-
Notifications
You must be signed in to change notification settings - Fork 3
/
cvpr_2021_data_efficient_tutorial.html
260 lines (202 loc) · 15.8 KB
/
cvpr_2021_data_efficient_tutorial.html
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
<!DOCTYPE html>
<html lang="en">
<head>
<title>VITA</title>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no">
<link href="https://fonts.googleapis.com/css?family=B612+Mono|Cabin:400,700&display=swap" rel="stylesheet">
<link rel="stylesheet" href="fonts/icomoon/style.css">
<link rel="stylesheet" href="https://stackpath.bootstrapcdn.com/bootstrap/4.4.1/css/bootstrap.min.css"
integrity="sha384-Vkoo8x4CGsO3+Hhxv8T/Q5PaXtkKtu6ug5TOeNV6gBiFeWPGFN9MuhOf23Q9Ifjh" crossorigin="anonymous">
<link rel="stylesheet" href="css/jquery-ui.css">
<link rel="stylesheet" href="css/owl.carousel.min.css">
<link rel="stylesheet" href="css/owl.theme.default.min.css">
<link rel="stylesheet" href="css/owl.theme.default.min.css">
<link rel="stylesheet" href="css/jquery.fancybox.min.css">
<link rel="stylesheet" href="fonts/flaticon/font/flaticon.css">
<link rel="stylesheet" href="css/aos.css">
<link href="css/jquery.mb.YTPlayer.min.css" media="all" rel="stylesheet" type="text/css">
<link rel="stylesheet" href="css/style.css">
<!-- HTML5 shim and Respond.js for IE8 support of HTML5 elements and media queries -->
<!--[if lt IE 9]>
<script src="https://oss.maxcdn.com/html5shiv/3.7.2/html5shiv.min.js"></script>
<script src="https://oss.maxcdn.com/respond/1.4.2/respond.min.js"></script>
<![endif]-->
</head>
<body data-spy="scroll" data-target=".site-navbar-target" data-offset="300">
<div class="site-section">
<div class="container">
<div class="row">
<div class="col-lg-12">
<div class="section-title" style="margin-bottom:20px">
<h2>CVPR 2021 Tutorial: Data- and Label-Efficient Learning in An Imperfect World</h2>
</div>
<div class="trend-entry d-flex">
<div class="trend-contents">
<div class="trend-contents">
<b style="color:rgb(68, 68, 68); font-size:22px"></b>
<div style="color:rgb(68, 68, 68); padding-bottom:50px" class="trend-contents">
<div style="width:70%; margin-left: 0;" class="row">
<p><div class="column_1">Time</div>
<div class="column_2">June 19, 2-7pm US Eastern time</div></p>
<p><div class="column_1">Location</div>
<div class="column_2">Zoom online</div></p>
</div>
</div>
</div>
<b style="color:rgb(68, 68, 68); font-size:22px">Topic Description</b>
<div style="color:rgb(68, 68, 68); padding-bottom:50px" class="trend-contents">
<p>Despite their empirical successes in computer vision, deep networks often demand “big” and carefully labeled training data. When being applied to studying complex problems in the real visual world, their performance is limited since both data and labels can be notoriously difficult, costly to obtain, or may come in various noisy, weak or long-tailed forms. For example, collecting image data in many scientific and engineering disciplines (astronomy, material science, geoscience, medicine and so on) often hinges on expensive and high-stake experiments. Further, the operation of data labeling in these applications is also tedious to scale up, whose complexity demands highly skilled professionals, creating challenges to use some cost-effective solutions such as crowdsourcing. Moreover, labeling data in large scale using crowdsourcing are often infeasible for proprietary or sensitive data. What is worse, additional labels are always needed when the trained models face changes in their operating environments and need to be adapted. For many problems, the labeled data required to adapt models to new environments approaches the amount required to train from scratch. Therefore, continuous data and label collection is also needed when the systems exhibit non-stationary properties or operate in varying environments.</p>
<p>In this tutorial, we address the grand challenge of data- and label-efficient visual learning in realistic and imperfect visual environments. We plan to focus on a comprehensive suite of state-of-the-art techniques to tackle this problem from multiple levels, including unsupervised/self-supervised learning, weakly-supervised learning, long tail visual recognition, domain adaptation and meta learning. We will also demonstrate those techniques in representative computer vision applications such as (interactive) segmentation, autonomous driving, and medical image understanding. The organizers will share their extensive experience on this topic and provide links to resources such as relevant datasets and source code. </p>
