-
Notifications
You must be signed in to change notification settings - Fork 4
/
Copy pathindex.html
367 lines (324 loc) · 9.74 KB
/
index.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
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
<script src="http://www.google.com/jsapi" type="text/javascript"></script>
<script type="text/javascript">google.load("jquery", "1.3.2");</script>
<style type="text/css">
body {
font-family: "HelveticaNeue-Light", "Helvetica Neue Light", "Helvetica Neue", Helvetica, Arial, "Lucida Grande", sans-serif;
font-weight: 300;
font-size: 24px;
margin-left: auto;
margin-right: auto;
width: 1100px;
}
h2 {
font-weight: 300;
text-align: center;
font-size: 30px;
}
.disclaimerbox {
background-color: #eee;
border: 1px solid #eeeeee;
border-radius: 10px;
-moz-border-radius: 10px;
-webkit-border-radius: 10px;
padding: 20px;
}
video.header-vid {
height: 140px;
border: 1px solid black;
border-radius: 10px;
-moz-border-radius: 10px;
-webkit-border-radius: 10px;
}
img.header-img {
height: 140px;
border: 1px solid black;
border-radius: 10px;
-moz-border-radius: 10px;
-webkit-border-radius: 10px;
}
img.rounded {
border: 0px solid #eeeeee;
border-radius: 10px;
-moz-border-radius: 10px;
-webkit-border-radius: 10px;
}
a:link,
a:visited {
color: #1367a7;
text-decoration: none;
}
a:hover {
color: #208799;
}
td.dl-link {
height: 160px;
text-align: center;
font-size: 22px;
}
.layered-paper-big {
/* modified from: http://css-tricks.com/snippets/css/layered-paper/ */
box-shadow:
0px 0px 1px 1px rgba(0, 0, 0, 0.35),
/* The top layer shadow */
5px 5px 0 0px #fff,
/* The second layer */
5px 5px 1px 1px rgba(0, 0, 0, 0.35),
/* The second layer shadow */
10px 10px 0 0px #fff,
/* The third layer */
10px 10px 1px 1px rgba(0, 0, 0, 0.35),
/* The third layer shadow */
15px 15px 0 0px #fff,
/* The fourth layer */
15px 15px 1px 1px rgba(0, 0, 0, 0.35),
/* The fourth layer shadow */
20px 20px 0 0px #fff,
/* The fifth layer */
20px 20px 1px 1px rgba(0, 0, 0, 0.35),
/* The fifth layer shadow */
25px 25px 0 0px #fff,
/* The fifth layer */
25px 25px 1px 1px rgba(0, 0, 0, 0.35);
/* The fifth layer shadow */
margin-left: 10px;
margin-right: 45px;
}
.paper-big {
/* modified from: http://css-tricks.com/snippets/css/layered-paper/ */
box-shadow:
0px 0px 1px 1px rgba(0, 0, 0, 0.35);
/* The top layer shadow */
margin-left: 10px;
margin-right: 45px;
}
.layered-paper {
/* modified from: http://css-tricks.com/snippets/css/layered-paper/ */
box-shadow:
0px 0px 1px 1px rgba(0, 0, 0, 0.35),
/* The top layer shadow */
5px 5px 0 0px #fff,
/* The second layer */
5px 5px 1px 1px rgba(0, 0, 0, 0.35),
/* The second layer shadow */
10px 10px 0 0px #fff,
/* The third layer */
10px 10px 1px 1px rgba(0, 0, 0, 0.35);
/* The third layer shadow */
margin-top: 5px;
margin-left: 10px;
margin-right: 30px;
margin-bottom: 5px;
}
.vert-cent {
position: relative;
top: 50%;
transform: translateY(-50%);
}
hr {
border: 0;
height: 1px;
background-image: linear-gradient(to right, rgba(0, 0, 0, 0), rgba(0, 0, 0, 0.75), rgba(0, 0, 0, 0));
}
p, li {
font-family: source-serif-pro, Georgia, Cambria, "Times New Roman", Times, serif;
font-size: 16px;
line-height: 1.4em;
text-align: justify;
}
p {
text-indent: 2em;
}
.caption {
font-size: 16px;
text-align: center;
text-indent: 0em;
}
main {
width: 800px;
margin-left: auto;
margin-right: auto;
margin-bottom: 200px;
}
.namerow {
display: flex;
justify-content: space-around;
margin-top: 1em;
margin-bottom: 1em;
}
.namerow > div {
font-size: 18px;
}
.schoolrow {
display: flex;
justify-content: space-evenly;
margin-top: 1em;
margin-bottom: 1em;
flex-wrap: wrap;
}
.schoolrow > div {
font-size: 16px;
}
.urlrow {
display: flex;
justify-content: space-evenly;
margin-top: 1em;
margin-bottom: 1em;
margin-left: 100px;
margin-right: 100px;
}
.paper {
display: flex;
align-items: center;
margin-left: 70px;
margin-right: 70px;
}
</style>
<html>
<head>
<title>Composed Image Retrieval with Text Feedback via Multi-Grained Uncertainty Regularization</title>
<meta property="og:title"
content="Composed Image Retrieval with Text Feedback via Multi-Grained Uncertainty Regularization. In ICLR, 2024." />
<meta property="og:url" content="https://monoxide-chen.github.io/uncertainty_retrieval/" />
<meta name='description'
content="The project page of Composed Image Retrieval with Text Feedback via Multi-Grained Uncertainty Regularization. In ICLR, 2024. Multi-Retreival. Yiyang Chen, Zhedong Zheng, Tat-seng Chua." />
<meta name="HandheldFriendly" content="True">
<meta name="MobileOptimized" content="320">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
</head>
<body>
<main>
<center>
<span style="font-size:34px">Composed Image Retrieval with Text Feedback via Multi-Grained Uncertainty
Regularization</span><br>
