red-datasets 0.0.3 → 0.0.4
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- checksums.yaml +4 -4
- data/README.md +38 -0
- data/doc/text/news.md +14 -0
- data/lib/datasets/cifar.rb +126 -0
- data/lib/datasets/dataset.rb +2 -0
- data/lib/datasets/version.rb +1 -1
- data/lib/datasets.rb +1 -0
- data/test/test-cifar.rb +209 -0
- metadata +5 -2
checksums.yaml
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
---
|
|
2
2
|
SHA256:
|
|
3
|
-
metadata.gz:
|
|
4
|
-
data.tar.gz:
|
|
3
|
+
metadata.gz: 56c3132aec99f8dab763d68d0b731df0e5b2c97ea9f88a847021210fda94b9ac
|
|
4
|
+
data.tar.gz: 20e11d381718b801dac9c155385837b770c5067646ac235ade9c0f06f4d23cb8
|
|
5
5
|
SHA512:
|
|
6
|
-
metadata.gz:
|
|
7
|
-
data.tar.gz:
|
|
6
|
+
metadata.gz: 1c3ccf7b09e6064cacc47c04dd972ef4ba9318278630446ae60911412e9dc5dced5c988423b78a1c0ae68d71ede6136ebd2ee9c86f4247fd56bbf940f7997f7e
|
|
7
|
+
data.tar.gz: bebe49f31022dfeab1eb7626ef1eaf722d5dedf373dab89ba79e3e2afb4849ace50b5098688e4d0480f22719a758a086c288c600cdcb358099a7aa9d1e60eab1
|
data/README.md
CHANGED
|
@@ -37,6 +37,44 @@ iris.each do |record|
|
|
|
37
37
|
end
|
|
38
38
|
```
|
|
39
39
|
|
|
40
|
+
Here is an example to access CIFAR dataset by `#each`:
|
|
41
|
+
|
|
42
|
+
**CIFAR-10**
|
|
43
|
+
|
|
44
|
+
```ruby
|
|
45
|
+
require "datasets"
|
|
46
|
+
|
|
47
|
+
cifar = Datasets::CIFAR.new(n_classes: 10, type: :train)
|
|
48
|
+
cifar.metadata
|
|
49
|
+
#=> #<struct Datasets::Metadata name="CIFAR-10", url="https://www.cs.toronto.edu/~kriz/cifar.html", licenses=nil, description="CIFAR-10 is 32x32 image dataset">licenses=nil, description="CIFAR-10 is 32x32 image datasets">
|
|
50
|
+
cifar.each do |record|
|
|
51
|
+
p record.pixels
|
|
52
|
+
# => [59, 43, 50, 68, 98, 119, 139, 145, 149, 143, .....]
|
|
53
|
+
p record.label
|
|
54
|
+
# => 6
|
|
55
|
+
end
|
|
56
|
+
end
|
|
57
|
+
```
|
|
58
|
+
|
|
59
|
+
**CIFAR-100**
|
|
60
|
+
|
|
61
|
+
```ruby
|
|
62
|
+
require "datasets"
|
|
63
|
+
|
|
64
|
+
cifar = Datasets::CIFAR.new(n_classes: 100, type: :test)
|
|
65
|
+
cifar.metadata
|
|
66
|
+
#=> #<struct Datasets::Metadata name="CIFAR-100", url="https://www.cs.toronto.edu/~kriz/cifar.html", licenses=nil, description="CIFAR-100 is 32x32 image dataset">
|
|
67
|
+
cifar.each do |record|
|
|
68
|
+
p record.pixels
|
|
69
|
+
#=> [199, 196, 195, 195, 196, 197, 198, 198, 199, .....]
|
|
70
|
+
p record.coarse_label
|
|
71
|
+
#=> 10
|
|
72
|
+
p record.fine_label
|
|
73
|
+
#=> 49
|
|
74
|
+
end
|
|
75
|
+
end
|
|
76
|
+
```
|
|
77
|
+
|
|
40
78
|
## License
|
|
41
79
|
|
|
42
80
|
The MIT license. See `LICENSE.txt` for details.
