-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathindex.html
More file actions
398 lines (352 loc) · 20.8 KB
/
index.html
File metadata and controls
398 lines (352 loc) · 20.8 KB
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
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
<!DOCTYPE html>
<html lang="en">
<head>
<style>
table {
font-family: arial, sans-serif;
border-collapse: collapse;
width: 100%;
}
td, th {
border: 2px solid #dddddd;
text-align: center;
padding: 8px;
}
tr:nth-child(even) {
background-color: #dddddd;
}
</style>
<meta charset="utf-8">
<meta http-equiv="X-UA-Compatible" content="IE=edge">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title>Moment Matching for Multi-Source Domain Adaptation</title>
<!-- Bootstrap -->
<link href="../css/bootstrap.min.css" rel="stylesheet">
<link href="../css/template.css" rel="stylesheet">
<!--link rel="icon" type="image/png" href="images/icons/favicon.ico"/-->
<!--link rel="stylesheet" type="text/css" href="vendor/bootstrap/css/bootstrap.min.css"-->
<!--link rel="stylesheet" type="text/css" href="fonts/font-awesome-4.7.0/css/font-awesome.min.css"-->
<!--link rel="stylesheet" type="text/css" href="vendor/animate/animate.css"-->
<!--link rel="stylesheet" type="text/css" href="vendor/select2/select2.min.css"-->
<!--link rel="stylesheet" type="text/css" href="vendor/perfect-scrollbar/perfect-scrollbar.css"-->
<!--link rel="stylesheet" type="text/css" href="css/util.css"-->
<!--link rel="stylesheet" type="text/css" href="css/main.css"-->
</head>
<body>
<nav class="navbar navbar-inverse navbar-fixed-top" role="navigation"
style="background-color:#34a853;border-color:#34a853">
<!-- Collect the nav links, forms, and other content for toggling -->
<div class="collapse navbar-collapse navbar-ex1-collapse">
<ul class="nav navbar-nav">
<li role="presentation"><a href="#overview">Overview</a></li>
<li role="presentation"><a href="#dataset">Dataset</a></li>
<li role="presentation"><a href="#code">Code</a></li>
<li role="presentation"><a href="#refs">Reference</a></li>
<!--<li><a href="./index.html#Contact">Contact</a></li>-->
</ul>
<!-- </div>--> <!--/.navbar-collapse -->
</div>
</nav>
<div class="container">
<div class="home-intro" style="padding: 5% 10%">
<div class="row">
<h1 align="center">Moment Matching for <br>Multi-Source Domain Adaptation</h1>
<div style="text-align:center">
Xingchao Peng, Qinxun Bai, Xide Xia, Zijun Huang, Kate Saenko, Bo Wang<br>
Our paper is accepted by ICCV 2019 as an <b>Oral</b> Presentation!
</div>
</div>
<div class="project-page">
<a name="abstract"></a>
<h2>Abstract</h2>
<p class="text-justify">
Conventional unsupervised domain adaptation (UDA) assumes that training data are sampled from a single
domain. This neglects the more practical scenario where
training data are collected from multiple sources, requiring multi-source domain adaptation. We make
three major contributions towards addressing this problem. First, we
propose a new deep learning approach, Moment Matching for Multi-Source Domain Adaptation (M3SDA), which
aims to transfer knowledge learned from multiple labeled source
domains to an unlabeled target domain by dynamically aligning moments of their feature distributions.
Second, we provide a sound theoretical analysis of moment-related error
bounds for multi-source domain adaptation. Third, we collect and annotate by far the largest UDA dataset
with six distinct domains and approximately 0.6 million images distributed
among 345 categories, addressing the gap in data availability for multi-source UDA research. Extensive
experiments are performed to demonstrate the effectiveness of
our proposed model, which outperforms existing state-ofthe-art methods by a large margin.