<p> The Economist once published a story titled, <a
href="https://www.economist.com/leaders/2017/05/06/the-worlds-most-valuable-resource-is-no-longer-oil-but-data">“The world's most valuable resource is no longer oil, but data.”</a> However, acquiring perfect data is usually inefficient or even hopeless for some research areas or applications such as segmentation, satellite / agriculture / medical imagery. Our aim is to efficiently leverage the existing data, either rich or scarce, and no matter they are labeled, weakly labeled, unlabeled, noisy, with domain gaps, etc., towards learning reliable recognition models for real world applications. We believe the topics covered by this tutorial will attract a wide range of researchers working on unsupervised / few-shot / weakly supervised / domain adaptive / meta learning, and label- or data-limited applications such as medical imagery from both academia and industry. </p>
</div>
<div class="section-title">
<b style="color:rgb(68, 68, 68); font-size:22px">Speakers</b>
</div>
<div class="row justify-content-md-left" style="margin-top: 20px">
<div class="col-md-4" style="margin-bottom: 40px">
<div class="card">
<a href="https://vita-group.github.io/"><img src="groups/atlaswang.jpg" width="100%" style="margin:auto"></a>
<div class="post-meta">
<span class="d-block" style="padding-top:10px"><a
href="https://vita-group.github.io/">Atlas Wang</a> </span>
<span class="d-block">Email: <a
href="mailto:[email protected]">atlaswang [at] utexas (dot) edu</a> </span>
</div>
</div>
</div>
<div class="col-md-4" style="margin-bottom: 40px">
<div class="card">
<a href="https://weiyc.github.io/"><img src="for-tur/wyc.jpg" width="110%" style="margin:auto"></a>
<div class="post-meta">
<span class="d-block" style="padding-top:10px"><a
href="https://weiyc.github.io/">Yunchao Wei</a> </span>
<span class="d-block">Email: <a
href="mailto:[email protected]">yunchao.wei [at] uts (dot) edu.au</a> </span>
</div>
</div>
</div>
<div class="col-md-4" style="margin-bottom: 40px">
<div class="card">
<a href="http://elisaricci.eu/"><img src="for-tur/Elisa.jpg" width="125%" style="margin:auto"></a>
<div class="post-meta">
<span class="d-block" style="padding-top:10px"><a
href="http://elisaricci.eu/">Elisa Ricci</a> </span>
<span class="d-block">Email: <a
href="mailto:[email protected]">e.ricci [at] unitn (dot) it</a> </span>
</div>
</div>
</div>
<div class="col-md-4" style="margin-bottom: 40px">
<div class="card">
<a href="http://boqinggong.info/"><img src="for-tur/boqing.jpg" width="100%" style="margin:auto"></a>
<div class="post-meta">
<span class="d-block" style="padding-top:10px"><a
href="http://boqinggong.info/">Boqing Gong</a> </span>
<span class="d-block">Email: <a
href="mailto:[email protected]">bgong [at] google (dot) com</a> </span>
</div>
</div>
</div>
<div class="col-md-4" style="margin-bottom: 40px">
<div class="card">
<a href="https://research.google/people/106719/"><img src="for-tur/ting.jpeg" width="128%" style="margin:auto"></a>
<div class="post-meta">
<span class="d-block" style="padding-top:10px"><a
href="https://research.google/people/106719/">Ting Chen</a> </span>
<span class="d-block">Email: <a
href="mailto:[email protected]">iamtingchen [at] google (dot) com</a> </span>
</div>
</div>
</div>
<div class="col-md-4" style="margin-bottom: 40px">
<div class="card">
<a href="https://xmengli999.github.io/"><img src="for-tur/meng.jpg" width="119%" style="margin:auto"></a>
<div class="post-meta">
<span class="d-block" style="padding-top:10px"><a
href="https://xmengli999.github.io/">Xiaomeng Li</a> </span>
<span class="d-block">Email: <a
href="mailto:[email protected]">eexmli [at] ust (dot) hk</a> </span>
</div>
</div>
</div>
</div>
<div class="trend-contents">
<b style="color:rgb(68, 68, 68); font-size:22px">Talk Outline (all in US eastern time)</b>
<div style="color:rgb(68, 68, 68)" class="trend-contents">
Every topic talk is scheduled to be 45 minutes, plus 5 minutes Q&A.
