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
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
|
<!doctype html>
<html>
<head>
<title>MegaPixels</title>
<meta charset="utf-8" />
<meta name="author" content="Adam Harvey" />
<meta name="description" content="LFW: Labeled Faces in The Wild" />
<meta name="referrer" content="no-referrer" />
<meta name="viewport" content="width=device-width, initial-scale=1.0, user-scalable=yes" />
<link rel='stylesheet' href='/assets/css/fonts.css' />
<link rel='stylesheet' href='/assets/css/tabulator.css' />
<link rel='stylesheet' href='/assets/css/css.css' />
<link rel='stylesheet' href='/assets/css/applets.css' />
</head>
<body>
<header>
<a class='slogan' href="/">
<div class='logo'></div>
<div class='site_name'>MegaPixels</div>
<span class='sub'>The Darkside of Datasets</span>
</a>
<div class='links'>
<a href="/datasets/">Datasets</a>
<a href="/research/">Research</a>
<a href="/about/">About</a>
</div>
</header>
<div class="content">
<section><h1>Labeled Faces in the Wild</h1>
</section><section><div class='meta'><div><div class='gray'>Created</div><div>2007</div></div><div><div class='gray'>Images</div><div>13,233</div></div><div><div class='gray'>People</div><div>5,749</div></div><div><div class='gray'>Created From</div><div>Yahoo News images</div></div><div><div class='gray'>Search available</div><div>Searchable</div></div></div></section><section><p>Labeled Faces in The Wild (LFW) is amongst the most widely used facial recognition training datasets in the world and is the first of its kind to be created entirely from images posted online. The LFW dataset includes 13,233 images of 5,749 people that were collected between 2002-2004. Use the tools below to check if you were included in this dataset or scroll down to read the analysis.</p>
</section><section><div class='applet' data-payload='{"command": "face_search"}'></div></section><section><div class='applet' data-payload='{"command": "name_search"}'></div></section><section><div class='applet' data-payload='{"command": "load file", "opt": "https://nyc3.digitaloceanspaces.com/megapixels/v1/datasets/lfw/assets/lfw_names_gender_kg_min.csv", "fields": "Name, Images, Gender, Description"}'></div></section><section class='images'><div class='image'><img src='https://nyc3.digitaloceanspaces.com/megapixels/v1/datasets/lfw/assets/lfw_feature.jpg' alt='Eighteen of the 5,749 people in the Labeled Faces in the Wild Dataset. The most widely used face dataset for benchmarking commercial face recognition algorithms.'><div class='caption'>Eighteen of the 5,749 people in the Labeled Faces in the Wild Dataset. The most widely used face dataset for benchmarking commercial face recognition algorithms.</div></div></section><section><h2>Intro</h2>
<p>Three paragraphs describing the LFW dataset in a format that can be easily replicated for the other datasets. Nothing too custom. An analysis of the initial research papers with context relative to all the other dataset papers.</p>
</section><section class='images'><div class='image'><img src='https://nyc3.digitaloceanspaces.com/megapixels/v1/datasets/lfw/assets/lfw_montage_everyone_nocrop_1920.jpg' alt=' all 5,749 people in the LFW Dataset sorted from most to least images collected.'><div class='caption'> all 5,749 people in the LFW Dataset sorted from most to least images collected.</div></div></section><section><h2>LFW by the Numbers</h2>
<ul>
<li>Was first published in 2007</li>
<li>Developed out of a prior dataset from Berkely called "Faces in the Wild" or "Names and Faces" [^lfw_original_paper]</li>
<li>Includes 13,233 images and 5,749 different people [^lfw_website]</li>
<li>There are about 3 men for every 1 woman (4,277 men and 1,472 women)[^lfw_website]</li>
<li>The person with the most images is George W. Bush with 530</li>
<li>Most people (70%) in the dataset have only 1 image</li>
<li>Thre are 1,680 people in the dataset with 2 or more images [^lfw_website]</li>
<li>Two out of 4 of the original authors received funding from the Office of Director of National Intelligence and IARPA for their 2016 LFW survey follow up report </li>
<li>The LFW dataset includes over 500 actors, 30 models, 10 presidents, 24 football players, 124 basketball players, 11 kings, and 2 queens</li>
<li>In all the LFW publications provided by the authors the words "ethics", "consent", and "privacy" appear 0 times [^lfw_original_paper], [^lfw_survey], [^lfw_tech_report] , [^lfw_website]</li>
<li>The word "future" appears 71 times</li>
</ul>
<h1>Facts</h1>
<ul>
<li>Was created for the purpose of improving "unconstrained face recognition" [^lfw_original_paper]</li>
<li>All images in LFW were obtained "in the wild" meaning without any consent from the subject or from the photographer</li>