</p>
<center>
<a
href="https://arxiv.org/pdf/1812.01754.pdf" target="_blank" class="btn btn-danger"
role="button">PDF</a>
</center>
</div>
<hr class="soften"></hr>
<div class="project-page">
<a name="overview"></a>
<h2>Overview</h2>
<br>
<center>
<img width="90%" alt="License" src="imgs/overview.png"></img>
</center>
<br>
</div>
<div class="project-page">
<a name="dataset"></a>
<h2>DomainNet Dataset</h2>
<br>
<center>
<img width="90%" alt="License" src="imgs/data_examples.png"></img>
<img width="90%" alt="License" src="imgs/statistics.png"></img>
</center>
<br>
</div>
<div class="project-page">
<a name="download"></a>
<h2>Download (cleaned version, recommended)</h2>
<p>Special Thanks to Guoliang Kang for pointing out the image duplications in our dataset.</p>
<br>
<div>
<table>
<thead>
<tr>
<th></th>
<th>Clipart</th>
<th>Infograph</th>
<th>Painting</th>
<th>Quickdraw</th>
<th>Real</th>
<th>Sketch</th>
</tr>
</thead>
<tbody>
<tr>
<td>Data</td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/groundtruth/clipart.zip">clipart.zip</a><b>(1.3G)</b></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/infograph.zip">infograph.zip</a><b>(4.0G)</b></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/groundtruth/painting.zip">painting.zip</a><b>(3.7G)</b></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/quickdraw.zip">quickdraw.zip</a><b>(439M)</b></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/real.zip">real.zip</a><b>(5.6G)</b></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/sketch.zip">sketch.zip</a><b>(2.5G)</b></td>
</tr>
<tr>
<td>Train</td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/domainnet/txt/clipart_train.txt">clipart_train.txt</a></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/domainnet/txt/infograph_train.txt">infograph_train.txt</a></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/domainnet/txt/painting_train.txt">painting_train.txt</a></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/domainnet/txt/quickdraw_train.txt">quickdraw_train.txt</a></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/domainnet/txt/real_train.txt">real_train.txt</a></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/domainnet/txt/sketch_train.txt">sketch_train.txt</a></td>
</tr>
<tr>
<td>Test</td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/domainnet/txt/clipart_test.txt">clipart_test.txt</a></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/domainnet/txt/infograph_test.txt">infograph_test.txt</a></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/domainnet/txt/painting_test.txt">painting_test.txt</a></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/domainnet/txt/quickdraw_test.txt">quickdraw_test.txt</a></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/domainnet/txt/real_test.txt">real_test.txt</a></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/domainnet/txt/sketch_test.txt">sketch_test.txt</a></td>
</tr>
<tr>
<td>#Original</td>
<td>48,837</td>
<td>53,201</td>
<td>75,759</td>
<td>172,500</td>
<td>175,327</td>
<td>70,386</td>
</tr>
<tr>
<td>#Cleaned</td>
<td>48,129</td>
<td>51,605</td>
<td>72,266</td>
<td>172,500</td>
<td>172,947</td>
<td>69,128</td>
</tr>
<tr>
<td>Removed</td>
<td>1.4%</td>
<td>2.9%</td>
<td>4.6%</td>
<td>0</td>
<td>1.4%</td>
<td>1.8%</td>
</tr>
</tbody>
</table>
</div>
<br>
<br>
<!--<li>We retain the <i>clipart</i> and <i>painting</i> subset as we are going to hold a domain adaptation challenge.
These two domains will be used as the testing domain.
</li>
<li>Train/Test images are structured in folders as {domain}/{cateogory}/{object_id}.jpg (or .png)</li>
</ul>-->
<br>
</div>
<div class="project-page">
<a name="download"></a>
<h2>Download (<a href="https://arxiv.org/pdf/1812.01754.pdf">ICCV19</a> version)</h2>
<br>
<div>
<table>
<thead>
<tr>
<th></th>
<th>Clipart</th>
<th>Infograph</th>
<th>Painting</th>
<th>Quickdraw</th>
<th>Real</th>
<th>Sketch</th>
</tr>
</thead>
<tbody>
<tr>
<td>Data</td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/groundtruth/clipart.zip">clipart.zip</a><b>(1.3G)</b></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/infograph.zip">infograph.zip</a><b>(4.0G)</b></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/groundtruth/painting.zip">painting.zip</a><b>(3.7G)</b></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/quickdraw.zip">quickdraw.zip</a><b>(439M)</b></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/real.zip">real.zip</a><b>(5.6G)</b></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/sketch.zip">sketch.zip</a><b>(2.5G)</b></td>
</tr>
<tr>
<td>Train</td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/groundtruth/txt/clipart_train.txt">clipart_train.txt</a></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/txt/infograph_train.txt">infograph_train.txt</a></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/groundtruth/txt/painting_train.txt">painting_train.txt</a></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/txt/quickdraw_train.txt">quickdraw_train.txt</a></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/txt/real_train.txt">real_train.txt</a></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/txt/sketch_train.txt">sketch_train.txt</a></td>
</tr>
<tr>
<td>Test</td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/groundtruth/txt/clipart_test.txt">clipart_test.txt</a></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/txt/infograph_test.txt">infograph_test.txt</a></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/groundtruth/txt/painting_test.txt">painting_test.txt</a></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/txt/quickdraw_test.txt">quickdraw_test.txt</a></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/txt/real_test.txt">real_test.txt</a></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/txt/sketch_test.txt">sketch_test.txt</a></td>
</tr>
</tbody>
</table>
</div>
<br>
<br>
<!--<li>We retain the <i>clipart</i> and <i>painting</i> subset as we are going to hold a domain adaptation challenge.
These two domains will be used as the testing domain.