<br />
<div style="width:100%; margin-left: 0;" class="row">
<p><div class="column_1">2:00 - 2:50 pm</div> <div class="column_2"> Topic I: Self-Supervised Visual Representation Learning - Ting Chen <a
href="https://utexas.box.com/s/4ey84llqfhp6dpciqkoya689jmnnuour">[Slides]</a> <a
href="https://utexas.box.com/s/ut6g4qwm4o7fpmciu3qizfy3yqxo6bmp">[Record]</a> </div></p>
<p><div class="column_1">2:50 - 3:35 pm</div> <div class="column_2"> Topic II: Long-Tailed Visual Recognition in the Wild - Boqing Gong <a
href="https://utexas.box.com/s/x5adt0j140i8y2boz69t0dgiyxgf9510">[Slides]</a> <a
href="https://utexas.box.com/s/tgicdrog4x3avsd2qpyuueeg4nrpp7po">[Record]</a> </div></p>
<p><div class="column_1">3:35 - 4:25 pm</div> <div class="column_2"> Topic III: More Robustness from Free Unlabeled Data - Atlas Wang <a
href="https://utexas.box.com/s/whatyr94gdq2k1paysigkcsyv9cr9vte">[Slides]</a> <a
href="https://utexas.box.com/s/h6tceficg2ikl0kbpthiup8aojzztoh1">[Record]</a> </div></p>
<p><div class="column_1">4:25 - 4:35 pm</div> <div class="column_2"> Break Time </div></p>
<p><div class="column_1">4:35 - 5:20 pm</div> <div class="column_2"> Topic IV: Towards Data-efficient Visual Segmentation - Yunchao Wei <a
href="https://utexas.box.com/s/56owvikw2ju10mlyv606qvv72bnn5a8n">[Slides]</a> <a
href="https://utexas.box.com/s/p4m3f3put58r9qnriry7fzh8em03mjge">[Record]</a> </div></p>
<p><div class="column_1">5:20 - 6:10 pm</div> <div class="column_2"> Topic V: Domain Adaptive Visual Understanding - Elisa Ricci <a
href="https://utexas.box.com/s/6mdtvt1wj8hsojzen918xekjyh43zxtx">[Slides]</a> <a
href="https://utexas.box.com/s/ph3xebwa2hri404k8sml5kfovgi4ak1k">[Record]</a> </div></p>
<p><div class="column_1">6:10 - 7:00 pm</div> <div class="column_2"> Topic VI: Data-Efficient Deep Learning for Medical Image Analysis - Xiaomeng Li <a
href="https://utexas.box.com/s/91l9u8a0g3lhhjcxhhgjnjn11p6mp73g">[Slides]</a> <a
href="https://utexas.box.com/s/88xqr0urcq5y0l5xtm7b25loh1y0lx7v">[Record]</a> </div></p>
</div>
</div>
</div>
<br />
</div>
</div>
</div>
</div>
<!-- END section -->
<div class="footer">
<div class="container">
<div class="row">
<div class="col-12">
<div class="copyright">
<p>
<!-- Link back to Colorlib can't be removed. Template is licensed under CC BY 3.0. -->
Copyright ©<script>document.write(new Date().getFullYear());</script>
All rights reserved | Built upon <a
href="https://colorlib.com" target="_blank">Colorlib</a>
<!-- Link back to Colorlib can't be removed. Template is licensed under CC BY 3.0. -->
</p>
</div>
</div>
</div>
</div>
</div>
</div>
<!-- .site-wrap -->
<!-- loader -->
<!-- <div id="loader" class="show fullscreen">
<svg class="circular" width="48px" height="48px">
<circle class="path-bg" cx="24" cy="24" r="22" fill="none" stroke-width="4" stroke="#eeeeee"/>
<circle class="path" cx="24" cy="24" r="22" fill="none" stroke-width="4" stroke-miterlimit="10"
stroke="#ff5e15"/>
</svg>
</div> -->
<script src="js/jquery-3.3.1.min.js"></script>
<script src="js/jquery-migrate-3.0.1.min.js"></script>
<script src="js/jquery-ui.js"></script>
<script src="js/popper.min.js"></script>
<script src="js/bootstrap.min.js"></script>
<script src="js/owl.carousel.min.js"></script>
<script src="js/jquery.stellar.min.js"></script>
<script src="js/jquery.countdown.min.js"></script>
<script src="js/bootstrap-datepicker.min.js"></script>
<script src="js/jquery.easing.1.3.js"></script>
<script src="js/aos.js"></script>
<script src="js/jquery.fancybox.min.js"></script>
<script src="js/jquery.sticky.js"></script>
<script src="js/jquery.mb.YTPlayer.min.js"></script>
<script src="js/main.js"></script>
</body>
</html>