<li>The faces were detected using the Viola-Jones haarcascade face detector [^lfw_website] [^lfw_survey]</li>
<li>Is considered the "most popular benchmark for face recognition" [^lfw_baidu]</li>
<li>Is "the most widely used evaluation set in the field of facial recognition" [^lfw_pingan]</li>
<li>Is used by several of the largest tech companies in the world including "Google, Facebook, Microsoft Research Asia, Baidu, Tencent, SenseTime, Face++ and Chinese University of Hong Kong." [^lfw_pingan]</li>
</ul>
<p>need citations</p>
<ul>
<li>All images were copied from Yahoo News between 2002 - 2004 [^lfw_original_paper]</li>
<li>SenseTime, who has relied on LFW for benchmarking their facial recognition performance, is the leading provider of surveillance to the Chinese Government (need citation)</li>
</ul>
</section><section class='images'><div class='image'><img src='https://nyc3.digitaloceanspaces.com/megapixels/v1/datasets/lfw/assets/lfw_montage_top1_640.jpg' alt=' former President George W. Bush'><div class='caption'> former President George W. Bush</div></div>
<div class='image'><img src='https://nyc3.digitaloceanspaces.com/megapixels/v1/datasets/lfw/assets/lfw_montage_top2_4_640.jpg' alt=' Colin Powel (236), Tony Blair (144), and Donald Rumsfeld (121)'><div class='caption'> Colin Powel (236), Tony Blair (144), and Donald Rumsfeld (121)</div></div></section><section><h2>People and Companies using the LFW Dataset</h2>
<p>This section describes who is using the dataset and for what purposes. It should include specific examples of people or companies with citations and screenshots. This section is followed up by the graph, the map, and then the supplementary material.</p>
<p>The LFW dataset is used by numerous companies for <a href="about/glossary#benchmarking">benchmarking</a> algorithms and in some cases <a href="about/glossary#training">training</a>. According to the benchmarking results page [^lfw_results] provided by the authors, over 2 dozen companies have contributed their benchmark results.</p>
<p>According to BiometricUpdate.com [^lfw_pingan], LFW is "the most widely used evaluation set in the field of facial recognition, LFW attracts a few dozen teams from around the globe including Google, Facebook, Microsoft Research Asia, Baidu, Tencent, SenseTime, Face++ and Chinese University of Hong Kong."</p>
<p>According to researchers at the Baidu Research – Institute of Deep Learning "LFW has been the most popular evaluation benchmark for face recognition, and played a very important role in facilitating the face recognition society to improve algorithm. [^lfw_baidu]."</p>
<p>In addition to commercial use as an evaluation tool, alll of the faces in LFW dataset are prepackaged into a popular machine learning code framework called scikit-learn.</p>
</section><section><div class='applet' data-payload='{"command": "load file", "opt": "https://nyc3.digitaloceanspaces.com/megapixels/v1/datasets/lfw/assets/lfw_commercial_use.csv", "fields": "name_display, company_url, example_url, country, description"}'></div></section><section><table>
<thead><tr>
<th style="text-align:left">Company</th>
<th style="text-align:left">Country</th>
<th style="text-align:left">Industries</th>
</tr>
</thead>
<tbody>
<tr>
<td style="text-align:left"><a href="http://www.aratek.co">Aratek</a></td>
<td style="text-align:left">China</td>
<td style="text-align:left">Biometric sensors for telecom, civil identification, finance, education, POS, and transportation</td>
</tr>
<tr>
<td style="text-align:left"><a href="http://www.aratek.co">Aratek</a></td>
<td style="text-align:left">China</td>
<td style="text-align:left">Biometric sensors for telecom, civil identification, finance, education, POS, and transportation</td>
</tr>
<tr>
<td style="text-align:left"><a href="http://www.aratek.co">Aratek</a></td>
<td style="text-align:left">China</td>
<td style="text-align:left">Biometric sensors for telecom, civil identification, finance, education, POS, and transportation</td>
</tr>
</tbody>
</table>
<p>Add 2-4 screenshots of companies mentioning LFW here</p>
</section><section class='images'><div class='image'><img src='https://nyc3.digitaloceanspaces.com/megapixels/v1/datasets/lfw/assets/lfw_screenshot_01.jpg' alt=' "PING AN Tech facial recognition receives high score in latest LFW test results"'><div class='caption'> "PING AN Tech facial recognition receives high score in latest LFW test results"</div></div>
<div class='image'><img src='https://nyc3.digitaloceanspaces.com/megapixels/v1/datasets/lfw/assets/lfw_screenshot_02.jpg' alt=' "Face Recognition Performance in LFW benchmark"'><div class='caption'> "Face Recognition Performance in LFW benchmark"</div></div>