|
data/doc/text/news.md
CHANGED
|
@@ -1,5 +1,19 @@
|
|
|
1
1
|
# News
|
|
2
2
|
|
|
3
|
+
## 0.0.4 - 2018-05-03
|
|
4
|
+
|
|
5
|
+
### Improvements
|
|
6
|
+
|
|
7
|
+
* `Datasets::Dataset`: Made enumerable.
|
|
8
|
+
|
|
9
|
+
* `Datasets::CIFAR`: Added the CIFAR dataset.
|
|
10
|
+
[GitHub#7][GitHub#8][GitHub#9][GitHub#10]
|
|
11
|
+
[Patch by Yusaku Hatanaka]
|
|
12
|
+
|
|
13
|
+
### Thanks
|
|
14
|
+
|
|
15
|
+
* Yusaku Hatanaka
|
|
16
|
+
|
|
3
17
|
## 0.0.3 - 2018-03-27
|
|
4
18
|
|
|
5
19
|
### Improvements
|
|
@@ -0,0 +1,126 @@
|
|
|
1
|
+
require "rubygems/package"
|
|
2
|
+
require "zlib"
|
|
3
|
+
|
|
4
|
+
require_relative "dataset"
|
|
5
|
+
|
|
6
|
+
module Datasets
|
|
7
|
+
class CIFAR < Dataset
|
|
8
|
+
class Record10 < Struct.new(:data, :label)
|
|
9
|
+
def pixels
|
|
10
|
+
data.unpack("C*")
|
|
11
|
+
end
|
|
12
|
+
end
|
|
13
|
+
|
|
14
|
+
class Record100 < Struct.new(:data, :coarse_label, :fine_label)
|
|
15
|
+
def pixels
|
|
16
|
+
data.unpack("C*")
|
|
17
|
+
end
|
|
18
|
+
end
|
|
19
|
+
|
|
20
|
+
def initialize(n_classes: 10, type: :train)
|
|
21
|
+
unless [10, 100].include?(n_classes)
|
|
22
|
+
raise 'Please set n_classes 10 or 100'
|
|
23
|
+
end
|
|
24
|
+
unless [:train, :test].include?(type)
|
|
25
|
+
raise 'Please set type :train or :test'
|
|
26
|
+
end
|
|
27
|
+
|
|
28
|
+
super()
|
|
29
|
+
|
|
30
|
+
@metadata.name = "CIFAR-#{n_classes}"
|
|
31
|
+
@metadata.url = "https://www.cs.toronto.edu/~kriz/cifar.html"
|
|
32
|
+
@metadata.description = "CIFAR-#{n_classes} is 32x32 image dataset"
|
|
33
|
+
|
|
34
|
+
@n_classes = n_classes
|
|
35
|
+
@type = type
|
|
36
|
+
end
|
|
37
|
+
|
|
38
|
+
def each(&block)
|
|
39
|
+
return to_enum(__method__) unless block_given?
|
|
40
|
+
|
|
41
|
+
data_path = cache_dir_path + "cifar-#{@n_classes}.tar.gz"
|
|
42
|
+
unless data_path.exist?
