-
Notifications
You must be signed in to change notification settings - Fork 2
/
S3Net.html
136 lines (134 loc) · 5.52 KB
/
S3Net.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
<!DOCTYPE html>
<html>
<head>
<meta charset="utf-8">
<title>S3Net</title>
<link rel="stylesheet" type="text/css" href="assets/scripts/bulma.min.css">
<link rel="stylesheet" type="text/css" href="assets/scripts/theme.css">
<link rel="stylesheet" type="text/css" href="https://cdn.bootcdn.net/ajax/libs/font-awesome/4.7.0/css/font-awesome.min.css">
</head>
<body>
<section class="hero is-light" style="">
<div class="hero-body" style="padding-top: 50px;">
<div class="container" style="text-align: center;margin-bottom:5px;">
<h1 class="title">
Selective, Structural, Subtle: Trilinear Spatial-Awareness
</h1>
<h1 class="title">
for Few-Shot Fine-Grained Visual Recognition
</h1>
<div class="author">Heng Wu<sup>1</sup></div>
<div class="author">Yifan Zhao<sup>1</sup></div>
<div class="author">Jia Li<sup>1,2</sup></div>
<div class="group">
<a href="http://cvteam.net/">CVTEAM</a>
</div>
<div class="aff">
<p><sup>1</sup>State Key Laboratory of Virtual Reality Technology and Systems, SCSE, Beihang University, Beijing, China</p>
<p><sup>2</sup>Pengcheng Laboratory, Shenzhen, China</p>
</div>
<div class="con">
<p style="font-size: 24px; margin-top:5px; margin-bottom: 15px;">
ICME 2021
</p>
</div>
<div class="columns">
<div class="column"></div>
<div class="column"></div>
<div class="column">
<a href="https://ieeexplore.ieee.org/abstract/document/9428223" target="_blank">
<p class="link">Paper</p>
</a>
</div>
<div class="column">
<a href="https://github.com/iCVTEAM/S3Net/" target="_blank">
<p class="link">Code</p>
</a>
</div>
<div class="column"></div>
<div class="column"></div>
</div>
</div>
</div>
</section>
<div style="text-align: center;">
<div class="container" style="max-width:850px">
<div style="text-align: center;">
<img src="assets/S3Net/head.png" class="centerImage">
</div>
</div>
<div class="head_cap">
<p style="color:gray;">
The framework of S3Net
</p>
</div>
</div>
<section class="hero">
<div class="hero-body">
<div class="container" style="max-width: 800px" >
<h1 style="">Abstract</h1>
<p style="text-align: justify; font-size: 17px;">
Few-shot learning aims to recognize the novel categories from
a few examples. However, most of the existing approaches usually
focus on general image classification and fail to handle subtle
differences between images. To alleviate this issue, we propose
a trilinear spatial-awareness network for fewshot-grained visual
recognition, called S3Net, which is composed of a spatial selection
module, structural pyramid descriptor, and subtle difference mining
module. Specifically,we first build the global relation to strengthen
the features by spatial selection module. The structural pyramid
descriptor then constructs a multi-scale representation for enhancing
the rich contextual information by exploiting different receptive
fields in the same feature layer. Furthermore, a similarity
loss based on local descriptors and a global classification loss
is design to help the network learn discrimination capability
by handling subtle differences in confused or near-duplicated
samples. Extensive experiments on 4 few-shot fine-grained
benchmarks demonstrate that our proposed approach is effective
and outperforms state-of-the-art models by large margins.
</p>
</div>
</div>
</section>
<section class="hero is-light" style="background-color:#FFFFFF;">
<div class="hero-body">
<div class="container" style="max-width:800px;margin-bottom:20px;">
<h1>
Performance Comparison
</h1>
</div>
<div class="container" style="max-width:800px">
<div style="text-align: center;">
<img src="assets/S3Net/perfcomp.png" class="centerImage">
</div>
</div>
</div>
</section>
<section class="hero" style="padding-top:0px;">
<div class="hero-body">
<div class="container" style="max-width:800px;">
<div class="card">
<header class="card-header">
<p class="card-header-title">
BibTex Citation
</p>
<a class="card-header-icon button-clipboard" style="border:0px; background: inherit;" data-clipboard-target="#bibtex-info" >
<i class="fa fa-copy" height="20px"></i>
</a>
</header>
<div class="card-content">
<pre style="background-color:inherit;padding: 0px;" id="bibtex-info">@article{9428223,
title={Selective, Structural, Subtle: Trilinear Spatial-Awareness for Few-Shot Fine-Grained Visual Recognition},
author={Wu, Heng and Zhao, Yifan and Li, Jia},
booktitle={2021 IEEE International Conference on Multimedia and Expo (ICME)},
pages={1-6},
year={2021},
}</pre>
</div>
</section>
<script type="text/javascript" src="assets/scripts/clipboard.min.js"></script>
<script>
new ClipboardJS('.button-clipboard');
</script>
</body>
</html>