self.images = list(map(lambda x: os.path.join(image_dir+'train', x), os.listdir(image_dir+'train')))
self.transform = transform
self.mode = mode
train_path = 'pretrain_ch.pkl'
if os.path.isfile(train_path):
self.train_dataset = torch.load(train_path)
else:
self.train_dataset = []
self.preprocess()
torch.save(self.train_dataset, train_path)
self.num_images = len(self.train_dataset)
def preprocess(self):
"""Preprocess the CelebA attribute file."""
random.seed(1234)
random.shuffle(self.images)
for i, img in enumerate(tqdm.tqdm(self.images)):
style_idx = int(img.split('_')[0][len(self.image_dir+'train/'):])
char_idx = int(img.split('_')[1][:-len(".png")])
style_trg_idx = []
char_trg_idx = []
style_target = random.choice([x for x in self.images
if str(style_idx)+'_' not in x and '_'+str(char_idx) in x])
style_trg_idx.append(int(style_target.split('_')[0][len(self.image_dir):]))
char_target = random.choice([x for x in self.images
if str(style_idx)+'_' in x and '_'+str(char_idx) not in x])
char_trg_idx.append(int(char_target.split('_')[1][:-len(".png")]))
self.train_dataset.append([img, style_target, char_target, style_idx, char_idx, style_trg_idx, char_trg_idx])
def __getitem__(self, index):
"""Return one image and its corresponding attribute label."""
dataset = self.train_dataset if self.mode == 'train' else self.test_dataset
src, style_trg, char_trg, src_style, src_char, trg_style, trg_char = dataset[index]
src = self.transform(Image.open(src))
style_trg = self.transform(Image.open(style_trg))
char_trg = self.transform(Image.open(char_trg))
return src, style_trg, char_trg, src_style, src_char,\
torch.LongTensor(trg_style), torch.LongTensor(trg_char)
def __len__(self):
"""Return the number of images."""
return self.num_images
class ImageFolder(data.Dataset):
"""Dataset class for the CelebA dataset."""
def __init__(self, image_dir, transform, mode):
"""Initialize and preprocess the CelebA dataset."""
self.image_dir = image_dir
self.train_images = list(map(lambda x: os.path.join(image_dir+'train', x), os.listdir(image_dir+'train')))
self.test_images = list(map(lambda x: os.path.join(image_dir+'test', x), os.listdir(image_dir+'test')))
self.transform = transform
self.mode = mode
test_path = 'test_eng.pkl'
if os.path.isfile(test_path):
self.test_dataset = torch.load(test_path)
else:
self.test_dataset = []
self.preprocess()
torch.save(self.test_dataset, test_path)
if mode == 'train':
self.num_images = len(self.train_images)
else:
self.num_images = len(self.test_dataset)
def preprocess(self):
"""Preprocess the CelebA attribute file."""
random.seed(1234)
random.shuffle(self.test_images)
for i, img in enumerate(self.test_images):
style_idx = int(img.split('_')[0][len(self.image_dir+'test/'):])
char_idx = int(img.split('_')[1][:-len(".png")])
target = random.choice([x for x in self.test_images
if str(style_idx)+'_' in x and '_'+str(char_idx) not in x])
char_trg_idx = int(target.split('_')[1][:-len(".png")])
self.test_dataset.append([img, style_idx, char_idx, target, char_trg_idx])
def __getitem__(self, index):
"""Return one image and its corresponding attribute label."""
if self.mode == 'train':
random.seed()
random.shuffle(self.train_images)
src = self.train_images[index]
src_style = int(src.split('_')[0][len(self.image_dir+'train/'):])
src_char = int(src.split('_')[1][:-len(".png")])
try:
trg = random.choice([x for x in self.train_images
if str(src_style)+'_' in x and '_'+str(src_char) not in x])
except:
trg = src
trg_char = int(trg.split('_')[1][:-len(".png")])
else:
src, src_style, src_char, trg, trg_char = self.test_dataset[index]
src = self.transform(Image.open(src))
trg = self.transform(Image.open(trg))
return src, src_style, src_char, \
trg, trg_char
def __len__(self):
"""Return the number of images."""
return self.num_images
def get_loader(image_dir, image_size=128,
batch_size=16, mode='train', num_workers=1):
"""Build and return a data loader."""
transform = []
transform.append(T.Resize(image_size))
transform.append(T.ToTensor())
#transform.append(T.Normalize(mean=(0.5, 0.5, 0.5), std=(0.5, 0.5, 0.5)))
transform = T.Compose(transform)
dataset = ImageFolder(image_dir, transform, mode)
#dataset = EncoderFolder(image_dir, transform, mode)
data_loader = data.DataLoader(dataset=dataset,
batch_size=batch_size,
shuffle=(mode=='train'),
num_workers=num_workers)
return data_loader