|
|
43
|
+
data_url = "https://www.cs.toronto.edu/~kriz/cifar-#{@n_classes}-binary.tar.gz"
|
|
44
|
+
download(data_path, data_url)
|
|
45
|
+
end
|
|
46
|
+
|
|
47
|
+
parse_data(data_path, &block)
|
|
48
|
+
end
|
|
49
|
+
|
|
50
|
+
private
|
|
51
|
+
|
|
52
|
+
def parse_data(data_path, &block)
|
|
53
|
+
open_tar(data_path) do |tar|
|
|
54
|
+
target_file_names.each do |target_file_name|
|
|
55
|
+
tar.seek(target_file_name) do |entry|
|
|
56
|
+
parse_entry(entry, &block)
|
|
57
|
+
end
|
|
58
|
+
end
|
|
59
|
+
end
|
|
60
|
+
end
|
|
61
|
+
|
|
62
|
+
def target_file_names
|
|
63
|
+
case @n_classes
|
|
64
|
+
when 10
|
|
65
|
+
prefix = 'cifar-10-batches-bin'
|
|
66
|
+
case @type
|
|
67
|
+
when :train
|
|
68
|
+
[
|
|
69
|
+
"#{prefix}/data_batch_1.bin",
|
|
70
|
+
"#{prefix}/data_batch_2.bin",
|
|
71
|
+
"#{prefix}/data_batch_3.bin",
|
|
72
|
+
"#{prefix}/data_batch_4.bin",
|
|
73
|
+
"#{prefix}/data_batch_5.bin",
|
|
74
|
+
]
|
|
75
|
+
when :test
|
|
76
|
+
[
|
|
77
|
+
"#{prefix}/test_batch.bin"
|
|
78
|
+
]
|
|
79
|
+
end
|
|
80
|
+
when 100
|
|
81
|
+
prefix = "cifar-100-binary"
|
|
82
|
+
case @type
|
|
83
|
+
when :train
|
|
84
|
+
[
|
|
85
|
+
"#{prefix}/train.bin",
|
|
86
|
+
]
|
|
87
|
+
when :test
|
|
88
|
+
[
|
|
89
|
+
"#{prefix}/test.bin",
|
|
90
|
+
]
|
|
91
|
+
end
|
|
92
|
+
end
|
|
93
|
+
end
|
|
94
|
+
|
|
95
|
+
def parse_entry(entry)
|
|
96
|
+
case @n_classes
|
|
97
|
+
when 10
|
|
98
|
+
loop do
|
|
99
|
+
label = entry.read(1)
|
|
100
|
+
break if label.nil?
|
|
101
|
+
label = label.unpack("C")[0]
|
|
102
|
+
data = entry.read(3072)
|
|
103
|
+
yield Record10.new(data, label)
|
|
104
|
+
end
|
|
105
|
+
when 100
|
|
106
|
+
loop do
|
|
107
|
+
coarse_label = entry.read(1)
|
|
108
|
+
break if coarse_label.nil?
|
|
109
|
+
coarse_label = coarse_label.unpack("C")[0]
|
|
110
|
+
fine_label = entry.read(1).unpack("C")[0]
|
|
111
|
+
data = entry.read(3072)
|
|
112
|
+
yield Record100.new(data, coarse_label, fine_label)
|
|
113
|
+
end
|
|
114
|
+
end
|
|
115
|
+
end
|
|
116
|
+
|
|
117
|
+
def open_tar(data_path)
|
|
118
|
+
Zlib::GzipReader.open(data_path) do |f|
|
|
119
|
+
Gem::Package::TarReader.new(f) do |tar|
|
|
120
|
+
yield(tar)
|
|
121
|
+
end
|
|
122
|
+
end
|
|
123
|
+
end
|
|
124
|
+
end
|
|
125
|
+
end
|
|
126
|
+
|
data/lib/datasets/dataset.rb
CHANGED
data/lib/datasets/version.rb
CHANGED
data/lib/datasets.rb
CHANGED
data/test/test-cifar.rb
ADDED
|
@@ -0,0 +1,209 @@
|
|
|
1
|
+
class CIFARTest < Test::Unit::TestCase
|
|
2
|
+
