-
Notifications
You must be signed in to change notification settings - Fork 2
/
Copy pathdataset_loaders.py
136 lines (107 loc) · 4.49 KB
/
dataset_loaders.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
import numpy as np
from torch.utils.data import random_split, Subset
from torchvision import transforms, datasets
def train_val_split(trainset, valset):
val_length = int(0.1 * len(trainset))
train_length = len(trainset) - val_length
idx = list(range(len(trainset)))
np.random.shuffle(idx)
train_idx = idx[:train_length]
val_idx = idx[train_length:]
train = Subset(trainset, train_idx)
val = Subset(valset, val_idx)
return train, val
def load_mnist(data_root):
ts = [
transforms.ToTensor(),
# transforms.Normalize((0.1307,), (0.3081,))
]
transform = transforms.Compose(ts)
trainset = datasets.MNIST(root=data_root,
train=True,
download=True,
transform=transform)
testset = datasets.MNIST(root=data_root,
train=False,
download=True,
transform=transform)
val_length = int(0.1 * len(trainset))
train, val = random_split(
trainset,
lengths=[len(trainset) - val_length, val_length])
return train, val, testset
def load_fashion_mnist(data_root):
ts = [
transforms.ToTensor(),
# transforms.Normalize((0.5,), (0.5,))
]
transform = transforms.Compose(ts)
trainset = datasets.FashionMNIST(root=data_root,
train=True,
download=True,
transform=transform)
testset = datasets.FashionMNIST(root=data_root,
train=False,
download=True,
transform=transform)
val_length = int(0.1 * len(trainset))
train, val = random_split(
trainset,
lengths=[len(trainset) - val_length, val_length])
return train, val, testset
def load_CIFAR10(data_root):
# validation = testing
transform_train = transforms.Compose([
transforms.RandomCrop(32, padding=4),
transforms.RandomHorizontalFlip(),
transforms.ToTensor(),
# transforms.Normalize((0.4914, 0.4822, 0.4465),
# (0.2023, 0.1994, 0.2010)),
])
transform_test = transforms.Compose([
transforms.ToTensor(),
# transforms.Normalize((0.4914, 0.4822, 0.4465),
# (0.2023, 0.1994, 0.2010)),
])
trainset = datasets.CIFAR10(root=data_root, train=True, download=True,
transform=transform_train)
valset = datasets.CIFAR10(root=data_root, train=True, download=True,
transform=transform_test)
testset = datasets.CIFAR10(root=data_root, train=False,
download=True, transform=transform_test)
# classes = ('plane', 'car', 'bird', 'cat',
# 'deer', 'dog', 'frog', 'horse', 'ship', 'truck')
# train, val = train_val_split(trainset, valset)
return trainset, testset, testset
def load_CIFAR3(data_root):
# validation = testing
transform_train = transforms.Compose([
transforms.RandomCrop(32, padding=4),
transforms.RandomHorizontalFlip(),
transforms.ToTensor(),
# transforms.Normalize((0.4914, 0.4822, 0.4465),
# (0.2023, 0.1994, 0.2010)),
])
transform_test = transforms.Compose([
transforms.ToTensor(),
# transforms.Normalize((0.4914, 0.4822, 0.4465),
# (0.2023, 0.1994, 0.2010)),
])
trainset = datasets.CIFAR10(root=data_root, train=True, download=True,
transform=transform_train)
valset = datasets.CIFAR10(root=data_root, train=True, download=True,
transform=transform_test)
testset = datasets.CIFAR10(root=data_root, train=False,
download=True, transform=transform_test,
)
reduced_trainset = reduce_to_n_classes(trainset, 3)
reduced_testset = reduce_to_n_classes(testset, 3)
# classes = ('plane', 'car', 'bird', 'cat',
# 'deer', 'dog', 'frog', 'horse', 'ship', 'truck')
# train, val = train_val_split(trainset, valset)
return reduced_trainset, reduced_testset, reduced_testset
def reduce_to_n_classes(dataset, n):
targets = np.array(dataset.targets)
dataset.data = dataset.data[targets < n]
dataset.targets = targets[targets < n]
return dataset