-
Notifications
You must be signed in to change notification settings - Fork 1
/
Data.py
176 lines (153 loc) · 9.3 KB
/
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
import torch
import h5py
from PIL import Image,ImageFile
ImageFile.LOAD_TRUNCATED_IMAGES = True
from torch.utils import data
from torch.utils.data import DataLoader
from torchvision import datasets, transforms
class Data(object):
def __init__(self, args,logger):
self.args = args
iteration = args.iteration
client = None
if args.dataset == 'pacs':
self.trainset, self.testset = None, None
if iteration == 0:
client = ['cartoon', 'sketch', 'art_painting','photo']
self.train_loader, self.test_loader,self.target_loader = get_pacs_loaders(args,client)
if iteration == 1:
client=[ 'photo','cartoon','sketch','art_painting']
self.train_loader, self.test_loader,self.target_loader = get_pacs_loaders(args,client)
if iteration == 2:
client=[ 'sketch','photo','art_painting','cartoon']
self.train_loader, self.test_loader,self.target_loader = get_pacs_loaders(args,client)
if iteration == 3:
client=[ 'photo','cartoon','art_painting','sketch']
self.train_loader, self.test_loader,self.target_loader = get_pacs_loaders(args,client)
if args.dataset == 'vlcs':
if iteration == 0:
self.train_loader, self.test_loader,self.target_loader = get_vlcs_loaders(args,
client=['SUN09', 'Caltech101','LabelMe','VOC2007'])
if iteration == 1:
self.train_loader, self.test_loader,self.target_loader = get_vlcs_loaders(args,
client=['Caltech101','VOC2007','SUN09', 'LabelMe'])
if iteration == 2:
self.train_loader, self.test_loader,self.target_loader = get_vlcs_loaders(args,
client=['SUN09', 'LabelMe','VOC2007','Caltech101'])
if iteration == 3:
self.train_loader, self.test_loader,self.target_loader = get_vlcs_loaders(args,
client=['LabelMe','Caltech101','VOC2007','SUN09'])
if args.dataset == 'office-home':
if iteration == 0:
self.train_loader, self.test_loader,self.target_loader = get_office_loaders(args,
client=['Real World','Product','Clipart','Art'])
if iteration == 1:
self.train_loader, self.test_loader,self.target_loader = get_office_loaders(args,
client=[ 'Real World','Product','Art','Clipart'])
if iteration == 2:
self.train_loader, self.test_loader,self.target_loader = get_office_loaders(args,
client=['Real World','Art','Clipart', 'Product'])
if iteration == 3:
self.train_loader, self.test_loader,self.target_loader = get_office_loaders(args,
client=['Clipart', 'Art','Product','Real World'])
logger.info('CLIENT_ORDER{}'.format(client))
class Loader_dataset(data.Dataset):
def __init__(self, path, tranforms = None):
self.path = path
self.dataset = datasets.ImageFolder(path, transform=tranforms)
self.length = self.dataset.__len__()
self.transform = tranforms
def __len__(self):
return self.length
def __getitem__(self, idx):
data, label = self.dataset.__getitem__(idx)
return data, label
def get_vlcs_loaders(args, client):
path_root = 'datasets/VLCS/'
trans0 = transforms.Compose([transforms.RandomResizedCrop(225, scale=(0.7, 1.0)),
transforms.RandomHorizontalFlip(),
transforms.RandomGrayscale(),
transforms.ToTensor(),
transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])])
trans1 = transforms.Compose([transforms.Resize([225, 225]),
transforms.ToTensor(),
transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])])
train_path, valid_path = {}, {}
train_datas, train_loaders = {}, {}
valid_datas, valid_loaders = {}, {}
for i in range(3):
train_path[i] = path_root + client[i] + '/train'
train_datas[i] = Loader_dataset(path=train_path[i], tranforms=trans0)
train_loaders[i] = DataLoader(train_datas[i], args.batch_size, True, num_workers=args.workers,pin_memory=args.pin)