<div class='image'><img src='https://nyc3.digitaloceanspaces.com/megapixels/v1/datasets/lfw/assets/lfw_screenshot_03.jpg' alt=' "The 1st place in face verification challenge, LFW"'><div class='caption'> "The 1st place in face verification challenge, LFW"</div></div></section><section><p>In benchmarking, companies use a dataset to evaluate their algorithms which are typically trained on other data. After training, researchers will use LFW as a benchmark to compare results with other algorithms.</p>
<p>For example, Baidu (est. net worth $13B) uses LFW to report results for their "Targeting Ultimate Accuracy: Face Recognition via Deep Embedding". According to the three Baidu researchers who produced the paper:</p>
<h2>Citations</h2>
<p>Overall, LFW has at least 456 citations from 123 countries. Sed ut perspiciatis, unde omnis iste natus error sit voluptatem accusantium doloremque laudantium, totam rem aperiam eaque ipsa, quae ab illo inventore veritatis et quasi architecto beatae vitae dicta sunt, explicabo. Nemo enim ipsam voluptatem, quia voluptas sit, aspernatur aut odit aut fugit, sed quia consequuntur magni dolores eos.</p>
<p>Sed ut perspiciatis, unde omnis iste natus error sit voluptatem accusantium doloremque laudantium, totam rem aperiam eaque ipsa, quae ab illo inventore veritatis et quasi architecto beatae vitae dicta sunt, explicabo. Nemo enim ipsam voluptatem, quia voluptas sit, aspernatur aut odit aut fugit, sed quia consequuntur magni dolores eos.</p>
</section><section class='images'><div class='image'><img src='https://nyc3.digitaloceanspaces.com/megapixels/v1/datasets/lfw/assets/temp_graph.jpg' alt='Distribution of citations per year per country for the top 5 countries with citations for the LFW Dataset'><div class='caption'>Distribution of citations per year per country for the top 5 countries with citations for the LFW Dataset</div></div></section><section class='images'><div class='image'><img src='https://nyc3.digitaloceanspaces.com/megapixels/v1/datasets/lfw/assets/temp_map.jpg' alt='Geographic distributions of citations for the LFW Dataset'><div class='caption'>Geographic distributions of citations for the LFW Dataset</div></div></section><section><h2>Conclusion</h2>
<p>The LFW face recognition training and evaluation dataset is a historically important face dataset as it was the first popular dataset to be created entirely from Internet images, paving the way for a global trend towards downloading anyone’s face from the Internet and adding it to a dataset. As will be evident with other datasets, LFW’s approach has now become the norm.</p>
<p>For all the 5,000 people in this datasets, their face is forever a part of facial recognition history. It would be impossible to remove anyone from the dataset because it is so ubiquitous. For their rest of the lives and forever after, these 5,000 people will continue to be used for training facial recognition surveillance.</p>
<h2>Right to Removal</h2>
<p>If you are affected by disclosure of your identity in this dataset please do contact the authors. Many have stated that they are willing to remove images upon request. The authors of the LFW dataset provide the following email for inquiries:</p>
<p>You can use the following message to request removal from the dataset:</p>
<p>To: Gary Huang <a href="mailto:mailto:gbhuang@cs.umass.edu">mailto:gbhuang@cs.umass.edu</a></p>
<p>Subject: Request for Removal from LFW Face Dataset</p>
<p>Dear [researcher name],</p>
<p>I am writing to you about the "Labeled Faces in The Wild Dataset". Recently I discovered that your dataset includes my identity and I no longer wish to be included in your dataset.</p>
<p>The dataset is being used thousands of companies around the world to improve facial recognition software including usage by governments for the purpose of law enforcement, national security, tracking consumers in retail environments, and tracking individuals through public spaces.</p>
<p>My name as it appears in your dataset is [your name]. Please remove all images from your dataset and inform your newsletter subscribers to likewise update their copies.</p>
<p>- [your name]</p>
<hr>
<h2>Supplementary Data</h2>
<p>Researchers, journ</p>
<table>
<thead><tr>
<th style="text-align:left">Title</th>
<th style="text-align:left">Organization</th>
<th style="text-align:left">Country</th>
<th style="text-align:left">Type</th>
</tr>
</thead>
<tbody>
<tr>
<td style="text-align:left">3D-aided face recognition from videos</td>
<td style="text-align:left">University of Lyon</td>
<td style="text-align:left">France</td>
<td style="text-align:left">edu</td>
</tr>
<tr>
<td style="text-align:left">A Community Detection Approach to Cleaning Extremely Large Face Database</td>
<td style="text-align:left">National University of Defense Technology, China</td>
