-
Notifications
You must be signed in to change notification settings - Fork 2
/
datasets.py
180 lines (119 loc) · 7.25 KB
/
datasets.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
import os
import numpy as np
import torch
from torch.utils.data import DataLoader, Subset, Dataset
from torchvision import transforms
from torchvision.datasets import CIFAR10, CIFAR100, ImageFolder
__all__ = ['cifar10_dataloaders', 'cifar100_dataloaders', 'tiny_imagenet_dataloaders',
'cifar10_dataloaders_eval', 'cifar100_dataloaders_eval', 'adv_image_dataset', 'tiny_imagenet_dataloaders_eval']
def cifar10_dataloaders(batch_size=64, data_dir = 'datasets/cifar10'):
train_transform = transforms.Compose([
transforms.RandomCrop(32, padding=4),
transforms.RandomHorizontalFlip(),
transforms.ToTensor(),
])
test_transform = transforms.Compose([
transforms.ToTensor(),
])
train_set = Subset(CIFAR10(data_dir, train=True, transform=train_transform, download=True), list(range(45000)))
val_set = Subset(CIFAR10(data_dir, train=True, transform=test_transform, download=True), list(range(45000, 50000)))
test_set = CIFAR10(data_dir, train=False, transform=test_transform, download=True)
train_loader = DataLoader(train_set, batch_size=batch_size, shuffle=True, num_workers=2, pin_memory=True)
val_loader = DataLoader(val_set, batch_size=batch_size, shuffle=False, num_workers=2, pin_memory=True)
test_loader = DataLoader(test_set, batch_size=batch_size, shuffle=False, num_workers=2, pin_memory=True)
indices = torch.randperm(len(train_set))[:100]
dataset = torch.utils.data.Subset(train_set, indices)
return train_loader, val_loader, test_loader
def cifar10_test_dataloaders(batch_size=64, data_dir = 'datasets/cifar10'):
test_transform = transforms.Compose([
transforms.ToTensor(),
])
test_set = CIFAR10(data_dir, train=False, transform=test_transform, download=True)
test_loader = DataLoader(test_set, batch_size=batch_size, shuffle=False, num_workers=2, pin_memory=True)
return test_loader
def cifar100_dataloaders(batch_size=64, data_dir = 'datasets/cifar100'):
train_transform = transforms.Compose([
transforms.RandomCrop(32, padding=4),
transforms.RandomHorizontalFlip(),
transforms.RandomRotation(15),
transforms.ToTensor(),
])
test_transform = transforms.Compose([
transforms.ToTensor(),
])
train_set = Subset(CIFAR100(data_dir, train=True, transform=train_transform, download=True), list(range(45000)))
val_set = Subset(CIFAR100(data_dir, train=True, transform=test_transform, download=True), list(range(45000, 50000)))
test_set = CIFAR100(data_dir, train=False, transform=test_transform, download=True)
train_loader = DataLoader(train_set, batch_size=batch_size, shuffle=True, num_workers=2, pin_memory=True)
val_loader = DataLoader(val_set, batch_size=batch_size, shuffle=False, num_workers=2, pin_memory=True)
test_loader = DataLoader(test_set, batch_size=batch_size, shuffle=False, num_workers=2, pin_memory=True)
return train_loader, val_loader, test_loader
def tiny_imagenet_dataloaders(batch_size=64, data_dir = 'datasets/tiny-imagenet-200', permutation_seed=10):
train_transform = transforms.Compose([
transforms.RandomCrop(64, padding=4),
transforms.RandomHorizontalFlip(),
transforms.ToTensor(),
])
test_transform = transforms.Compose([
transforms.ToTensor(),
])
train_path = os.path.join(data_dir, 'training')
val_path = os.path.join(data_dir, 'validation')
np.random.seed(permutation_seed)
split_permutation = list(np.random.permutation(100000))
train_set = Subset(ImageFolder(train_path, transform=train_transform), split_permutation[:90000])
val_set = Subset(ImageFolder(train_path, transform=test_transform), split_permutation[90000:])
test_set = ImageFolder(val_path, transform=test_transform)
train_loader = DataLoader(train_set, batch_size=batch_size, shuffle=True, num_workers=2, pin_memory=True)
val_loader = DataLoader(val_set, batch_size=batch_size, shuffle=False, num_workers=2, pin_memory=True)
test_loader = DataLoader(test_set, batch_size=batch_size, shuffle=False, num_workers=2, pin_memory=True)
return train_loader, val_loader, test_loader
def tiny_imagenet_dataloaders_eval(batch_size=64, data_dir = 'datasets/tiny-imagenet-200', permutation_seed=10):
test_transform = transforms.Compose([
transforms.ToTensor(),
])
train_path = os.path.join(data_dir, 'training')
val_path = os.path.join(data_dir, 'validation')
np.random.seed(permutation_seed)
split_permutation = list(np.random.permutation(100000))
train_set = Subset(ImageFolder(train_path, transform=test_transform), split_permutation[:90000])
val_set = Subset(ImageFolder(train_path, transform=test_transform), split_permutation[90000:])
test_set = ImageFolder(val_path, transform=test_transform)
train_loader = DataLoader(train_set, batch_size=batch_size, shuffle=False, num_workers=2, pin_memory=True)
val_loader = DataLoader(val_set, batch_size=batch_size, shuffle=False, num_workers=2, pin_memory=True)
test_loader = DataLoader(test_set, batch_size=batch_size, shuffle=False, num_workers=2, pin_memory=True)
return train_loader, val_loader, test_loader
def cifar100_dataloaders_eval(batch_size=64, data_dir='datasets/cifar100', data_num=45000):
test_transform = transforms.Compose([
transforms.ToTensor(),
])
train_set = Subset(CIFAR100(data_dir, train=True, transform=test_transform, download=True), list(range(data_num)))
val_set = Subset(CIFAR100(data_dir, train=True, transform=test_transform, download=True), list(range(45000, 50000)))
test_set = CIFAR100(data_dir, train=False, transform=test_transform, download=True)
train_loader = DataLoader(train_set, batch_size=batch_size, shuffle=False, num_workers=2, pin_memory=True)
val_loader = DataLoader(val_set, batch_size=batch_size, shuffle=False, num_workers=2, pin_memory=True)
test_loader = DataLoader(test_set, batch_size=batch_size, shuffle=False, num_workers=2, pin_memory=True)
return train_loader, val_loader, test_loader
def cifar10_dataloaders_eval(batch_size=64, data_dir='datasets/cifar10', data_num=45000):
test_transform = transforms.Compose([
transforms.ToTensor(),
])
train_set = Subset(CIFAR10(data_dir, train=True, transform=test_transform, download=True), list(range(data_num)))
val_set = Subset(CIFAR10(data_dir, train=True, transform=test_transform, download=True), list(range(45000, 50000)))
test_set = CIFAR10(data_dir, train=False, transform=test_transform, download=True)
train_loader = DataLoader(train_set, batch_size=batch_size, shuffle=False, num_workers=2, pin_memory=True)
val_loader = DataLoader(val_set, batch_size=batch_size, shuffle=False, num_workers=2, pin_memory=True)
test_loader = DataLoader(test_set, batch_size=batch_size, shuffle=False, num_workers=2, pin_memory=True)
return train_loader, val_loader, test_loader
class adv_image_dataset(Dataset):
def __init__(self, data):
super(adv_image_dataset, self).__init__()
self.image = data['data']
self.target = data['label']
self.number = self.image.shape[0]
def __len__(self):
return self.number
def __getitem__(self, index):
img = self.image[index]
label = self.target[index]
return img, label