include Helper::Sandbox
|
|
3
|
+
|
|
4
|
+
def setup_raw_data(data)
|
|
5
|
+
setup_sandbox
|
|
6
|
+
|
|
7
|
+
def @dataset.cache_dir_path
|
|
8
|
+
@cache_dir_path
|
|
9
|
+
end
|
|
10
|
+
def @dataset.cache_dir_path=(path)
|
|
11
|
+
@cache_dir_path = path
|
|
12
|
+
end
|
|
13
|
+
@dataset.cache_dir_path = @tmp_dir
|
|
14
|
+
|
|
15
|
+
def @dataset.data=(data)
|
|
16
|
+
@data = data
|
|
17
|
+
end
|
|
18
|
+
@dataset.data = data
|
|
19
|
+
|
|
20
|
+
def @dataset.download(output_path, url)
|
|
21
|
+
Zlib::GzipWriter.open(output_path) do |gz|
|
|
22
|
+
Gem::Package::TarWriter.new(gz) do |tar|
|
|
23
|
+
@data.each do |path, content|
|
|
24
|
+
if content == :directory
|
|
25
|
+
tar.mkdir(path, 0755)
|
|
26
|
+
else
|
|
27
|
+
tar.add_file_simple(path, 0644, content.bytesize) do |file|
|
|
28
|
+
file.write(content)
|
|
29
|
+
end
|
|
30
|
+
end
|
|
31
|
+
end
|
|
32
|
+
end
|
|
33
|
+
end
|
|
34
|
+
end
|
|
35
|
+
end
|
|
36
|
+
|
|
37
|
+
def teardown
|
|
38
|
+
teardown_sandbox
|
|
39
|
+
end
|
|
40
|
+
|
|
41
|
+
sub_test_case("cifar-10") do
|
|
42
|
+
def create_data(label, pixel)
|
|
43
|
+
[label].pack("C") + ([pixel] * 3072).pack("C*")
|
|
44
|
+
end
|
|
45
|
+
|
|
46
|
+
sub_test_case("train") do
|
|
47
|
+
def setup
|
|
48
|
+
@dataset = Datasets::CIFAR.new(n_classes: 10, type: :train)
|
|
49
|
+
directory = "cifar-10-batches-bin"
|
|
50
|
+
setup_raw_data(directory => :directory,
|
|
51
|
+
"#{directory}/data_batch_1.bin" => create_data(1, 10),
|
|
52
|
+
"#{directory}/data_batch_2.bin" => create_data(2, 20),
|
|
53
|
+
"#{directory}/data_batch_3.bin" => create_data(3, 30),
|
|
54
|
+
"#{directory}/data_batch_4.bin" => create_data(4, 40),
|
|
55
|
+
"#{directory}/data_batch_5.bin" => create_data(5, 50),
|
|
56
|
+
"#{directory}/data_batch_6.bin" => create_data(6, 60))
|
|
57
|
+
end
|
|
58
|
+
|
|
59
|
+
test("#each") do
|
|
60
|
+
raw_dataset = @dataset.collect do |record|
|
|
61
|
+
{
|
|
62
|
+
:label => record.label,
|
|
63
|
+
:data => record.data,
|
|
64
|
+
:pixels => record.pixels
|
|
65
|
+
}
|
|
66
|
+
end
|
|
67
|
+
assert_equal([
|
|
68
|
+
{
|
|
69
|
+
:label => 1,
|
|
70
|
+
:data => ([10] * 3072).pack("C*"),
|
|
71
|
+
:pixels => [10] * 3072
|
|
72
|
+
},
|
|
73
|
+
{
|
|
74
|
+
:label => 2,
|
|
75
|
+
:data => ([20] * 3072).pack("C*"),
|
|
76
|
+
:pixels => [20] * 3072
|
|
77
|
+
},
|
|
78
|
+
{
|
|
79
|
+
:label => 3,
|
|
80
|
+
:data => ([30] * 3072).pack("C*"),
|
|
81
|
+
:pixels => [30] * 3072
|
|
82
|
+
},
|
|
83
|
+
{
|
|
84
|
+
:label => 4,