<td style="text-align:left">China</td>
<td style="text-align:left">edu</td>
</tr>
<tr>
<td style="text-align:left">3D-aided face recognition from videos</td>
<td style="text-align:left">University of Lyon</td>
<td style="text-align:left">France</td>
<td style="text-align:left">edu</td>
</tr>
<tr>
<td style="text-align:left">3D-aided face recognition from videos</td>
<td style="text-align:left">University of Lyon</td>
<td style="text-align:left">France</td>
<td style="text-align:left">edu</td>
</tr>
<tr>
<td style="text-align:left">3D-aided face recognition from videos</td>
<td style="text-align:left">University of Lyon</td>
<td style="text-align:left">France</td>
<td style="text-align:left">edu</td>
</tr>
<tr>
<td style="text-align:left">3D-aided face recognition from videos</td>
<td style="text-align:left">University of Lyon</td>
<td style="text-align:left">France</td>
<td style="text-align:left">edu</td>
</tr>
<tr>
<td style="text-align:left">3D-aided face recognition from videos</td>
<td style="text-align:left">University of Lyon</td>
<td style="text-align:left">France</td>
<td style="text-align:left">edu</td>
</tr>
<tr>
<td style="text-align:left">3D-aided face recognition from videos</td>
<td style="text-align:left">University of Lyon</td>
<td style="text-align:left">France</td>
<td style="text-align:left">edu</td>
</tr>
<tr>
<td style="text-align:left">3D-aided face recognition from videos</td>
<td style="text-align:left">University of Lyon</td>
<td style="text-align:left">France</td>
<td style="text-align:left">edu</td>
</tr>
<tr>
<td style="text-align:left">3D-aided face recognition from videos</td>
<td style="text-align:left">University of Lyon</td>
<td style="text-align:left">France</td>
<td style="text-align:left">edu</td>
</tr>
<tr>
<td style="text-align:left">3D-aided face recognition from videos</td>
<td style="text-align:left">University of Lyon</td>
<td style="text-align:left">France</td>
<td style="text-align:left">edu</td>
</tr>
<tr>
<td style="text-align:left">3D-aided face recognition from videos</td>
<td style="text-align:left">University of Lyon</td>
<td style="text-align:left">France</td>
<td style="text-align:left">edu</td>
</tr>
<tr>
<td style="text-align:left">3D-aided face recognition from videos</td>
<td style="text-align:left">University of Lyon</td>
<td style="text-align:left">France</td>
<td style="text-align:left">edu</td>
</tr>
<tr>
<td style="text-align:left">3D-aided face recognition from videos</td>
<td style="text-align:left">University of Lyon</td>
<td style="text-align:left">France</td>
<td style="text-align:left">edu</td>
</tr>
<tr>
<td style="text-align:left">3D-aided face recognition from videos</td>
<td style="text-align:left">University of Lyon</td>
<td style="text-align:left">France</td>
<td style="text-align:left">edu</td>
</tr>
</tbody>
</table>
<h2>Code</h2>
</section><section><pre><code class="lang-python">#!/usr/bin/python
import numpy as np
from sklearn.datasets import fetch_lfw_people
import imageio
import imutils
# download LFW dataset (first run takes a while)
lfw_people = fetch_lfw_people(min_faces_per_person=1, resize=1, color=True, funneled=False)
# introspect dataset
n_samples, h, w, c = lfw_people.images.shape
print('{:,} images at {}x{}'.format(n_samples, w, h))
cols, rows = (176, 76)
n_ims = cols * rows
# build montages
im_scale = 0.5
ims = lfw_people.images[:n_ims
montages = imutils.build_montages(ims, (int(w*im_scale, int(h*im_scale)), (cols, rows))
montage = montages[0]
# save full montage image
imageio.imwrite('lfw_montage_full.png', montage)
# make a smaller version
montage_960 = imutils.resize(montage, width=960)
imageio.imwrite('lfw_montage_960.jpg', montage_960)
</code></pre>
</section><section><h2>Disclaimer</h2>
<p>MegaPixels is an educational art project designed to encourage discourse about facial recognition datasets. Any ethical or legal issues should be directed to the researcher's parent organizations. Except where necessary for contact or clarity, the names of researchers have been subsituted by their parent organization. In no way does this project aim to villify researchers who produced the datasets.</p>
<p>Read more about <a href="about/code-of-conduct">MegaPixels Code of Conduct</a></p>
<div class="footnotes">
<hr>
<ol></ol>
</div>
</section>
</div>
<footer>
<div>
<a href="/">MegaPixels.cc</a>
<a href="/about/disclaimer/">Disclaimer</a>
<a href="/about/terms/">Terms of Use</a>
<a href="/about/privacy/">Privacy</a>
<a href="/about/">About</a>
<a href="/about/team/">Team</a>
</div>
<div>
MegaPixels ©2017-19 Adam R. Harvey /
<a href="https://ahprojects.com">ahprojects.com</a>
</div>
</footer>
</body>
<script src="/assets/js/dist/index.js"></script>
</html>
|
