"official/vision/tasks/image_classification.py" did not exist on "c52a287fdb81d7548b71a33944e7471f9352e748"
lsun.py 4.39 KB
Newer Older
soumith's avatar
soumith committed
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
import torch.utils.data as data
from PIL import Image
import os
import os.path
import StringIO
import string
import sys
if sys.version_info[0] == 2:
    import cPickle as pickle
else:
    import pickle

class LSUNClassDataset(data.Dataset):
    def __init__(self, db_path, transform=None, target_transform=None):
        import lmdb
        self.db_path = db_path
        self.env = lmdb.open(db_path, map_size=1099511627776,
                        max_readers=100, readonly=True)
        with self.env.begin(write=False) as txn:
            self.length = txn.stat()['entries']
        cache_file = '_cache_' + db_path.replace('/', '_')
        if os.path.isfile(cache_file):
            self.keys = pickle.load( open( cache_file, "rb" ) )
        else:
            with self.env.begin(write=False) as txn:
                self.keys = [ key for key, _ in txn.cursor() ]
            pickle.dump( self.keys, open( cache_file, "wb" ) )
        self.transform = transform
        self.target_transform = target_transform

    def __getitem__(self, index):
        img, target = None, None
        env = self.env
        with env.begin(write=False) as txn:
            imgbuf = txn.get(self.keys[index])

        buf = StringIO.StringIO()
        buf.write(imgbuf)
        buf.seek(0)
        img = Image.open(buf).convert('RGB')

        if self.transform is not None:
            img = self.transform(img)

        if self.target_transform is not None:
            target = self.target_transform(target)

        return img, target

    def __len__(self):
        return self.length

    def __repr__(self):
        return self.__class__.__name__ + ' (' + self.db_path + ')'

class LSUNDataset(data.Dataset):
    """
    db_path = root directory for the database files
    classes = 'train' | 'val' | 'test' | ['bedroom_train', 'church_train', ...]
    """
    def __init__(self, db_path, classes='train',
                 transform=None, target_transform=None):
        categories = ['bedroom', 'bridge', 'church_outdoor', 'classroom',
                      'conference_room', 'dining_room', 'kitchen',
                      'living_room', 'restaurant', 'tower']
        dset_opts = ['train', 'val', 'test']
        self.db_path = db_path
        if type(classes) == str and classes in dset_opts:
            classes = [c + '_' + classes for c in categories]
        if type(classes) == list:
            for c in classes:
                c_short = c.split('_')
                c_short.pop(len(c_short) - 1)
                c_short = string.join(c_short, '_')
                if c_short not in categories:
                    raise(ValueError('Unknown LSUN class: ' + c_short + '.'\
                          'Options are: ' + str(categories)))
                c_short = c.split('_')
                c_short = c_short.pop(len(c_short) - 1)
                if c_short not in dset_opts:
                    raise(ValueError('Unknown postfix: ' + c_short + '.'\
                          'Options are: ' + str(dset_opts)))
        else:
            raise(ValueError('Unknown option for classes'))
        self.classes = classes

        # for each class, create an LSUNClassDataset
        self.dbs = []
        for c in self.classes:
            self.dbs.append(LSUNClassDataset(
                db_path = db_path + '/' + c + '_lmdb',
                transform = transform))

        self.indices = []
        count = 0
        for db in self.dbs:
            count += len(db)
            self.indices.append(count)

        self.length = count
        self.target_transform = target_transform

    def __getitem__(self, index):
        target = 0
        sub = 0
        for ind in self.indices:
            if index < ind:
                break
            target += 1
            sub += ind

        db = self.dbs[target]
        index = index - sub

        if self.target_transform is not None:
            target = self.target_transform(target)

        return db[index], target

    def __len__(self):
        return self.length

    def __repr__(self):
        return self.__class__.__name__ + ' (' + self.db_path + ')'

if __name__ == '__main__':
    #lsun = LSUNClassDataset(db_path='/home/soumith/local/lsun/train/bedroom_train_lmdb')
    #a = lsun[0]
    lsun = LSUNDataset(db_path='/home/soumith/local/lsun/train',
                       classes=['bedroom_train', 'church_outdoor_train'])
    print(lsun.classes)
    print(lsun.dbs)
    a, t = lsun[len(lsun)-1]
    print(a)
    print(t)