|
|
85
|
+
:data => ([40] * 3072).pack("C*"),
|
|
86
|
+
:pixels => [40] * 3072
|
|
87
|
+
},
|
|
88
|
+
{
|
|
89
|
+
:label => 5,
|
|
90
|
+
:data => ([50] * 3072).pack("C*"),
|
|
91
|
+
:pixels => [50] * 3072
|
|
92
|
+
},
|
|
93
|
+
],
|
|
94
|
+
raw_dataset)
|
|
95
|
+
end
|
|
96
|
+
end
|
|
97
|
+
|
|
98
|
+
sub_test_case("test") do
|
|
99
|
+
def setup
|
|
100
|
+
@dataset = Datasets::CIFAR.new(n_classes: 10, type: :test)
|
|
101
|
+
directory = "cifar-10-batches-bin"
|
|
102
|
+
data = create_data(1, 100) + create_data(2, 200)
|
|
103
|
+
setup_raw_data(directory => :directory,
|
|
104
|
+
"#{directory}/test_batch.bin" => data)
|
|
105
|
+
end
|
|
106
|
+
|
|
107
|
+
test("#each") do
|
|
108
|
+
raw_dataset = @dataset.collect do |record|
|
|
109
|
+
{
|
|
110
|
+
:label => record.label,
|
|
111
|
+
:data => record.data,
|
|
112
|
+
:pixels => record.pixels,
|
|
113
|
+
}
|
|
114
|
+
end
|
|
115
|
+
assert_equal([
|
|
116
|
+
{
|
|
117
|
+
:label => 1,
|
|
118
|
+
:data => ([100] * 3072).pack("C*"),
|
|
119
|
+
:pixels => [100] * 3072
|
|
120
|
+
},
|
|
121
|
+
{
|
|
122
|
+
:label => 2,
|
|
123
|
+
:data => ([200] * 3072).pack("C*"),
|
|
124
|
+
:pixels => [200] * 3072
|
|
125
|
+
},
|
|
126
|
+
],
|
|
127
|
+
raw_dataset)
|
|
128
|
+
end
|
|
129
|
+
end
|
|
130
|
+
end
|
|
131
|
+
|
|
132
|
+
sub_test_case("cifar-100") do
|
|
133
|
+
def create_data(coarse_label, fine_label, pixel)
|
|
134
|
+
[coarse_label, fine_label].pack("C*") + ([pixel] * 3072).pack("C*")
|
|
135
|
+
end
|
|
136
|
+
|
|
137
|
+
sub_test_case("train") do
|
|
138
|
+
def setup
|
|
139
|
+
@dataset = Datasets::CIFAR.new(n_classes: 100, type: :train)
|
|
140
|
+
directory = "cifar-100-binary"
|
|
141
|
+
data = create_data(1, 11, 10) + create_data(2, 22, 20)
|
|
142
|
+
setup_raw_data(directory => :directory,
|
|
143
|
+
"#{directory}/train.bin" => data)
|
|
144
|
+
end
|
|
145
|
+
|
|
146
|
+
test("#each") do
|
|
147
|
+
raw_dataset = @dataset.collect do |record|
|
|
148
|
+
{
|
|
149
|
+
:coarse_label => record.coarse_label,
|
|
150
|
+
:fine_label => record.fine_label,
|
|
151
|
+
:data => record.data,
|
|
152
|
+
:pixels => record.pixels,
|
|
153
|
+
}
|
|
154
|
+
end
|
|
155
|
+
assert_equal([
|
|
156
|
+
{
|
|
157
|
+
:coarse_label => 1,
|
|
158
|
+
:fine_label => 11,
|
|
159
|
+
:data => ([10] * 3072).pack("C*"),
|
|
160
|
+
:pixels => [10] * 3072
|
|
161
|
+
},
|
|
162
|
+
{
|
|
163
|
+
:coarse_label => 2,
|
|
164
|
+
:fine_label => 22,
|
|
165
|
+
:data => ([20] * 3072).pack("C*"),
|
|
166
