-
Notifications
You must be signed in to change notification settings - Fork 0
/
index.html
272 lines (249 loc) · 10.8 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
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<meta http-equiv="X-UA-Compatible" content="IE=edge">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title>Burst Image Super-Resolution with Base Frame Selection</title>
<!-- Bootstrap -->
<link href="css/bootstrap-4.4.1.css" rel="stylesheet">
<link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/font-awesome/4.7.0/css/font-awesome.min.css">
<!-- <link href='http://fonts.googleapis.com/css?family=Open+Sans:400italic,700italic,800italic,400,700,800' rel='stylesheet' type='text/css'>
<link rel="stylesheet" type="text/css" href="css/project.css" media="screen" />
<link rel="stylesheet" type="text/css" media="screen" href="css/iconize.css" />
<script src="js/google-code-prettify/prettify.js"></script> -->
<script>
MathJax = {
tex: {
inlineMath: [['$', '$'], ['\\(', '\\)']]
},
svg: {
fontCache: 'global'
}
};
</script>
<script type="text/javascript" id="MathJax-script" async
src="https://cdn.jsdelivr.net/npm/mathjax@3/es5/tex-svg.js">
</script>
</head>
<!-- cover -->
<section>
<div class="jumbotron text-center mt-0">
<div class="container">
<div class="row">
<div class="col-12">
<h2>Burst Image Super-Resolution with Base Frame Selection</h2>
<h4 style="color:#5a6268;">CVPR 2024 Workshop (9<sup>th</sup> NTIRE)</h4>
<hr>
<h6>
<a href="" target="_blank">Sanghyun Kim*</a>,
<a href="https://mlee47.github.io/" target="_blank">Min Jung Lee*</a>,
<a href="https://woo525.github.io/" target="_blank">Woohyeok Kim</a>,
<a href="https://hesedjds.github.io/" target="_blank">Deunsol Jung</a>,
<a href="https://rimchang.github.io/" target="_blank">Jaesung Rim</a>,
<a href="https://www.scho.pe.kr/" target="_blank">Sunghyun Cho</a>,
<a href="https://cvlab.postech.ac.kr/~mcho/" target="_blank">Minsu Cho</a>
</h6>
<p>
Pohang University of Science and Technology (POSTECH)
</p>
<div class="row justify-content-center">
<div class="column">
<p class="mb-5"><a class="btn btn-large btn-light" href="https://arxiv.org/abs/2406.17869" role="button"
target="_blank">
<i class="fa fa-file"></i> Paper</a> </p>
</div>
<!-- <div class="column">
<p class="mb-5"><a class="btn btn-large btn-light" href="https://github.com/sua-choi/CMS"
role="button" target="_blank">
<i class="fa fa-github-alt"></i> Code </a> </p>
</div> -->
</div>
</div>
</div>
</div>
</div>
</section>
<!-- abstract -->
<section>
<div class="container">
<div class="row">
<div class="col-12 text-center">
<br>
<h2>Abstract</h2>
<!-- <hr style="margin-top:0px"> -->
<p class="text-left">
Burst image super-resolution has been a topic of active research in recent years due to its ability to obtain a high resolution image using complementary information between multiple frames in the burst.
In this work, we explore using burst shots with non-uniform exposures to confront real-world practical scenarios by introducing a new benchmark dataset, dubbed Non-uniformly Exposed Burst Image (NEBI), that includes the burst frames at varying exposure times to obtain a broader range of irradiance and motion characteristics within a scene.
As burst shots with non-uniform exposures exhibit varying levels of degradation, fusing information of the burst shots into the first frame as a base frame may not result in optimal image quality.
To address this limitation, we propose a Frame Selection Network~(FSN) for non-uniform scenarios. This network seamlessly integrates into existing super-resolution methods in a plug-and-play manner with low computational cost.
The comparative analysis reveals the effectiveness of the non-uniform setting for the practical scenario and our FSN on synthetic-/real- NEBI datasets.
</p>
<br>
</div>
</div>
</div>
</section>
<br>
<br>
<section>
<div class="container">
<div class="row">
<div class="col-12 text-center">
<h2>Motivation</h2>
<!-- <hr style="margin-top:0px"> -->
</div>
<br>
<div class="col-12 text-center">
<h4>Why do we need non-uniformly exposed burst?</h4>
</div>
<div class="col-12" style="display:flex; align-items:center; margin-bottom:20px;">
<img src="images/teaser1.png" style="width:50%; margin-top:10px; margin-right:20px;">
<div>
When burst photography uses the same exposure times for all frames, it often results in poor quality images due to camera noise and motion blur if the exposure time isn't optimal.
It's hard to determine the best exposure time in real-world scenarios, making this approach less practical.
However, using burst shots with varied exposure times can help reconstruct high-resolution images as though the optimal exposure time was used.
