-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathhotel_dataloader.py
219 lines (172 loc) · 8.62 KB
/
hotel_dataloader.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
import random
import numpy as np
import torch
from PIL import Image
from torch.utils.data import Dataset
from utils import get_shuffled_data, loadDataToMem
from torchvision.utils import save_image
class HotelTrain(Dataset):
def __init__(self, args, transform=None, mode='train', save_pictures=False):
super(HotelTrain, self).__init__()
np.random.seed(args.seed)
self.transform = transform
self.save_pictures = save_pictures
self.class1 = 0
self.image1 = None
self.no_negative = args.no_negative
self.datas, self.num_classes, self.length, self.labels, _ = loadDataToMem(args.dataset_path, args.dataset_name,
args.dataset_split_type,
mode=mode,
split_file_name=args.splits_file_name,
portion=args.portion)
self.shuffled_data = get_shuffled_data(datas=self.datas, seed=args.seed)
print('hotel train classes: ', self.num_classes)
print('hotel train length: ', self.length)
def __len__(self):
return self.length
def __getitem__(self, index):
# image1 = random.choice(self.dataset.imgs)
label = None
img1 = None
img2 = None
# get image from same class
if index % (self.no_negative + 1) == 0:
label = 1.0
idx1 = random.randint(0, self.num_classes - 1)
self.class1 = self.labels[idx1]
class2 = self.class1
self.image1 = Image.open(random.choice(self.datas[self.class1]))
image2 = Image.open(random.choice(self.datas[class2]))
# get image from different class
else:
label = 0.0
# idx1 = random.randint(0, self.num_classes - 1)
idx2 = random.randint(0, self.num_classes - 1)
class2 = self.labels[idx2]
while self.class1 == class2:
idx2 = random.randint(0, self.num_classes - 1)
class2 = self.labels[idx2]
# class1 = self.labels[idx1]
# image1 = Image.open(random.choice(self.datas[self.class1]))
image2 = Image.open(random.choice(self.datas[class2]))
image1 = self.image1.convert('RGB')
image2 = image2.convert('RGB')
save = False
if self.transform:
if self.save_pictures and random.random() < 0.0001:
save = True
img1_random = random.randint(0, 1000)
img2_random = random.randint(0, 1000)
image1.save(f'hotel_imagesamples/train/train_{self.class1}_{img1_random}_before.png')
image2.save(f'hotel_imagesamples/train/train_{class2}_{img2_random}_before.png')
image2 = self.transform(image2)
image1 = self.transform(image1)
if save:
save_image(image1, f'hotel_imagesamples/train/train_{self.class1}_{img1_random}_after.png')
save_image(image2, f'hotel_imagesamples/train/train_{class2}_{img2_random}_after.png')
return image1, image2, torch.from_numpy(np.array([label], dtype=np.float32))
def _get_single_item(self, index):
label, image_path = self.shuffled_data[index]
image = Image.open(image_path)
image = image.convert('RGB')
if self.transform:
image = self.transform(image)
return image, torch.from_numpy(np.array(label, dtype=np.float32))
def get_k_samples(self, k=100):
ks = np.random.randint(len(self.shuffled_data), size=k)
imgs = []
lbls = []
for i in ks:
img, lbl = self._get_single_item(i)
imgs.append(img)
lbls.append(lbl)
return imgs, lbls
class HotelTest(Dataset):
def __init__(self, args, transform=None, mode='test_seen', save_pictures=False):
np.random.seed(args.seed)
super(HotelTest, self).__init__()
self.transform = transform
self.save_pictures = save_pictures
self.times = args.times
self.way = args.way
self.img1 = None
self.c1 = None
self.datas, self.num_classes, _, self.labels, self.datas_bg = loadDataToMem(args.dataset_path,
args.dataset_name,
args.dataset_split_type, mode=mode,
split_file_name=args.splits_file_name,
portion=args.portion)
print(f'hotel {mode} classes: ', self.num_classes)
print(f'hotel {mode} length: ', self.__len__())
def __len__(self):
return self.times * self.way
def __getitem__(self, index):
idx = index % self.way
label = None
# generate image pair from same class
if idx == 0:
self.c1 = self.labels[random.randint(0, self.num_classes - 1)]
c2 = self.c1
self.img1 = Image.open(random.choice(self.datas[self.c1])).convert('RGB')
img2 = Image.open(random.choice(self.datas[c2])).convert('RGB')
# generate image pair from different class
else:
c2 = list(self.datas_bg.keys())[random.randint(0, len(self.datas_bg.keys()) - 1)]
while self.c1 == c2:
c2 = list(self.datas_bg.keys())[random.randint(0, len(self.datas_bg.keys()) - 1)]
img2 = Image.open(random.choice(self.datas_bg[c2])[0]).convert('RGB')
save = False
if self.transform:
if self.save_pictures and random.random() < 0.001:
save = True
img1_random = random.randint(0, 1000)
img2_random = random.randint(0, 1000)
self.img1.save(f'hotel_imagesamples/val/val_{self.c1}_{img1_random}_before.png')
img2.save(f'hotel_imagesamples/val/val_{c2}_{img2_random}_before.png')
img1 = self.transform(self.img1)
img2 = self.transform(img2)
if save:
save_image(img1, f'hotel_imagesamples/val/val_{self.c1}_{img1_random}_after.png')
save_image(img2, f'hotel_imagesamples/val/val_{c2}_{img2_random}_after.png')
return img1, img2
class Hotel_DB(Dataset):
def __init__(self, args, transform=None, mode='test'):
np.random.seed(args.seed)
super(Hotel_DB, self).__init__()
self.transform = transform
total = True
if 'seen' in mode: # mode == *_seen or *_unseen
mode_tmp = mode
total = False
else:
mode_tmp = mode + '_seen'
total = True
self.datas, self.num_classes, _, self.labels, self.datas_bg = loadDataToMem(args.dataset_path,
args.dataset_name,
args.dataset_split_type,
mode=mode_tmp,
split_file_name=args.splits_file_name,
portion=args.portion)
# if total:
self.all_shuffled_data = get_shuffled_data(self.datas_bg,
seed=args.seed,
one_hot=False,
both_seen_unseen=True,
shuffle=False)
# else: # todo
# self.all_shuffled_data = get_shuffled_data(self.datas, seed=args.seed, one_hot=False)
print(f'hotel {mode} classes: ', self.num_classes)
print(f'hotel {mode} length: ', self.__len__())
def __len__(self):
return len(self.all_shuffled_data)
def __getitem__(self, index):
lbl = self.all_shuffled_data[index][0]
img = Image.open(self.all_shuffled_data[index][1]).convert('RGB')
bl = self.all_shuffled_data[index][2]
path = self.all_shuffled_data[index][1].split('/')
id = path[-4]
id += '-' + path[-3]
id += '-' + path[-1].split('.')[0]
if self.transform:
img = self.transform(img)
return img, lbl, bl, id