|
+
:pixels => [20] * 3072
|
|
167
|
+
},
|
|
168
|
+
],
|
|
169
|
+
raw_dataset)
|
|
170
|
+
end
|
|
171
|
+
end
|
|
172
|
+
|
|
173
|
+
sub_test_case("test") do
|
|
174
|
+
def setup
|
|
175
|
+
@dataset = Datasets::CIFAR.new(n_classes: 100, type: :test)
|
|
176
|
+
directory = "cifar-100-binary"
|
|
177
|
+
data = create_data(1, 11, 100) + create_data(6, 66, 200)
|
|
178
|
+
setup_raw_data(directory => :directory,
|
|
179
|
+
"#{directory}/test.bin" => data)
|
|
180
|
+
end
|
|
181
|
+
|
|
182
|
+
test("#each") do
|
|
183
|
+
raw_dataset = @dataset.collect do |record|
|
|
184
|
+
{
|
|
185
|
+
:coarse_label => record.coarse_label,
|
|
186
|
+
:fine_label => record.fine_label,
|
|
187
|
+
:data => record.data,
|
|
188
|
+
:pixels => record.pixels,
|
|
189
|
+
}
|
|
190
|
+
end
|
|
191
|
+
assert_equal([
|
|
192
|
+
{
|
|
193
|
+
:coarse_label => 1,
|
|
194
|
+
:fine_label => 11,
|
|
195
|
+
:data => ([100] * 3072).pack("C*"),
|
|
196
|
+
:pixels => [100] * 3072
|
|
197
|
+
},
|
|
198
|
+
{
|
|
199
|
+
:coarse_label => 6,
|
|
200
|
+
:fine_label => 66,
|
|
201
|
+
:data => ([200] * 3072).pack("C*"),
|
|
202
|
+
:pixels => [200] * 3072
|
|
203
|
+
},
|
|
204
|
+
],
|
|
205
|
+
raw_dataset)
|
|
206
|
+
end
|
|
207
|
+
end
|
|
208
|
+
end
|
|
209
|
+
end
|
metadata
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
|
2
2
|
name: red-datasets
|
|
3
3
|
version: !ruby/object:Gem::Version
|
|
4
|
-
version: 0.0.
|
|
4
|
+
version: 0.0.4
|
|
5
5
|
platform: ruby
|
|
6
6
|
authors:
|
|
7
7
|
- tomisuker
|
|
@@ -9,7 +9,7 @@ authors:
|
|
|
9
9
|
autorequire:
|
|
10
10
|
bindir: bin
|
|
11
11
|
cert_chain: []
|
|
12
|
-
date: 2018-03
|
|
12
|
+
date: 2018-05-03 00:00:00.000000000 Z
|
|
13
13
|
dependencies:
|
|
14
14
|
- !ruby/object:Gem::Dependency
|
|
15
15
|
name: bundler
|
|
@@ -99,6 +99,7 @@ files:
|
|
|
99
99
|
- Rakefile
|
|
100
100
|
- doc/text/news.md
|
|
101
101
|
- lib/datasets.rb
|
|
102
|
+
- lib/datasets/cifar.rb
|
|
102
103
|
- lib/datasets/dataset.rb
|
|
103
104
|
- lib/datasets/downloader.rb
|
|
104
105
|
- lib/datasets/iris.rb
|
|
@@ -109,6 +110,7 @@ files:
|
|
|
109
110
|
- red-datasets.gemspec
|
|
110
111
|
- test/helper.rb
|
|
111
112
|
- test/run-test.rb
|
|
113
|
+
- test/test-cifar.rb
|
|
112
114
|
- test/test-iris.rb
|
|
113
115
|
- test/test-table.rb
|
|
114
116
|
- test/test-wikipedia.rb
|
|
@@ -142,3 +144,4 @@ test_files:
|
|
|
142
144
|
- test/helper.rb
|
|
143
145
|
- test/run-test.rb
|
|
144
146
|
- test/test-table.rb
|
|
147
|
+
- test/test-cifar.rb
|