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
|
"""Modified Olivetti faces dataset.
The original database was available from (now defunct)
https://www.cl.cam.ac.uk/research/dtg/attarchive/facedatabase.html
The version retrieved here comes in MATLAB format from the personal
web page of Sam Roweis:
https://cs.nyu.edu/~roweis/
"""
# Copyright (c) 2011 David Warde-Farley <wardefar at iro dot umontreal dot ca>
# License: BSD 3 clause
from os.path import dirname, exists, join
from os import makedirs, remove
import numpy as np
from scipy.io.matlab import loadmat
from .base import get_data_home
from .base import _fetch_remote
from .base import RemoteFileMetadata
from .base import _pkl_filepath
from ..utils import _joblib
from ..utils import check_random_state, Bunch
# The original data can be found at:
# https://cs.nyu.edu/~roweis/data/olivettifaces.mat
FACES = RemoteFileMetadata(
filename='olivettifaces.mat',
url='https://ndownloader.figshare.com/files/5976027',
checksum=('b612fb967f2dc77c9c62d3e1266e0c73'
'd5fca46a4b8906c18e454d41af987794'))
def fetch_olivetti_faces(data_home=None, shuffle=False, random_state=0,
download_if_missing=True):
"""Load the Olivetti faces data-set from AT&T (classification).
Download it if necessary.
================= =====================
Classes 40
Samples total 400
Dimensionality 4096
Features real, between 0 and 1
================= =====================
Read more in the :ref:`User Guide <olivetti_faces_dataset>`.
Parameters
----------
data_home : optional, default: None
Specify another download and cache folder for the datasets. By default
all scikit-learn data is stored in '~/scikit_learn_data' subfolders.
shuffle : boolean, optional
If True the order of the dataset is shuffled to avoid having
images of the same person grouped.
random_state : int, RandomState instance or None (default=0)
Determines random number generation for dataset shuffling. Pass an int
for reproducible output across multiple function calls.
See :term:`Glossary <random_state>`.
download_if_missing : optional, True by default
If False, raise a IOError if the data is not locally available
instead of trying to download the data from the source site.
Returns
-------
An object with the following attributes:
data : numpy array of shape (400, 4096)
Each row corresponds to a ravelled face image of original size
64 x 64 pixels.
images : numpy array of shape (400, 64, 64)
Each row is a face image corresponding to one of the 40 subjects
of the dataset.
target : numpy array of shape (400, )
Labels associated to each face image. Those labels are ranging from
0-39 and correspond to the Subject IDs.
DESCR : string
Description of the modified Olivetti Faces Dataset.
"""
data_home = get_data_home(data_home=data_home)
if not exists(data_home):
makedirs(data_home)
filepath = _pkl_filepath(data_home, 'olivetti.pkz')
if not exists(filepath):
if not download_if_missing:
raise IOError("Data not found and `download_if_missing` is False")
print('downloading Olivetti faces from %s to %s'
% (FACES.url, data_home))
mat_path = _fetch_remote(FACES, dirname=data_home)
mfile = loadmat(file_name=mat_path)
# delete raw .mat data
remove(mat_path)
faces = mfile['faces'].T.copy()
_joblib.dump(faces, filepath, compress=6)
del mfile
else:
faces = _joblib.load(filepath)
# We want floating point data, but float32 is enough (there is only
# one byte of precision in the original uint8s anyway)
faces = np.float32(faces)
faces = faces - faces.min()
faces /= faces.max()
faces = faces.reshape((400, 64, 64)).transpose(0, 2, 1)
# 10 images per class, 400 images total, each class is contiguous.
target = np.array([i // 10 for i in range(400)])
if shuffle:
random_state = check_random_state(random_state)
order = random_state.permutation(len(faces))
faces = faces[order]
target = target[order]
module_path = dirname(__file__)
with open(join(module_path, 'descr', 'olivetti_faces.rst')) as rst_file:
fdescr = rst_file.read()
return Bunch(data=faces.reshape(len(faces), -1),
images=faces,
target=target,
DESCR=fdescr)
|