</div>
</div>
<div class="col-12 text-center">
<h4>Why do we need a Frame Selection Network?</h4>
</div>
<div class="col-12" style="display:flex; align-items:center; margin-bottom:20px;">
<img src="images/teaser2.png" style="width:50%; margin-top:10px; margin-right:20px;">
<div>
Burst shots with non-uniform exposures display varying degrees of quality, affecting the alignment and fusion of features between the base and subsequent frames, which can reduce overall image quality.
Previous methods typically use the first frame as the base frame, ignoring its potential negative effects on image restoration.
To address this issue, we propose a Frame Selection Network (FSN) that identifies the most suitable base frame to enhance overall image quality. </div>
</div>
</div>
</div>
</section>
<br>
<br>
<br>
<section>
<div class="container">
<div class="row">
<div class="col-12 text-center">
<h2>Benchmark: Synthetic-/Real-NEBI</h2>
</div>
<div class="col-12">
<p class="text-left">
<img src="images/benchmark.png" style="width:100%; margin-top:10px; margin-bottom:10px;">
<br>
In the burst sequence, from left to right, the exposure time increases, leading to reduced noise and increased blur.
<br>
</p>
</div>
</div>
</div>
</section>
<br>
<br>
<br>
<section>
<div class="container">
<div class="row">
<div class="col-12 text-center">
<h2>Method</h2>
</div>
<div class="col-12">
<p class="text-left">
<!-- <h4>Evaluation on GCD</h4> -->
<img src="images/method.png" style="width:100%; margin-top:10px; margin-bottom:10px;">
<br>
FSN first constructs the image feature of each frame using CNN.
The constructed image feature is fed into our <b>C</b>orrelation-based <b>M</b>otion <b>A</b>ware (CMA) blocks, which update the image feature based on motion information extracted from the burst frames.
To extract motion information from the burst frames, we propose <b>F</b>eature <b>C</b>orrelation <b>M</b>odule (FCM), which computes the local correlation along both spatial and temporal axes.
<br>
</p>
</div>
</div>
</div>
</section>
<br>
<br>
<br>
<section>
<div class="container">
<div class="row">
<div class="col-12 text-center">
<h2>Experiments</h2>
</div>
<div class="col-12">
<p class="text-left">
<div class="col-12 text-center">
<h4>Evaluation on Synthetic-/Real-NEBI</h4>
</div>
<img src="images/experiment1.png" style="width:60%; margin-top:10px; display: block; margin-left: auto; margin-right: auto;">
<br>
Comparison of existing burst super-resolution models and their variants incorporating our Feature Selection Network (FSN), along with an evaluation against the Auto-Exposure (AE) algorithm.
<br>
<br>
<div class="col-12 text-center">
<h4>Training strategies on Real-NEBI</h4>
</div>
<img src="images/experiment2.png" style="width:40%; margin-top:10px; display: block; margin-left: auto; margin-right: auto;">
<br>
When training a Frame Selection Network (FSN) with different targets like PSNR, SSIM, and LPIPS using synthetic data, and then evaluating it on real data, the training target affects the performance.
<br>
<div class="col-12 text-center">
<h4>Ablation study</h4>
</div>
<img src="images/ablation1.png" style="width:40%; margin-top:10px; display: block; margin-left: auto; margin-right: auto;">
<br>
Comparison of the performance of a baseline super-resolution network against variants that incorporate varying numbers of CMA blocks.
<br>
</p>
</div>
</div>
</div>
</section>
<br>
<br>
<br>
<section>
<div class="container">
<div class="row">
<div class="col-12 text-center">
<h2>Qualitative results</h2>
</div>
<div class="col-12">
<p class="text-left">
<img src="images/qual.png" style="width:100%; margin-top:10px; margin-bottom:10px;">
BIPNet, when combined with our frame selector, improves the clarity of high-frequency image details by effectively merging complementary information from multiple frames into the chosen base frame.
Conversely, without the frame selector, BIPNet tends to produce blurry images due to significant degradation in the initial frame.
<div class="col-12">
<p class="text-left">
</p>
</div>
</div>
</div>
</section>
<br>
<!-- citing -->
<div class="container">
<div class="row ">
<div class="col-12">
<h3>Citation</h3>
<hr style="margin-top:0px">
<pre style="background-color: #e9eeef;padding: 1.25em 1.5em">
<code>
@inproceedings{kim2024burst,
title={Burst Image Super-Resolution with Base Frame Selection},
author={Kim, Sanghyun and Lee, Minjung and Kim, Woohyeok and Jung, Deunsol and Rim, Jaesung and Cho, Sunghyun and Cho, Minsu},
booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
pages={5940--5949},
year={2024}
}
</code>
</pre>
<hr>
</div>
</div>
</div>
<footer class="text-center" style="margin-bottom:10px">
Thanks to <a href="https://lioryariv.github.io/" target="_blank">Lior Yariv</a> for the website template.
</footer>
</body>
</html>