valid_path[i] = path_root + client[i] + '/val'
valid_datas[i] = Loader_dataset(path=valid_path[i], tranforms=trans1)
valid_loaders[i] = DataLoader(valid_datas[i], args.batch_size, True, num_workers=args.workers,pin_memory=args.pin)
target_path = path_root + client[3] + '/val'
target_data = Loader_dataset(target_path, trans1)
target_loader = DataLoader(target_data, args.batch_size, True, num_workers=args.workers,pin_memory=args.pin)
print(client,'\n')
return train_loaders, valid_loaders, target_loader
class Loader_dataset_pacs(data.Dataset):
def __init__(self, path, tranforms):
self.path = path
hdf = h5py.File(self.path, 'r')
self.length = len(hdf['labels']) # <KeysViewHDF5 ['images', 'labels']>
self.transform = tranforms
hdf.close()
def __len__(self):
return self.length
def __getitem__(self, idx):
hdf = h5py.File(self.path, 'r')
y = hdf['labels'][idx]
data_pil = Image.fromarray(hdf['images'][idx, :, :, :].astype('uint8'), 'RGB')
hdf.close()
data = self.transform(data_pil)
return data, torch.tensor(y).long().squeeze()-1
def get_pacs_loaders(args, client = ['cartoon', 'sketch', 'art_painting','photo']):
path_root = 'datasets/PACS/'
trans0 = transforms.Compose([transforms.RandomResizedCrop(222, scale=(0.7, 1.0)),
transforms.RandomHorizontalFlip(),
transforms.RandomGrayscale(),
transforms.ToTensor(),
transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])])
trans1 = transforms.Compose([transforms.Resize([222, 222]),
transforms.ToTensor(),
transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])])
train_path, valid_path = {}, {}
train_datas, train_loaders = {}, {}
valid_datas, valid_loaders = {}, {}
for i in range(3):
train_path[i] = path_root + client[i] + '_train.hdf5'
train_datas[i] = Loader_dataset_pacs(path=train_path[i], tranforms=trans0)
train_loaders[i] = DataLoader(train_datas[i], args.batch_size, True, num_workers=args.workers, pin_memory=args.pin)
valid_path[i] = path_root + client[i] + '_val.hdf5'
valid_datas[i] = Loader_dataset_pacs(path=valid_path[i], tranforms=trans1)
valid_loaders[i] = DataLoader(valid_datas[i], args.batch_size, True, num_workers=args.workers, pin_memory=args.pin)
target_path = path_root + client[3] + '_test.hdf5'
target_data = Loader_dataset_pacs(target_path, trans1)
target_loader = DataLoader(target_data, args.batch_size, True, num_workers=args.workers, pin_memory=args.pin)
return train_loaders, valid_loaders, target_loader
def get_office_loaders(args, client):
path_root = 'datasets/OfficeHome/'
trans0 = transforms.Compose([transforms.RandomResizedCrop(225, scale=(0.7, 1.0)),
transforms.RandomHorizontalFlip(),
transforms.RandomGrayscale(),
transforms.ToTensor(),
transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])])
trans1 = transforms.Compose([transforms.Resize([225, 225]),
transforms.ToTensor(),
transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])])
train_path, valid_path = {}, {}
train_datas, train_loaders = {}, {}
valid_datas, valid_loaders = {}, {}
for i in range(3):
train_path[i] = path_root + client[i] + '/train'
train_datas[i] = Loader_dataset(path=train_path[i], tranforms=trans0)
train_loaders[i] = DataLoader(train_datas[i], args.batch_size, True, num_workers=args.workers,pin_memory=args.pin)
valid_path[i] = path_root + client[i] + '/val'
valid_datas[i] = Loader_dataset(path=valid_path[i], tranforms=trans1)
valid_loaders[i] = DataLoader(valid_datas[i], args.batch_size, True, num_workers=args.workers,pin_memory=args.pin)
target_path = path_root + client[3] + '/val'
target_data = Loader_dataset(target_path, trans1)
target_loader = DataLoader(target_data, args.batch_size, True, num_workers=args.workers,pin_memory=args.pin)
return train_loaders, valid_loaders, target_loader