</li>
<li>Train/Test images are structured in folders as {domain}/{cateogory}/{object_id}.jpg (or .png)</li>
</ul>-->
<br>
</div>
<div class="project-page">
<a name="chellenge-download"></a>
<h2>Download (<a href="http://ai.bu.edu/visda-2019/">VisDA 2019 Challenge</a> version) </h2>
<br>
<div>
<table>
<thead>
<tr>
<th></th>
<th>Clipart</th>
<th>Infograph</th>
<th>Painting</th>
<th>Quickdraw</th>
<th>Real</th>
<th>Sketch</th>
</tr>
</thead>
<tbody>
<tr>
<td>Data</td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/clipart.zip">clipart.zip</a><b>(1.3G)</b></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/infograph.zip">infograph.zip</a><b>(4.0G)</b></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/painting.zip">painting.zip</a><b>(3.7G)</b></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/quickdraw.zip">quickdraw.zip</a><b>(439M)</b></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/real.zip">real.zip</a><b>(5.6G)</b></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/sketch.zip">sketch.zip</a><b>(2.5G)</b></td>
</tr>
<tr>
<td>Train</td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/txt/clipart_train.txt">clipart_train.txt</a></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/txt/infograph_train.txt">infograph_train.txt</a></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/txt/painting_train.txt">painting_train.txt</a></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/txt/quickdraw_train.txt">quickdraw_train.txt</a></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/txt/real_train.txt">real_train.txt</a></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/txt/sketch_train.txt">sketch_train.txt</a></td>
</tr>
<tr>
<td>Test</td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/txt/clipart_test.txt">clipart_test.txt</a></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/txt/infograph_test.txt">infograph_test.txt</a></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/txt/painting_test.txt">painting_test.txt</a></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/txt/quickdraw_test.txt">quickdraw_test.txt</a></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/txt/real_test.txt">real_test.txt</a></td>
<td><a href="http://csr.bu.edu/ftp/visda/2019/multi-source/txt/sketch_test.txt">sketch_test.txt</a></td>
</tr>
</tbody>
</table>
</div>
<br>
<br>
<ul>
<!--<li>We retain the <i>clipart</i> and <i>painting</i> subset as we are going to hold a domain adaptation challenge.
These two domains will be used as the testing domain.
</li>
<li>Train/Test images are structured in folders as {domain}/{cateogory}/{object_id}.jpg (or .png)</li>
</ul>-->
<li>We have released the training/testing images of <i>clipart</i> and <i>painting</i> subset now! </li>
<li>For <i>Infograph</i>, <i>Quickdraw</i>, <i>Real</i>, <i>Sketch</i> domains, the train/test images are structured in folders as {domain}/{cateogory}/{object_id}.jpg (or .png)</li>
<li>For <i>clipart</i> and <i>painting</i> domains, the train/test images are structured as {domain}/{train/test}/{trunk_ID}/{MD5}.jpg</li>
<li>To prepare for the <a href="https://competitions.codalab.org/competitions/20256">submission</a>, please output the predictions to a "result.txt" file, with one prediciton ranging from 0 to 344 per line (37706 lines for 37706 test samples). The order of the testing images is listed in
<a href="http://csr.bu.edu/ftp/visda/2019/multi-source/txt/submission_list.txt">submission_list.txt</a>.
</li>
<br>
</div>
<hr class="soften"></hr>
<div class="project-page">
<a name="code"></a>
<h2>Code</h2>
<p class="lead">
</br>
<a href="https://github.com/VisionLearningGroup/VisionLearningGroup.github.io/tree/master/M3SDA/code_MSDA_digit" target="_blank">[Code]
<!--detection]</a>-->
<!--<a href="https://github.com/VisionLearningGroup/taskcv-2017-public" target="_blank">[classification and-->
<!--segmentation]</a>-->
</br>
</p>
</div>
<hr class="soften"></hr>
<div class="project-page">
<a name="refs"></a>
<h2>Reference</h2>
<p class="lead">
If you find this useful in your work please consider citing:
<div class="highlight">
<pre> <code>
@inproceedings{peng2019moment,
title={Moment matching for multi-source domain adaptation},
author={Peng, Xingchao and Bai, Qinxun and Xia, Xide and Huang, Zijun and Saenko, Kate and Wang, Bo},
booktitle={Proceedings of the IEEE International Conference on Computer Vision},
pages={1406--1415},
year={2019}
}
</code> </pre>
</div>
</div>
<div class="project-page">
<h2>Fair Use Notice</h2>
<p>This dataset contains some copyrighted material whose use has not been specifically authorized by the copyright owners. In an effort to advance scientific research, we make this material available for academic research. We believe this constitutes a <strong>fair use</strong> of any such copyrighted material as provided for in section 107 of the US Copyright Law. In accordance with Title 17 U.S.C. Section 107, the material on this site is distributed without profit for non-commercial research and educational purposes. For more information on fair use please <a href="https://www.law.cornell.edu/uscode/text/17/107" title="Fair Use Notice">click here</a>. If you wish to use copyrighted material on this site or in our dataset for purposes of your own that go beyond non-commercial research and academic purposes, you must obtain permission directly from the copyright owner.
<em>(adapted from <a href="http://people.cs.pitt.edu/~chris/photographer/">Christopher Thomas</a>)</em></p>
</div>
</div> <!--container-->
<!-- jQuery (necessary for Bootstrap's JavaScript plugins) -->
<!--<script src="https://ajax.googleapis.com/ajax/libs/jquery/1.11.2/jquery.min.js"></script>-->
<!-- Include all compiled plugins (below), or include individual files as needed -->
<!--<script src="js/bootstrap.min.js"></script>
<script src="js/toggle.js"></script> -->
</body>
</html>