-
Notifications
You must be signed in to change notification settings - Fork 0
/
dataset.py
146 lines (119 loc) · 5.33 KB
/
dataset.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
# Copyright 2018 The TensorFlow Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""tf.data.Dataset interface to the MNIST dataset."""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import gzip
import os
import shutil
import tempfile
import numpy as np
from six.moves import urllib
import tensorflow as tf
def read32(bytestream):
"""Read 4 bytes from bytestream as an unsigned 32-bit integer."""
dt = np.dtype(np.uint32).newbyteorder('>')
return np.frombuffer(bytestream.read(4), dtype=dt)[0]
def check_image_file_header(filename):
"""Validate that filename corresponds to images for the MNIST dataset."""
with tf.gfile.Open(filename, 'rb') as f:
magic = read32(f)
read32(f) # num_images, unused
rows = read32(f)
cols = read32(f)
if magic != 2051:
raise ValueError('Invalid magic number %d in MNIST file %s' % (magic,
f.name))
if rows != 28 or cols != 28:
raise ValueError(
'Invalid MNIST file %s: Expected 28x28 images, found %dx%d' %
(f.name, rows, cols))
def check_labels_file_header(filename):
"""Validate that filename corresponds to labels for the MNIST dataset."""
with tf.gfile.Open(filename, 'rb') as f:
magic = read32(f)
read32(f) # num_items, unused
if magic != 2049:
raise ValueError('Invalid magic number %d in MNIST file %s' % (magic,
f.name))
def download(directory, filename):
"""Download (and unzip) a file from the MNIST dataset if not already done."""
filepath = os.path.join(directory, filename)
if tf.gfile.Exists(filepath):
return filepath
if not tf.gfile.Exists(directory):
tf.gfile.MakeDirs(directory)
# CVDF mirror of http://yann.lecun.com/exdb/mnist/
url = 'https://storage.googleapis.com/cvdf-datasets/mnist/' + filename + '.gz'
_, zipped_filepath = tempfile.mkstemp(suffix='.gz')
print('Downloading %s to %s' % (url, zipped_filepath))
urllib.request.urlretrieve(url, zipped_filepath)
with gzip.open(zipped_filepath, 'rb') as f_in, \
tf.gfile.Open(filepath, 'wb') as f_out:
shutil.copyfileobj(f_in, f_out)
os.remove(zipped_filepath)
return filepath
def dataset(directory, images_file, labels_file, batch_size=512):
"""Download and parse MNIST dataset."""
images_file = download(directory, images_file)
labels_file = download(directory, labels_file)
check_image_file_header(images_file)
check_labels_file_header(labels_file)
def decode_image(image):
# Normalize from [0, 255] to [0.0, 1.0]
image = tf.decode_raw(image, tf.uint8)
image = tf.cast(image, tf.float32)
image = tf.reshape(image, [784])
return image / 255.0
def decode_label(label):
label = tf.decode_raw(label, tf.uint8) # tf.string -> [tf.uint8]
label = tf.reshape(label, []) # label is a scalar
return tf.to_int32(label)
images = tf.data.FixedLengthRecordDataset(
images_file, 28 * 28, header_bytes=16).map(decode_image)
labels = tf.data.FixedLengthRecordDataset(
labels_file, 1, header_bytes=8).map(decode_label)
return tf.data.Dataset.zip((images, labels)).batch(batch_size)
def train(directory):
"""tf.data.Dataset object for MNIST training data."""
return dataset(directory, 'train-images-idx3-ubyte',
'train-labels-idx1-ubyte').apply(
tf.contrib.data.shuffle_and_repeat(200))
def test(directory):
"""tf.data.Dataset object for MNIST test data."""
return dataset(directory, 't10k-images-idx3-ubyte',
't10k-labels-idx1-ubyte', batch_size=10000)
def get_iterators(train_directory, test_directory):
dataset = {'training': None, 'test': None}
iterator = {'training': None, 'test': None}
# Define training and test iterators to feed iterator above.
dataset['training'] = train(train_directory)
iterator['training'] = dataset['training'].make_initializable_iterator()
handle = iterator['training'].string_handle()
tf.add_to_collection('trn_iterator_handles', handle)
tf.add_to_collection('trn_iterator_inits', iterator['training'].initializer)
dataset['test'] = test(test_directory)
iterator['test'] = dataset['test'].make_initializable_iterator()
handle = iterator['test'].string_handle()
tf.add_to_collection('test_iterator_handles', handle)
tf.add_to_collection('test_iterator_inits', iterator['test'].initializer)
# Create feedable iterator.
handle = tf.placeholder(tf.string, shape=[])
tf.add_to_collection('feedable_iterator_handles', handle)
iterator = tf.data.Iterator.from_string_handle(handle,
dataset['training'].output_types,
dataset['training'].output_shapes)
next_elem = iterator.get_next()
return next_elem