<div class="namerow">
<div>
<a href="https://scholar.google.com/citations?user=hnxvC5UAAAAJ">Yiyang Chen</a><sup>1,2</sup>
</div>
<div>
<a href="http://zdzheng.xyz/">Zhedong Zheng</a><sup>3</sup>
</div>
<div>
<a href="https://jiwei0523.github.io/">Wei Ji</a><sup>1</sup>
</div>
<div>
<a href="https://scholar.google.com/citations?user=1W2Tio4AAAAJ">Leigang Qu</a><sup>1</sup>
</div>
<div>
<a href="https://www.chuatatseng.com/">Tat-seng Chua</a><sup>1</sup>
</div>
</div>
<div class="schoolrow">
<div style="margin-right: 2em;">
<sup>1</sup>Sea-NExT Joint Lab, National University of Singapore
</div>
<div>
<sup>2</sup>Tsinghua University
</div>
<div>
<sup>3</sup>Faculty of Science and Technology, and Institute of Collaborative Innovation, University of Macau
</div>
</div>
<div class="urlrow">
<div>
Code<a href='https://github.com/Monoxide-Chen/uncertainty_retrieval'>[GitHub]</a>
</div>
<div>
Paper<a href="https://arxiv.org/abs/2211.07394"> [arXiv]</a>
</div>
<div>
Cite<a href="bibtex.txt"> [BibTeX]</a>
</div>
</div>
</center>
<hr>
<h2>Abstract</h2>
<p>
We investigate composed image retrieval with text feedback. Users gradually look for the target of interest by
moving from coarse to fine-grained feedback. However, existing methods merely focus on the latter, i.e.,
fine-grained search, by harnessing positive and negative pairs during training. This pair-based paradigm only
considers the one-to-one distance between a pair of specific points, which is not aligned with the one-to-many
coarse-grained retrieval process and compromises the recall rate.
</p>
<p style="margin-bottom:0.5em;">
In an attempt to fill this gap, we introduce a unified learning approach to simultaneously modeling the coarse- and fine-grained retrieval by considering the
multi-grained uncertainty. The key idea underpinning the proposed method is to integrate fine- and
coarse-grained retrieval as matching data points with small and large fluctuations, respectively. Specifically,
our method contains two modules: uncertainty modeling and uncertainty regularization.
<!-- <br> -->
</p>
<ul style="margin-top:0.5em;">
<li>The uncertainty modeling simulates the multi-grained queries by introducing identically distributed fluctuations in the feature
space.
</li>
<li>
Based on the uncertainty modeling, we further introduce uncertainty regularization to adapt the
matching objective according to the fluctuation range. Compared with existing methods, the proposed strategy
explicitly prevents the model from pushing away potential candidates in the early stage, and thus improves the
recall rate. On the three public datasets, i.e., FashionIQ, Fashion200k, and Shoes, the proposed method has
achieved +4.03%, +3.38%, and +2.40% Recall@50 accuracy over a strong baseline, respectively.
</li>
</ul>
<!-- <br> -->
<hr>
<h2>Architecture</h2>
<p>
Our main contributions are the uncertainty modeling via augmenter, and the uncertainty regularization
for coarse matching. Our model applies both the fine-grained matching and the proposed coarse-grained
uncertainty regularization, facilitating the model training.
</p>
<p>
<center>
<img class="round" style="width:800px" src="pipeline.png" /></a>
</center>
</p>
<p class="caption">
The overview of our network.
</p>
<hr>
<h2>Results</h2>
<p>
Without loss of generability, we verify the effectiveness of the proposed method on the fashion datasets, which collect
the feedback from customers easily, including <strong>FashionIQ</strong>, <strong>Fashion200k</strong> and <strong>Shoes</strong>.
Each image in these fashion datasets is tagged with descriptive texts as product description, such as 'similar style t-shirt
but white logo print'.
</p>
<p>
<center>
<img class="round" style="width:700px" src="result1.png" /></a>
</center>
</p>
<p class="caption">
Results on FashionIQ.
</p>
<p>
<center>
<img class="round" style="width:600px" src="result2.png" /></a>
</center>
</p>
<p class="caption">
Results on Fashion200k and Shoes.
</p>
<hr>
<h2>Paper</h2>
<div class="paper">
<div>
<img class="layered-paper-big" style="height:175px" src="paper01.png" />
</div>
<div>
<span style="font-size:12pt">Y. Chen, Z. Zheng, W. Ji, L. Qu, T. Chua.</span><br>
<b><span style="font-size:12pt">Composed Image Retrieval with Text Feedback via Multi-Grained Uncertainty
Regularization.</b></span><br>
<span style="font-size:12pt">ICLR, 2024 <a href="https://arxiv.org/abs/2211.07394">[ArXiv]</a></span>.
</div>
</div>
<!-- Global site tag (gtag.js) - Google Analytics -->
<script async src="https://www.googletagmanager.com/gtag/js?id=UA-75863369-5"></script>
<script>
window.dataLayer = window.dataLayer || [];
function gtag() { dataLayer.push(arguments); }
gtag('js', new Date());
gtag('config', 'UA-75863369-5');
</script>
</main>
</body>
</html>